348 files changed, 95729 insertions, 3012 deletions

diff --git a/.github/workflows/premerge.yaml b/.github/workflows/premerge.yaml
index 03c0c01..951fc16 100644
--- a/.github/workflows/premerge.yaml
+++ b/.github/workflows/premerge.yaml
@@ -62,6 +62,7 @@ jobs:
         with:
           fetch-depth: 2
       - name: Build and Test
+        timeout-minutes: 120
         continue-on-error: ${{ runner.arch == 'ARM64' }}
         run: |
           git config --global --add safe.directory '*'
@@ -149,6 +150,7 @@ jobs:
           echo "windows-runtimes=${runtimes_to_build}" >> $GITHUB_OUTPUT
           echo "windows-runtimes-check-targets=${runtimes_check_targets}" >> $GITHUB_OUTPUT
       - name: Build and Test
+        timeout-minutes: 180
         if: ${{ steps.vars.outputs.windows-projects != '' }}
         shell: cmd
         run: |
diff --git a/clang-tools-extra/clang-tidy/abseil/RedundantStrcatCallsCheck.cpp b/clang-tools-extra/clang-tidy/abseil/RedundantStrcatCallsCheck.cpp
index d7cc0ca..a58c041 100644
--- a/clang-tools-extra/clang-tidy/abseil/RedundantStrcatCallsCheck.cpp
+++ b/clang-tools-extra/clang-tidy/abseil/RedundantStrcatCallsCheck.cpp
@@ -45,7 +45,10 @@ struct StrCatCheckResult {
   std::vector<FixItHint> Hints;
 };
 
-void removeCallLeaveArgs(const CallExpr *Call, StrCatCheckResult *CheckResult) {
+} // namespace
+
+static void removeCallLeaveArgs(const CallExpr *Call,
+                                StrCatCheckResult *CheckResult) {
   if (Call->getNumArgs() == 0)
     return;
   // Remove 'Foo('
@@ -58,9 +61,9 @@ void removeCallLeaveArgs(const CallExpr *Call, StrCatCheckResult *CheckResult) {
           Call->getRParenLoc(), Call->getEndLoc().getLocWithOffset(1))));
 }
 
-const clang::CallExpr *processArgument(const Expr *Arg,
-                                       const MatchFinder::MatchResult &Result,
-                                       StrCatCheckResult *CheckResult) {
+static const clang::CallExpr *
+processArgument(const Expr *Arg, const MatchFinder::MatchResult &Result,
+                StrCatCheckResult *CheckResult) {
   const auto IsAlphanum = hasDeclaration(cxxMethodDecl(hasName("AlphaNum")));
   static const auto *const Strcat = new auto(hasName("::absl::StrCat"));
   const auto IsStrcat = cxxBindTemporaryExpr(
@@ -78,8 +81,8 @@ const clang::CallExpr *processArgument(const Expr *Arg,
   return nullptr;
 }
 
-StrCatCheckResult processCall(const CallExpr *RootCall, bool IsAppend,
-                              const MatchFinder::MatchResult &Result) {
+static StrCatCheckResult processCall(const CallExpr *RootCall, bool IsAppend,
+                                     const MatchFinder::MatchResult &Result) {
   StrCatCheckResult CheckResult;
   std::deque<const CallExpr *> CallsToProcess = {RootCall};
 
@@ -101,7 +104,6 @@ StrCatCheckResult processCall(const CallExpr *RootCall, bool IsAppend,
   }
   return CheckResult;
 }
-} // namespace
 
 void RedundantStrcatCallsCheck::check(const MatchFinder::MatchResult &Result) {
   bool IsAppend = false;
diff --git a/clang-tools-extra/clang-tidy/bugprone/MultipleStatementMacroCheck.cpp b/clang-tools-extra/clang-tidy/bugprone/MultipleStatementMacroCheck.cpp
index 390f3dd..54ed899 100644
--- a/clang-tools-extra/clang-tidy/bugprone/MultipleStatementMacroCheck.cpp
+++ b/clang-tools-extra/clang-tidy/bugprone/MultipleStatementMacroCheck.cpp
@@ -18,8 +18,11 @@ namespace {
 
 AST_MATCHER(Expr, isInMacro) { return Node.getBeginLoc().isMacroID(); }
 
+} // namespace
+
 /// Find the next statement after `S`.
-const Stmt *nextStmt(const MatchFinder::MatchResult &Result, const Stmt *S) {
+static const Stmt *nextStmt(const MatchFinder::MatchResult &Result,
+                            const Stmt *S) {
   auto Parents = Result.Context->getParents(*S);
   if (Parents.empty())
     return nullptr;
@@ -40,8 +43,8 @@ using ExpansionRanges = std::vector<SourceRange>;
 /// \brief Get all the macro expansion ranges related to `Loc`.
 ///
 /// The result is ordered from most inner to most outer.
-ExpansionRanges getExpansionRanges(SourceLocation Loc,
-                                   const MatchFinder::MatchResult &Result) {
+static ExpansionRanges
+getExpansionRanges(SourceLocation Loc, const MatchFinder::MatchResult &Result) {
   ExpansionRanges Locs;
   while (Loc.isMacroID()) {
     Locs.push_back(
@@ -51,8 +54,6 @@ ExpansionRanges getExpansionRanges(SourceLocation Loc,
   return Locs;
 }
 
-} // namespace
-
 void MultipleStatementMacroCheck::registerMatchers(MatchFinder *Finder) {
   const auto Inner = expr(isInMacro(), unless(compoundStmt())).bind("inner");
   Finder->addMatcher(
diff --git a/clang-tools-extra/clang-tidy/cppcoreguidelines/ProTypeMemberInitCheck.cpp b/clang-tools-extra/clang-tidy/cppcoreguidelines/ProTypeMemberInitCheck.cpp
index 37d737a..1ac9b8b 100644
--- a/clang-tools-extra/clang-tidy/cppcoreguidelines/ProTypeMemberInitCheck.cpp
+++ b/clang-tools-extra/clang-tidy/cppcoreguidelines/ProTypeMemberInitCheck.cpp
@@ -28,10 +28,13 @@ AST_MATCHER(CXXRecordDecl, hasDefaultConstructor) {
   return Node.hasDefaultConstructor();
 }
 
+} // namespace
+
 // Iterate over all the fields in a record type, both direct and indirect (e.g.
 // if the record contains an anonymous struct).
 template <typename T, typename Func>
-void forEachField(const RecordDecl &Record, const T &Fields, const Func &Fn) {
+static void forEachField(const RecordDecl &Record, const T &Fields,
+                         const Func &Fn) {
   for (const FieldDecl *F : Fields) {
     if (F->isAnonymousStructOrUnion()) {
       if (const CXXRecordDecl *R = F->getType()->getAsCXXRecordDecl())
@@ -43,8 +46,9 @@ void forEachField(const RecordDecl &Record, const T &Fields, const Func &Fn) {
 }
 
 template <typename T, typename Func>
-void forEachFieldWithFilter(const RecordDecl &Record, const T &Fields,
-                            bool &AnyMemberHasInitPerUnion, const Func &Fn) {
+static void forEachFieldWithFilter(const RecordDecl &Record, const T &Fields,
+                                   bool &AnyMemberHasInitPerUnion,
+                                   const Func &Fn) {
   for (const FieldDecl *F : Fields) {
     if (F->isAnonymousStructOrUnion()) {
       if (const CXXRecordDecl *R = F->getType()->getAsCXXRecordDecl()) {
@@ -59,8 +63,9 @@ void forEachFieldWithFilter(const RecordDecl &Record, const T &Fields,
   }
 }
 
-void removeFieldInitialized(const FieldDecl *M,
-                            SmallPtrSetImpl<const FieldDecl *> &FieldDecls) {
+static void
+removeFieldInitialized(const FieldDecl *M,
+                       SmallPtrSetImpl<const FieldDecl *> &FieldDecls) {
   const RecordDecl *R = M->getParent();
   if (R && R->isUnion()) {
     // Erase all members in a union if any member of it is initialized.
@@ -70,9 +75,9 @@ void removeFieldInitialized(const FieldDecl *M,
     FieldDecls.erase(M);
 }
 
-void removeFieldsInitializedInBody(
-    const Stmt &Stmt, ASTContext &Context,
-    SmallPtrSetImpl<const FieldDecl *> &FieldDecls) {
+static void
+removeFieldsInitializedInBody(const Stmt &Stmt, ASTContext &Context,
+                              SmallPtrSetImpl<const FieldDecl *> &FieldDecls) {
   auto Matches =
       match(findAll(binaryOperator(
                 hasOperatorName("="),
@@ -82,9 +87,9 @@ void removeFieldsInitializedInBody(
     removeFieldInitialized(Match.getNodeAs<FieldDecl>("fieldDecl"), FieldDecls);
 }
 
-StringRef getName(const FieldDecl *Field) { return Field->getName(); }
+static StringRef getName(const FieldDecl *Field) { return Field->getName(); }
 
-StringRef getName(const RecordDecl *Record) {
+static StringRef getName(const RecordDecl *Record) {
   // Get the typedef name if this is a C-style anonymous struct and typedef.
   if (const TypedefNameDecl *Typedef = Record->getTypedefNameForAnonDecl())
     return Typedef->getName();
@@ -94,7 +99,7 @@ StringRef getName(const RecordDecl *Record) {
 // Creates comma separated list of decls requiring initialization in order of
 // declaration.
 template <typename R, typename T>
-std::string
+static std::string
 toCommaSeparatedString(const R &OrderedDecls,
                        const SmallPtrSetImpl<const T *> &DeclsToInit) {
   SmallVector<StringRef, 16> Names;
@@ -105,12 +110,14 @@ toCommaSeparatedString(const R &OrderedDecls,
   return llvm::join(Names.begin(), Names.end(), ", ");
 }
 
-SourceLocation getLocationForEndOfToken(const ASTContext &Context,
-                                        SourceLocation Location) {
+static SourceLocation getLocationForEndOfToken(const ASTContext &Context,
+                                               SourceLocation Location) {
   return Lexer::getLocForEndOfToken(Location, 0, Context.getSourceManager(),
                                     Context.getLangOpts());
 }
 
+namespace {
+
 // There are 3 kinds of insertion placements:
 enum class InitializerPlacement {
   // 1. The fields are inserted after an existing CXXCtorInitializer stored in
@@ -187,15 +194,17 @@ struct InitializerInsertion {
   SmallVector<std::string, 4> Initializers;
 };
 
+} // namespace
+
 // Convenience utility to get a RecordDecl from a QualType.
-const RecordDecl *getCanonicalRecordDecl(const QualType &Type) {
+static const RecordDecl *getCanonicalRecordDecl(const QualType &Type) {
   if (const auto *RT = Type->getAsCanonical<RecordType>())
     return RT->getDecl();
   return nullptr;
 }
 
 template <typename R, typename T>
-SmallVector<InitializerInsertion, 16>
+static SmallVector<InitializerInsertion, 16>
 computeInsertions(const CXXConstructorDecl::init_const_range &Inits,
                   const R &OrderedDecls,
                   const SmallPtrSetImpl<const T *> &DeclsToInit) {
@@ -239,8 +248,9 @@ computeInsertions(const CXXConstructorDecl::init_const_range &Inits,
 
 // Gets the list of bases and members that could possibly be initialized, in
 // order as they appear in the class declaration.
-void getInitializationsInOrder(const CXXRecordDecl &ClassDecl,
-                               SmallVectorImpl<const NamedDecl *> &Decls) {
+static void
+getInitializationsInOrder(const CXXRecordDecl &ClassDecl,
+                          SmallVectorImpl<const NamedDecl *> &Decls) {
   Decls.clear();
   for (const auto &Base : ClassDecl.bases()) {
     // Decl may be null if the base class is a template parameter.
@@ -253,9 +263,10 @@ void getInitializationsInOrder(const CXXRecordDecl &ClassDecl,
 }
 
 template <typename T>
-void fixInitializerList(const ASTContext &Context, DiagnosticBuilder &Diag,
-                        const CXXConstructorDecl *Ctor,
-                        const SmallPtrSetImpl<const T *> &DeclsToInit) {
+static void fixInitializerList(const ASTContext &Context,
+                               DiagnosticBuilder &Diag,
+                               const CXXConstructorDecl *Ctor,
+                               const SmallPtrSetImpl<const T *> &DeclsToInit) {
   // Do not propose fixes in macros since we cannot place them correctly.
   if (Ctor->getBeginLoc().isMacroID())
     return;
@@ -271,8 +282,6 @@ void fixInitializerList(const ASTContext &Context, DiagnosticBuilder &Diag,
   }
 }
 
-} // anonymous namespace
-
 ProTypeMemberInitCheck::ProTypeMemberInitCheck(StringRef Name,
                                                ClangTidyContext *Context)
     : ClangTidyCheck(Name, Context),
diff --git a/clang-tools-extra/clang-tidy/misc/CoroutineHostileRAIICheck.cpp b/clang-tools-extra/clang-tidy/misc/CoroutineHostileRAIICheck.cpp
index 8ec7695..3b9b8e0 100644
--- a/clang-tools-extra/clang-tidy/misc/CoroutineHostileRAIICheck.cpp
+++ b/clang-tools-extra/clang-tidy/misc/CoroutineHostileRAIICheck.cpp
@@ -60,12 +60,12 @@ AST_MATCHER_P(CoawaitExpr, awaitable, ast_matchers::internal::Matcher<Expr>,
     return InnerMatcher.matches(*E, Finder, Builder);
   return false;
 }
+} // namespace
 
-auto typeWithNameIn(const std::vector<StringRef> &Names) {
+static auto typeWithNameIn(const std::vector<StringRef> &Names) {
   return hasType(
       hasCanonicalType(hasDeclaration(namedDecl(hasAnyName(Names)))));
 }
-} // namespace
 
 CoroutineHostileRAIICheck::CoroutineHostileRAIICheck(StringRef Name,
                                                      ClangTidyContext *Context)
diff --git a/clang-tools-extra/clang-tidy/misc/NoRecursionCheck.cpp b/clang-tools-extra/clang-tidy/misc/NoRecursionCheck.cpp
index 0d7667c..035598d 100644
--- a/clang-tools-extra/clang-tidy/misc/NoRecursionCheck.cpp
+++ b/clang-tools-extra/clang-tidy/misc/NoRecursionCheck.cpp
@@ -151,10 +151,12 @@ constexpr unsigned SmallSCCSize = 32;
 using CallStackTy =
     llvm::SmallVector<CallGraphNode::CallRecord, SmallCallStackSize>;
 
+} // namespace
+
 // In given SCC, find *some* call stack that will be cyclic.
 // This will only find *one* such stack, it might not be the smallest one,
 // and there may be other loops.
-CallStackTy pathfindSomeCycle(ArrayRef<CallGraphNode *> SCC) {
+static CallStackTy pathfindSomeCycle(ArrayRef<CallGraphNode *> SCC) {
   // We'll need to be able to performantly look up whether some CallGraphNode
   // is in SCC or not, so cache all the SCC elements in a set.
   const ImmutableSmallSet<CallGraphNode *, SmallSCCSize> SCCElts(SCC);
@@ -190,8 +192,6 @@ CallStackTy pathfindSomeCycle(ArrayRef<CallGraphNode *> SCC) {
   return CallStack;
 }
 
-} // namespace
-
 void NoRecursionCheck::registerMatchers(MatchFinder *Finder) {
   Finder->addMatcher(translationUnitDecl().bind("TUDecl"), this);
 }
diff --git a/clang-tools-extra/clang-tidy/readability/UppercaseLiteralSuffixCheck.cpp b/clang-tools-extra/clang-tidy/readability/UppercaseLiteralSuffixCheck.cpp
index c1dc209..740a68d 100644
--- a/clang-tools-extra/clang-tidy/readability/UppercaseLiteralSuffixCheck.cpp
+++ b/clang-tools-extra/clang-tidy/readability/UppercaseLiteralSuffixCheck.cpp
@@ -55,8 +55,10 @@ struct NewSuffix {
   std::optional<FixItHint> FixIt;
 };
 
-std::optional<SourceLocation> getMacroAwareLocation(SourceLocation Loc,
-                                                    const SourceManager &SM) {
+} // namespace
+
+static std::optional<SourceLocation>
+getMacroAwareLocation(SourceLocation Loc, const SourceManager &SM) {
   // Do nothing if the provided location is invalid.
   if (Loc.isInvalid())
     return std::nullopt;
@@ -67,8 +69,8 @@ std::optional<SourceLocation> getMacroAwareLocation(SourceLocation Loc,
   return SpellingLoc;
 }
 
-std::optional<SourceRange> getMacroAwareSourceRange(SourceRange Loc,
-                                                    const SourceManager &SM) {
+static std::optional<SourceRange>
+getMacroAwareSourceRange(SourceRange Loc, const SourceManager &SM) {
   std::optional<SourceLocation> Begin =
       getMacroAwareLocation(Loc.getBegin(), SM);
   std::optional<SourceLocation> End = getMacroAwareLocation(Loc.getEnd(), SM);
@@ -77,7 +79,7 @@ std::optional<SourceRange> getMacroAwareSourceRange(SourceRange Loc,
   return SourceRange(*Begin, *End);
 }
 
-std::optional<std::string>
+static std::optional<std::string>
 getNewSuffix(llvm::StringRef OldSuffix,
              const std::vector<StringRef> &NewSuffixes) {
   // If there is no config, just uppercase the entirety of the suffix.
@@ -96,7 +98,7 @@ getNewSuffix(llvm::StringRef OldSuffix,
 }
 
 template <typename LiteralType>
-std::optional<NewSuffix>
+static std::optional<NewSuffix>
 shouldReplaceLiteralSuffix(const Expr &Literal,
                            const std::vector<StringRef> &NewSuffixes,
                            const SourceManager &SM, const LangOptions &LO) {
@@ -174,8 +176,6 @@ shouldReplaceLiteralSuffix(const Expr &Literal,
   return ReplacementDsc;
 }
 
-} // namespace
-
 UppercaseLiteralSuffixCheck::UppercaseLiteralSuffixCheck(
     StringRef Name, ClangTidyContext *Context)
     : ClangTidyCheck(Name, Context),
diff --git a/clang-tools-extra/clang-tidy/utils/DeclRefExprUtils.cpp b/clang-tools-extra/clang-tidy/utils/DeclRefExprUtils.cpp
index 57453ad..a5b0883 100644
--- a/clang-tools-extra/clang-tidy/utils/DeclRefExprUtils.cpp
+++ b/clang-tools-extra/clang-tidy/utils/DeclRefExprUtils.cpp
@@ -19,9 +19,8 @@ namespace clang::tidy::utils::decl_ref_expr {
 using namespace ::clang::ast_matchers;
 using llvm::SmallPtrSet;
 
-namespace {
-
-template <typename S> bool isSetDifferenceEmpty(const S &S1, const S &S2) {
+template <typename S>
+static bool isSetDifferenceEmpty(const S &S1, const S &S2) {
   for (auto E : S1)
     if (S2.count(E) == 0)
       return false;
@@ -30,15 +29,15 @@ template <typename S> bool isSetDifferenceEmpty(const S &S1, const S &S2) {
 
 // Extracts all Nodes keyed by ID from Matches and inserts them into Nodes.
 template <typename Node>
-void extractNodesByIdTo(ArrayRef<BoundNodes> Matches, StringRef ID,
-                        SmallPtrSet<const Node *, 16> &Nodes) {
+static void extractNodesByIdTo(ArrayRef<BoundNodes> Matches, StringRef ID,
+                               SmallPtrSet<const Node *, 16> &Nodes) {
   for (const auto &Match : Matches)
     Nodes.insert(Match.getNodeAs<Node>(ID));
 }
 
 // Returns true if both types refer to the same type,
 // ignoring the const-qualifier.
-bool isSameTypeIgnoringConst(QualType A, QualType B) {
+static bool isSameTypeIgnoringConst(QualType A, QualType B) {
   A = A.getCanonicalType();
   B = B.getCanonicalType();
   A.addConst();
@@ -47,7 +46,8 @@ bool isSameTypeIgnoringConst(QualType A, QualType B) {
 }
 
 // Returns true if `D` and `O` have the same parameter types.
-bool hasSameParameterTypes(const CXXMethodDecl &D, const CXXMethodDecl &O) {
+static bool hasSameParameterTypes(const CXXMethodDecl &D,
+                                  const CXXMethodDecl &O) {
   if (D.getNumParams() != O.getNumParams())
     return false;
   for (int I = 0, E = D.getNumParams(); I < E; ++I) {
@@ -60,7 +60,7 @@ bool hasSameParameterTypes(const CXXMethodDecl &D, const CXXMethodDecl &O) {
 
 // If `D` has a const-qualified overload with otherwise identical
 // ref-qualifiers and parameter types, returns that overload.
-const CXXMethodDecl *findConstOverload(const CXXMethodDecl &D) {
+static const CXXMethodDecl *findConstOverload(const CXXMethodDecl &D) {
   assert(!D.isConst());
 
   DeclContext::lookup_result LookupResult =
@@ -81,7 +81,7 @@ const CXXMethodDecl *findConstOverload(const CXXMethodDecl &D) {
 
 // Returns true if both types are pointers or reference to the same type,
 // ignoring the const-qualifier.
-bool pointsToSameTypeIgnoringConst(QualType A, QualType B) {
+static bool pointsToSameTypeIgnoringConst(QualType A, QualType B) {
   assert(A->isPointerType() || A->isReferenceType());
   assert(B->isPointerType() || B->isReferenceType());
   return isSameTypeIgnoringConst(A->getPointeeType(), B->getPointeeType());
@@ -122,7 +122,7 @@ bool pointsToSameTypeIgnoringConst(QualType A, QualType B) {
 //
 // This function checks (A) ad (B), but the caller should make sure that the
 // object is not mutated through the return value.
-bool isLikelyShallowConst(const CXXMethodDecl &M) {
+static bool isLikelyShallowConst(const CXXMethodDecl &M) {
   assert(!M.isConst());
   // The method can mutate our variable.
 
@@ -146,6 +146,8 @@ bool isLikelyShallowConst(const CXXMethodDecl &M) {
   return isSameTypeIgnoringConst(CallTy, OverloadTy);
 }
 
+namespace {
+
 // A matcher that matches DeclRefExprs that are used in ways such that the
 // underlying declaration is not modified.
 // If the declaration is of pointer type, `Indirections` specifies the level
diff --git a/clang-tools-extra/clang-tidy/utils/DesignatedInitializers.cpp b/clang-tools-extra/clang-tidy/utils/DesignatedInitializers.cpp
index 044f89b..b068ae2 100644
--- a/clang-tools-extra/clang-tidy/utils/DesignatedInitializers.cpp
+++ b/clang-tools-extra/clang-tidy/utils/DesignatedInitializers.cpp
@@ -19,8 +19,6 @@
 
 namespace clang::tidy::utils {
 
-namespace {
-
 /// Returns true if Name is reserved, like _Foo or __Vector_base.
 static inline bool isReservedName(llvm::StringRef Name) {
   // This doesn't catch all cases, but the most common.
@@ -28,6 +26,8 @@ static inline bool isReservedName(llvm::StringRef Name) {
          (isUppercase(Name[1]) || Name[1] == '_');
 }
 
+namespace {
+
 // Helper class to iterate over the designator names of an aggregate type.
 //
 // For an array type, yields [0], [1], [2]...
@@ -112,6 +112,8 @@ private:
   RecordDecl::field_iterator FieldsEnd;
 };
 
+} // namespace
+
 // Collect designator labels describing the elements of an init list.
 //
 // This function contributes the designators of some (sub)object, which is
@@ -127,10 +129,9 @@ private:
 // '.a:' is produced directly without recursing into the written sublist.
 // (The written sublist will have a separate collectDesignators() call later).
 // Recursion with Prefix='.b' and Sem = {3, ImplicitValue} produces '.b.x:'.
-void collectDesignators(const InitListExpr *Sem,
-                        llvm::DenseMap<SourceLocation, std::string> &Out,
-                        const llvm::DenseSet<SourceLocation> &NestedBraces,
-                        std::string &Prefix) {
+static void collectDesignators(
+    const InitListExpr *Sem, llvm::DenseMap<SourceLocation, std::string> &Out,
+    const llvm::DenseSet<SourceLocation> &NestedBraces, std::string &Prefix) {
   if (!Sem || Sem->isTransparent())
     return;
   assert(Sem->isSemanticForm());
@@ -170,8 +171,6 @@ void collectDesignators(const InitListExpr *Sem,
   }
 }
 
-} // namespace
-
 llvm::DenseMap<SourceLocation, std::string>
 getUnwrittenDesignators(const InitListExpr *Syn) {
   assert(Syn->isSyntacticForm());
diff --git a/clang/docs/ClangFormatStyleOptions.rst b/clang/docs/ClangFormatStyleOptions.rst
index b746df5..570cab2 100644
--- a/clang/docs/ClangFormatStyleOptions.rst
+++ b/clang/docs/ClangFormatStyleOptions.rst
@@ -245,7 +245,7 @@ the configuration (without a prefix: ``Auto``).
     .. note::
 
      This currently only applies to braced initializer lists (when
-     ``Cpp11BracedListStyle`` is ``true``) and parentheses.
+     ``Cpp11BracedListStyle`` is not ``Block``) and parentheses.
 
 
 
@@ -3816,29 +3816,72 @@ the configuration (without a prefix: ``Auto``).
 
 .. _Cpp11BracedListStyle:
 
-**Cpp11BracedListStyle** (``Boolean``) :versionbadge:`clang-format 3.4` :ref:`¶ <Cpp11BracedListStyle>`
-  If ``true``, format braced lists as best suited for C++11 braced
-  lists.
+**Cpp11BracedListStyle** (``BracedListStyle``) :versionbadge:`clang-format 3.4` :ref:`¶ <Cpp11BracedListStyle>`
+  The style to handle braced lists.
 
-  Important differences:
+  Possible values:
 
-  * No spaces inside the braced list.
-  * No line break before the closing brace.
-  * Indentation with the continuation indent, not with the block indent.
+  * ``BLS_Block`` (in configuration: ``Block``)
+    Best suited for pre C++11 braced lists.
 
-  Fundamentally, C++11 braced lists are formatted exactly like function
-  calls would be formatted in their place. If the braced list follows a name
-  (e.g. a type or variable name), clang-format formats as if the ``{}`` were
-  the parentheses of a function call with that name. If there is no name,
-  a zero-length name is assumed.
+    * Spaces inside the braced list.
+    * Line break before the closing brace.
+    * Indentation with the block indent.
+
+
+    .. code-block:: c++
+
+       vector<int> x{ 1, 2, 3, 4 };
+       vector<T> x{ {}, {}, {}, {} };
+       f(MyMap[{ composite, key }]);
+       new int[3]{ 1, 2, 3 };
+       Type name{ // Comment
+                  value
+       };
+
+  * ``BLS_FunctionCall`` (in configuration: ``FunctionCall``)
+    Best suited for C++11 braced lists.
+
+    * No spaces inside the braced list.
+    * No line break before the closing brace.
+    * Indentation with the continuation indent.
+
+    Fundamentally, C++11 braced lists are formatted exactly like function
+    calls would be formatted in their place. If the braced list follows a
+    name (e.g. a type or variable name), clang-format formats as if the
+    ``{}`` were the parentheses of a function call with that name. If there
+    is no name, a zero-length name is assumed.
+
+    .. code-block:: c++
+
+       vector<int> x{1, 2, 3, 4};
+       vector<T> x{{}, {}, {}, {}};
+       f(MyMap[{composite, key}]);
+       new int[3]{1, 2, 3};
+       Type name{ // Comment
+           value};
+
+  * ``BLS_AlignFirstComment`` (in configuration: ``AlignFirstComment``)
+    Same as ``FunctionCall``, except for the handling of a comment at the
+    begin, it then aligns everything following with the comment.
+
+    * No spaces inside the braced list. (Even for a comment at the first
+      position.)
+    * No line break before the closing brace.
+    * Indentation with the continuation indent, except when followed by a
+      line comment, then it uses the block indent.
+
+
+    .. code-block:: c++
+
+       vector<int> x{1, 2, 3, 4};
+       vector<T> x{{}, {}, {}, {}};
+       f(MyMap[{composite, key}]);
+       new int[3]{1, 2, 3};
+       Type name{// Comment
+                 value};
 
-  .. code-block:: c++
 
-     true:                                  false:
-     vector<int> x{1, 2, 3, 4};     vs.     vector<int> x{ 1, 2, 3, 4 };
-     vector<T> x{{}, {}, {}, {}};           vector<T> x{ {}, {}, {}, {} };
-     f(MyMap[{composite, key}]);            f(MyMap[{ composite, key }]);
-     new int[3]{1, 2, 3};                   new int[3]{ 1, 2, 3 };
 
 .. _DeriveLineEnding:
 
@@ -6625,7 +6668,7 @@ the configuration (without a prefix: ``Auto``).
   .. note::
 
    This option doesn't apply to initializer braces if
-   ``Cpp11BracedListStyle`` is set to ``true``.
+   ``Cpp11BracedListStyle`` is not ``Block``.
 
   Possible values:
 
diff --git a/clang/docs/InternalsManual.rst b/clang/docs/InternalsManual.rst
index c677ddfa..eff46ab 100644
--- a/clang/docs/InternalsManual.rst
+++ b/clang/docs/InternalsManual.rst
@@ -10,7 +10,7 @@ Introduction
 
 This document describes some of the more important APIs and internal design
 decisions made in the Clang C front-end.  The purpose of this document is to
-both capture some of this high level information and also describe some of the
+both capture some of this high-level information and also describe some of the
 design decisions behind it.  This is meant for people interested in hacking on
 Clang, not for end-users.  The description below is categorized by libraries,
 and does not describe any of the clients of the libraries.
@@ -20,7 +20,7 @@ LLVM Support Library
 
 The LLVM ``libSupport`` library provides many underlying libraries and
 `data-structures <https://llvm.org/docs/ProgrammersManual.html>`_, including
-command line option processing, various containers and a system abstraction
+command line option processing, various containers, and a system abstraction
 layer, which is used for file system access.
 
 The Clang "Basic" Library
@@ -34,7 +34,7 @@ and information about the subset of the language being compiled for.
 Part of this infrastructure is specific to C (such as the ``TargetInfo``
 class), other parts could be reused for other non-C-based languages
 (``SourceLocation``, ``SourceManager``, ``Diagnostics``, ``FileManager``).
-When and if there is future demand we can figure out if it makes sense to
+When and if there is future demand, we can figure out if it makes sense to
 introduce a new library, move the general classes somewhere else, or introduce
 some other solution.
 
@@ -96,7 +96,7 @@ The ``EXTENSION`` and ``EXTWARN`` severities are used for extensions to the
 language that Clang accepts.  This means that Clang fully understands and can
 represent them in the AST, but we produce diagnostics to tell the user their
 code is non-portable.  The difference is that the former are ignored by
-default, and the later warn by default.  The ``WARNING`` severity is used for
+default, and the latter warn by default.  The ``WARNING`` severity is used for
 constructs that are valid in the currently selected source language but that
 are dubious in some way.  The ``REMARK`` severity provides generic information
 about the compilation that is not necessarily related to any dubious code.  The
@@ -106,7 +106,7 @@ These *severities* are mapped into a smaller set (the ``Diagnostic::Level``
 enum, {``Ignored``, ``Note``, ``Remark``, ``Warning``, ``Error``, ``Fatal``}) of
 output
 *levels* by the diagnostics subsystem based on various configuration options.
-Clang internally supports a fully fine grained mapping mechanism that allows
+Clang internally supports a fully fine-grained mapping mechanism that allows
 you to map almost any diagnostic to the output level that you want.  The only
 diagnostics that cannot be mapped are ``NOTE``\ s, which always follow the
 severity of the previously emitted diagnostic and ``ERROR``\ s, which can only
@@ -116,18 +116,18 @@ example).
 Diagnostic mappings are used in many ways.  For example, if the user specifies
 ``-pedantic``, ``EXTENSION`` maps to ``Warning``, if they specify
 ``-pedantic-errors``, it turns into ``Error``.  This is used to implement
-options like ``-Wunused_macros``, ``-Wundef`` etc.
+options like ``-Wunused_macros``, ``-Wundef``, etc.
 
 Mapping to ``Fatal`` should only be used for diagnostics that are considered so
 severe that error recovery won't be able to recover sensibly from them (thus
-spewing a ton of bogus errors).  One example of this class of error are failure
+spewing a ton of bogus errors).  One example of this class of error is failure
 to ``#include`` a file.
 
 Diagnostic Wording
 ^^^^^^^^^^^^^^^^^^
 The wording used for a diagnostic is critical because it is the only way for a
 user to know how to correct their code. Use the following suggestions when
-wording a diagnostic.
+wording a diagnostic:
 
 * Diagnostics in Clang do not start with a capital letter and do not end with
   punctuation.
@@ -162,7 +162,7 @@ wording a diagnostic.
   cannot be null in well-defined C++ code``.
 * Prefer diagnostic wording without contractions whenever possible. The single
   quote in a contraction can be visually distracting due to its use with
-  syntactic constructs and contractions can be harder to understand for non-
+  syntactic constructs, and contractions can be harder to understand for non-
   native English speakers.
 
 The Format String
@@ -195,14 +195,14 @@ the C++ code that :ref:`produces them <internals-producing-diag>`, and are
 referenced by ``%0`` .. ``%9``.  If you have more than 10 arguments to your
 diagnostic, you are doing something wrong :).  Unlike ``printf``, there is no
 requirement that arguments to the diagnostic end up in the output in the same
-order as they are specified, you could have a format string with "``%1 %0``"
+order as they are specified; you could have a format string with "``%1 %0``"
 that swaps them, for example.  The text in between the percent and digit are
 formatting instructions.  If there are no instructions, the argument is just
 turned into a string and substituted in.
 
 Here are some "best practices" for writing the English format string:
 
-* Keep the string short.  It should ideally fit in the 80 column limit of the
+* Keep the string short.  It should ideally fit in the 80-column limit of the
   ``DiagnosticKinds.td`` file.  This avoids the diagnostic wrapping when
   printed, and forces you to think about the important point you are conveying
   with the diagnostic.
@@ -227,7 +227,7 @@ used to achieve this sort of thing in a localizable way, see below.
 Formatting a Diagnostic Argument
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-Arguments to diagnostics are fully typed internally, and come from a couple
+Arguments to diagnostics are fully typed internally and come from a couple of
 different classes: integers, types, names, and random strings.  Depending on
 the class of the argument, it can be optionally formatted in different ways.
 This gives the ``DiagnosticConsumer`` information about what the argument means
@@ -268,7 +268,7 @@ Description:
   This format specifier is used to merge multiple related diagnostics together
   into one common one, without requiring the difference to be specified as an
   English string argument.  Instead of specifying the string, the diagnostic
-  gets an integer argument and the format string selects the numbered option.
+  gets an integer argument, and the format string selects the numbered option.
   In this case, the "``%0``" value must be an integer in the range [0..2].  If
   it is 0, it prints "unary", if it is 1 it prints "binary" if it is 2, it
   prints "unary or binary".  This allows other language translations to
@@ -287,7 +287,7 @@ Description:
   additionally generates a namespace, enumeration, and enumerator list based on
   the format string given. In the above case, a namespace is generated named
   ``FrobbleKind`` that has an unscoped enumeration with the enumerators
-  ``VarDecl`` and ``FuncDecl`` which correspond to the values 0 and 1. This
+  ``VarDecl`` and ``FuncDecl``, which correspond to the values 0 and 1. This
   permits a clearer use of the ``Diag`` in source code, as the above could be
   called as: ``Diag(Loc, diag::frobble) << diag::FrobbleKind::VarDecl``.
 
@@ -407,7 +407,7 @@ Example:
     def note_ovl_candidate : Note<
       "candidate %sub{select_ovl_candidate}3,2,1 not viable">;
 
-  and will act as if it was written
+  and will act as if it were written
   ``"candidate %select{function|constructor}3%select{| template| %1}2 not viable"``.
 Description:
   This format specifier is used to avoid repeating strings verbatim in multiple
@@ -447,7 +447,7 @@ For example, the binary expression error comes from code like this:
       << lex->getType() << rex->getType()
       << lex->getSourceRange() << rex->getSourceRange();
 
-This shows that use of the ``Diag`` method: it takes a location (a
+This shows the use of the ``Diag`` method: it takes a location (a
 :ref:`SourceLocation <SourceLocation>` object) and a diagnostic enum value
 (which matches the name from ``Diagnostic*Kinds.td``).  If the diagnostic takes
 arguments, they are specified with the ``<<`` operator: the first argument
@@ -586,7 +586,7 @@ Strangely enough, the ``SourceLocation`` class represents a location within the
 source code of the program.  Important design points include:
 
 #. ``sizeof(SourceLocation)`` must be extremely small, as these are embedded
-   into many AST nodes and are passed around often.  Currently it is 32 bits.
+   into many AST nodes and are passed around often.  Currently, it is 32 bits.
 #. ``SourceLocation`` must be a simple value object that can be efficiently
    copied.
 #. We should be able to represent a source location for any byte of any input
@@ -605,7 +605,7 @@ In practice, the ``SourceLocation`` works together with the ``SourceManager``
 class to encode two pieces of information about a location: its spelling
 location and its expansion location.  For most tokens, these will be the
 same.  However, for a macro expansion (or tokens that came from a ``_Pragma``
-directive) these will describe the location of the characters corresponding to
+directive), these will describe the location of the characters corresponding to
 the token and the location where the token was used (i.e., the macro
 expansion point or the location of the ``_Pragma`` itself).
 
@@ -621,7 +621,7 @@ token.  This concept maps directly to the "spelling location" for the token.
 .. mostly taken from https://discourse.llvm.org/t/code-ranges-of-tokens-ast-elements/16893/2
 
 Clang represents most source ranges by [first, last], where "first" and "last"
-each point to the beginning of their respective tokens.  For example consider
+each point to the beginning of their respective tokens.  For example, consider
 the ``SourceRange`` of the following statement:
 
 .. code-block:: text
@@ -632,7 +632,7 @@ the ``SourceRange`` of the following statement:
 To map from this representation to a character-based representation, the "last"
 location needs to be adjusted to point to (or past) the end of that token with
 either ``Lexer::MeasureTokenLength()`` or ``Lexer::getLocForEndOfToken()``.  For
-the rare cases where character-level source ranges information is needed we use
+the rare cases where character-level source ranges information is needed, we use
 the ``CharSourceRange`` class.
 
 The Driver Library
@@ -651,17 +651,17 @@ The Frontend Library
 ====================
 
 The Frontend library contains functionality useful for building tools on top of
-the Clang libraries, for example several methods for outputting diagnostics.
+the Clang libraries, including several methods for outputting diagnostics.
 
 Compiler Invocation
 -------------------
 
 One of the classes provided by the Frontend library is ``CompilerInvocation``,
-which holds information that describe current invocation of the Clang ``-cc1``
+which holds information that describes the current invocation of the Clang ``-cc1``
 frontend. The information typically comes from the command line constructed by
 the Clang driver or from clients performing custom initialization. The data
 structure is split into logical units used by different parts of the compiler,
-for example ``PreprocessorOptions``, ``LanguageOptions`` or ``CodeGenOptions``.
+for example, ``PreprocessorOptions``, ``LanguageOptions``, or ``CodeGenOptions``.
 
 Command Line Interface
 ----------------------
@@ -698,7 +698,7 @@ Adding new Command Line Option
 ------------------------------
 
 When adding a new command line option, the first place of interest is the header
-file declaring the corresponding options class (e.g. ``CodeGenOptions.h`` for
+file declaring the corresponding options class (e.g., ``CodeGenOptions.h`` for
 command line option that affects the code generation). Create new member
 variable for the option value:
 
@@ -739,7 +739,7 @@ The helper classes take a list of acceptable prefixes of the option (e.g.
 Then, specify additional attributes via mix-ins:
 
 * ``HelpText`` holds the text that will be printed besides the option name when
-  the user requests help (e.g. via ``clang --help``).
+  the user requests help (e.g., via ``clang --help``).
 * ``Group`` specifies the "category" of options this option belongs to. This is
   used by various tools to categorize and sometimes filter options.
 * ``Flags`` may contain "tags" associated with the option. These may affect how
@@ -779,7 +779,7 @@ use them to construct the ``-cc1`` job:
     }
 
 The last step is implementing the ``-cc1`` command line argument
-parsing/generation that initializes/serializes the option class (in our case
+parsing/generation that initializes/serializes the option class (in our case,
 ``CodeGenOptions``) stored within ``CompilerInvocation``. This can be done
 automatically by using the marshalling annotations on the option definition:
 
@@ -946,13 +946,13 @@ described below. All of them take a key path argument and possibly other
 information required for parsing or generating the command line argument.
 
 **Note:** The marshalling infrastructure is not intended for driver-only
-options. Only options of the ``-cc1`` frontend need to be marshalled to/from
+options. Only options of the ``-cc1`` frontend need to be marshalled to/from a
 ``CompilerInvocation`` instance.
 
 **Positive Flag**
 
 The key path defaults to ``false`` and is set to ``true`` when the flag is
-present on command line.
+present on the command line.
 
 .. code-block:: text
 
@@ -963,7 +963,7 @@ present on command line.
 **Negative Flag**
 
 The key path defaults to ``true`` and is set to ``false`` when the flag is
-present on command line.
+present on the command line.
 
 .. code-block:: text
 
@@ -1041,7 +1041,7 @@ and the result is assigned to the key path on success.
 
 The key path defaults to the value specified in ``MarshallingInfoEnum`` prefixed
 by the contents of ``NormalizedValuesScope`` and ``::``. This ensures correct
-reference to an enum case is formed even if the enum resides in different
+reference to an enum case is formed even if the enum resides in a different
 namespace or is an enum class. If the value present on the command line does not
 match any of the comma-separated values from ``Values``, an error diagnostic is
 issued. Otherwise, the corresponding element from ``NormalizedValues`` at the
@@ -1410,7 +1410,7 @@ or a clear engineering tradeoff -- should desugar minimally and wrap the result
 in a construct representing the original source form.
 
 For example, ``CXXForRangeStmt`` directly represents the syntactic form of a
-range-based for statement, but also holds a semantic representation of the
+range-based for statement but also holds a semantic representation of the
 range declaration and iterator declarations. It does not contain a
 fully-desugared ``ForStmt``, however.
 
@@ -1425,7 +1425,7 @@ with the same or similar semantics.
 The ``Type`` class and its subclasses
 -------------------------------------
 
-The ``Type`` class (and its subclasses) are an important part of the AST.
+The ``Type`` class (and its subclasses) is an important part of the AST.
 Types are accessed through the ``ASTContext`` class, which implicitly creates
 and uniques them as they are needed.  Types have a couple of non-obvious
 features: 1) they do not capture type qualifiers like ``const`` or ``volatile``
@@ -1474,7 +1474,7 @@ various operators (for example, the type of ``*Y`` is "``foo``", not
 is an instance of the ``TypedefType`` class, which indicates that the type of
 these expressions is a typedef for "``foo``".
 
-Representing types like this is great for diagnostics, because the
+Representing types like this is great for diagnostics because the
 user-specified type is always immediately available.  There are two problems
 with this: first, various semantic checks need to make judgements about the
 *actual structure* of a type, ignoring typedefs.  Second, we need an efficient
@@ -1521,7 +1521,7 @@ know it exists.  To continue the example, the result type of the indirection
 operator is the pointee type of the subexpression.  In order to determine the
 type, we need to get the instance of ``PointerType`` that best captures the
 typedef information in the program.  If the type of the expression is literally
-a ``PointerType``, we can return that, otherwise we have to dig through the
+a ``PointerType``, we can return that; otherwise, we have to dig through the
 typedefs to find the pointer type.  For example, if the subexpression had type
 "``foo*``", we could return that type as the result.  If the subexpression had
 type "``bar``", we want to return "``foo*``" (note that we do *not* want
@@ -1552,7 +1552,7 @@ that sets a bit), and remove one or more type qualifiers (just return a
 ``QualType`` with the bitfield set to empty).
 
 Further, because the bits are stored outside of the type itself, we do not need
-to create duplicates of types with different sets of qualifiers (i.e. there is
+to create duplicates of types with different sets of qualifiers (i.e., there is
 only a single heap allocated "``int``" type: "``const int``" and "``volatile
 const int``" both point to the same heap allocated "``int``" type).  This
 reduces the heap size used to represent bits and also means we do not have to
@@ -1972,7 +1972,7 @@ and optimize code for it, but it's used as parsing continues to detect further
 errors in the input. Clang-based tools also depend on such ASTs, and IDEs in
 particular benefit from a high-quality AST for broken code.
 
-In presence of errors, clang uses a few error-recovery strategies to present the
+In the presence of errors, clang uses a few error-recovery strategies to present the
 broken code in the AST:
 
 - correcting errors: in cases where clang is confident about the fix, it
@@ -1981,7 +1981,7 @@ broken code in the AST:
   provide more accurate subsequent diagnostics. Typo correction is a typical
   example.
 - representing invalid node: the invalid node is preserved in the AST in some
-  form, e.g. when the "declaration" part of the declaration contains semantic
+  form, e.g., when the "declaration" part of the declaration contains semantic
   errors, the Decl node is marked as invalid.
 - dropping invalid node: this often happens for errors that we don’t have
   graceful recovery. Prior to Recovery AST, a mismatched-argument function call
@@ -1994,9 +1994,9 @@ for broken code.
 Recovery AST
 ^^^^^^^^^^^^
 
-The idea of Recovery AST is to use recovery nodes which act as a placeholder to
+The idea of Recovery AST is to use recovery nodes, which act as a placeholder to
 maintain the rough structure of the parsing tree, preserve locations and
-children but have no language semantics attached to them.
+children, but have no language semantics attached to them.
 
 For example, consider the following mismatched function call:
 
@@ -2031,10 +2031,10 @@ With Recovery AST, the AST looks like:
           `-DeclRefExpr <col:9> 'int' lvalue ParmVar 'abc' 'int'
 
 
-An alternative is to use existing Exprs, e.g. CallExpr for the above example.
-This would capture more call details (e.g. locations of parentheses) and allow
+An alternative is to use existing Exprs, e.g., CallExpr for the above example.
+This would capture more call details (e.g., locations of parentheses) and allow
 it to be treated uniformly with valid CallExprs. However, jamming the data we
-have into CallExpr forces us to weaken its invariants, e.g. arg count may be
+have into CallExpr forces us to weaken its invariants, e.g., arg count may be
 wrong. This would introduce a huge burden on consumers of the AST to handle such
 "impossible" cases. So when we're representing (rather than correcting) errors,
 we use a distinct recovery node type with extremely weak invariants instead.
@@ -2048,7 +2048,7 @@ Types and dependence
 ^^^^^^^^^^^^^^^^^^^^
 
 ``RecoveryExpr`` is an ``Expr``, so it must have a type. In many cases the true
-type can't really be known until the code is corrected (e.g. a call to a
+type can't really be known until the code is corrected (e.g., a call to a
 function that doesn't exist). And it means that we can't properly perform type
 checks on some containing constructs, such as ``return 42 + unknownFunction()``.
 
@@ -2058,7 +2058,7 @@ mean dependence on a template parameter or how an error is repaired. The
 ``DependentTy``, and this suppresses type-based analysis in the same way it
 would inside a template.
 
-In cases where we are confident about the concrete type (e.g. the return type
+In cases where we are confident about the concrete type (e.g., the return type
 for a broken non-overloaded function call), the ``RecoveryExpr`` will have this
 type. This allows more code to be typechecked, and produces a better AST and
 more diagnostics. For example:
@@ -2071,7 +2071,7 @@ more diagnostics. For example:
 Whether or not the ``RecoveryExpr`` has a dependent type, it is always
 considered value-dependent, because its value isn't well-defined until the error
 is resolved. Among other things, this means that clang doesn't emit more errors
-where a RecoveryExpr is used as a constant (e.g. array size), but also won't try
+where a RecoveryExpr is used as a constant (e.g., array size), but also won't try
 to evaluate it.
 
 ContainsErrors bit
@@ -2122,7 +2122,7 @@ cycles. One example of a cycle is the connection between a
 ``ClassTemplateDecl`` and its "templated" ``CXXRecordDecl``. The *templated*
 ``CXXRecordDecl`` represents all the fields and methods inside the class
 template, while the ``ClassTemplateDecl`` holds the information which is
-related to being a template, i.e. template arguments, etc. We can get the
+related to being a template, i.e., template arguments, etc. We can get the
 *templated* class (the ``CXXRecordDecl``) of a ``ClassTemplateDecl`` with
 ``ClassTemplateDecl::getTemplatedDecl()``. And we can get back a pointer of the
 "described" class template from the *templated* class:
@@ -2145,7 +2145,7 @@ we skip the copy.
 The informal definition of structural equivalency is the following:
 Two nodes are **structurally equivalent** if they are
 
-- builtin types and refer to the same type, e.g. ``int`` and ``int`` are
+- builtin types and refer to the same type, e.g., ``int`` and ``int`` are
   structurally equivalent,
 - function types and all their parameters have structurally equivalent types,
 - record types and all their fields in order of their definition have the same
@@ -2162,7 +2162,7 @@ mentioned properties, we have to check for equivalent template
 parameters/arguments, etc.
 
 The structural equivalent check can be and is used independently from the
-ASTImporter, e.g. the ``clang::Sema`` class uses it also.
+ASTImporter, e.g., the ``clang::Sema`` class uses it also.
 
 The equivalence of nodes may depend on the equivalency of other pairs of nodes.
 Thus, the check is implemented as a parallel graph traversal. We traverse
@@ -2195,7 +2195,7 @@ Redeclaration Chains
 ^^^^^^^^^^^^^^^^^^^^
 
 The early version of the ``ASTImporter``'s merge mechanism squashed the
-declarations, i.e. it aimed to have only one declaration instead of maintaining
+declarations, i.e., it aimed to have only one declaration instead of maintaining
 a whole redeclaration chain. This early approach simply skipped importing a
 function prototype, but it imported a definition. To demonstrate the problem
 with this approach let's consider an empty "to" context and the following
@@ -2225,7 +2225,7 @@ another definition, we will use the existing definition. However, we can import
 prototype(s): we chain the newly imported prototype(s) to the existing
 definition. Whenever we import a new prototype from a third context, that will
 be added to the end of the redeclaration chain. This may result in long
-redeclaration chains in certain cases, e.g. if we import from several
+redeclaration chains in certain cases, e.g., if we import from several
 translation units which include the same header with the prototype.
 
 .. Squashing prototypes
@@ -2290,7 +2290,7 @@ Traversal during the Import
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 The node specific import mechanisms are implemented in
-``ASTNodeImporter::VisitNode()`` functions, e.g. ``VisitFunctionDecl()``.
+``ASTNodeImporter::VisitNode()`` functions, e.g., ``VisitFunctionDecl()``.
 When we import a declaration then first we import everything which is needed to
 call the constructor of that declaration node. Everything which can be set
 later is set after the node is created. For example, in case of  a
@@ -2490,7 +2490,7 @@ In case of LLDB, an implementation of the ``ExternalASTSource`` interface is
 attached to the AST context which is related to the parsed expression. This
 implementation of the ``ExternalASTSource`` interface is realized with the help
 of the ``ASTImporter`` class. This way, LLDB can reuse Clang's parsing
-machinery while synthesizing the underlying AST from the debug data (e.g. from
+machinery while synthesizing the underlying AST from the debug data (e.g., from
 DWARF). From the view of the ``ASTImporter`` this means both the "to" and the
 "from" context may have declaration contexts with external lexical storage. If
 a ``DeclContext`` in the "to" AST context has external lexical storage then we
@@ -2573,7 +2573,7 @@ conflict error (ODR violation in C++). In this case, we return with an
 clients of the ``ASTImporter`` may require a different, perhaps less
 conservative and more liberal error handling strategy.
 
-E.g. static analysis clients may benefit if the node is created even if there
+E.g., static analysis clients may benefit if the node is created even if there
 is a name conflict. During the CTU analysis of certain projects, we recognized
 that there are global declarations which collide with declarations from other
 translation units, but they are not referenced outside from their translation
@@ -2916,7 +2916,7 @@ Any error during satisfaction is recorded in ``ConstraintSatisfaction``.
 for nested requirements, ``ConstraintSatisfaction`` is stored (including
 diagnostics) in the AST, which is something we might want to improve.
 
-When an atomic constraint is not satified, we try to substitute into any
+When an atomic constraint is not satisfied, we try to substitute into any
 enclosing concept-id using the same mechanism described above, for
 diagnostics purpose, and inject that in the ``ConstraintSatisfaction``.
 
@@ -3584,7 +3584,7 @@ be specified by appending a ``+`` to the number. For example:
   void f(); // expected-note 0+ {{previous declaration is here}}
   void g(); // expected-note 1+ {{previous declaration is here}}
 
-In the first example, the diagnostic becomes optional, i.e. it will be
+In the first example, the diagnostic becomes optional, i.e., it will be
 swallowed if it occurs, but will not generate an error if it does not occur. In
 the second example, the diagnostic must occur at least once. As a short-hand,
 "one or more" can be specified simply by ``+``. For example:
diff --git a/clang/include/clang/Basic/BuiltinsX86.td b/clang/include/clang/Basic/BuiltinsX86.td
index 62c70fba..d03c778 100644
--- a/clang/include/clang/Basic/BuiltinsX86.td
+++ b/clang/include/clang/Basic/BuiltinsX86.td
@@ -124,13 +124,13 @@ let Attributes = [Const, NoThrow, RequiredVectorWidth<128>] in {
   }
 
   let Features = "ssse3" in {
-    def pmulhrsw128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
     def psignb128 : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>)">;
     def psignw128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
     def psignd128 : X86Builtin<"_Vector<4, int>(_Vector<4, int>, _Vector<4, int>)">;
   }
 
   let Features = "ssse3", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
+    def pmulhrsw128 : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>)">;
     def pmaddubsw128 : X86Builtin<"_Vector<8, short>(_Vector<16, char>, _Vector<16, char>)">;
     def pshufb128 : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>)">;
   }
@@ -608,7 +608,6 @@ let Features = "avx2", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] i
   def palignr256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>, _Constant int)">;
 
   def pmovmskb256 : X86Builtin<"int(_Vector<32, char>)">;
-  def pmulhrsw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
   def psadbw256 : X86Builtin<"_Vector<4, long long int>(_Vector<32, char>, _Vector<32, char>)">;
   def psignb256 : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>)">;
   def psignw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
@@ -661,6 +660,7 @@ let Features = "avx2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWi
   def psrawi256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, int)">;
   def psradi256 : X86Builtin<"_Vector<8, int>(_Vector<8, int>, int)">;
 
+  def pmulhrsw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
   def pmulhuw256 : X86Builtin<"_Vector<16, unsigned short>(_Vector<16, unsigned short>, _Vector<16, unsigned short>)">;
   def pmulhw256 : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>)">;
 
@@ -1386,13 +1386,10 @@ let Features = "avx512bitalg", Attributes = [NoThrow, Const, RequiredVectorWidth
   def vpshufbitqmb512_mask : X86Builtin<"unsigned long long int(_Vector<64, char>, _Vector<64, char>, unsigned long long int)">;
 }
 
-let Features = "avx512bw", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in {
-  def pmulhrsw512 : X86Builtin<"_Vector<32, short>(_Vector<32, short>, _Vector<32, short>)">;
-}
-
 let Features = "avx512bw", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in {
   def pavgb512 : X86Builtin<"_Vector<64, unsigned char>(_Vector<64, unsigned char>, _Vector<64, unsigned char>)">;
   def pavgw512 : X86Builtin<"_Vector<32, unsigned short>(_Vector<32, unsigned short>, _Vector<32, unsigned short>)">;
+  def pmulhrsw512 : X86Builtin<"_Vector<32, short>(_Vector<32, short>, _Vector<32, short>)">;
   def pmulhuw512 : X86Builtin<"_Vector<32, unsigned short>(_Vector<32, unsigned short>, _Vector<32, unsigned short>)">;
   def pmulhw512 : X86Builtin<"_Vector<32, short>(_Vector<32, short>, _Vector<32, short>)">;
 }
diff --git a/clang/include/clang/Basic/riscv_vector.td b/clang/include/clang/Basic/riscv_vector.td
index 07a8724..96d8300 100644
--- a/clang/include/clang/Basic/riscv_vector.td
+++ b/clang/include/clang/Basic/riscv_vector.td
@@ -1013,9 +1013,9 @@ let ManualCodegen = [{
 }] in {
   let HasFRMRoundModeOp = true in {
     // 13.2. Vector Single-Width Floating-Point Add/Subtract Instructions
-    defm vfadd  : RVVFloatingBinBuiltinSetRoundingMode;
-    defm vfsub  : RVVFloatingBinBuiltinSetRoundingMode;
-    defm vfrsub : RVVFloatingBinVFBuiltinSetRoundingMode;
+    defm vfadd  : RVVFloatingBinBuiltinSetRoundingMode<HasBF=1>;
+    defm vfsub  : RVVFloatingBinBuiltinSetRoundingMode<HasBF=1>;
+    defm vfrsub : RVVFloatingBinVFBuiltinSetRoundingMode<HasBF=1>;
 
     // 13.3. Vector Widening Floating-Point Add/Subtract Instructions
     // Widening FP add/subtract, 2*SEW = 2*SEW +/- SEW
@@ -1023,14 +1023,14 @@ let ManualCodegen = [{
     defm vfwsub : RVVFloatingWidenOp0BinBuiltinSetRoundingMode;
 
     // 13.4. Vector Single-Width Floating-Point Multiply/Divide Instructions
-    defm vfmul  : RVVFloatingBinBuiltinSetRoundingMode;
+    defm vfmul  : RVVFloatingBinBuiltinSetRoundingMode<HasBF=1>;
     defm vfdiv  : RVVFloatingBinBuiltinSetRoundingMode;
     defm vfrdiv : RVVFloatingBinVFBuiltinSetRoundingMode;
   }
   // 13.2. Vector Single-Width Floating-Point Add/Subtract Instructions
-  defm vfadd  : RVVFloatingBinBuiltinSet;
-  defm vfsub  : RVVFloatingBinBuiltinSet;
-  defm vfrsub : RVVFloatingBinVFBuiltinSet;
+  defm vfadd  : RVVFloatingBinBuiltinSet<HasBF=1>;
+  defm vfsub  : RVVFloatingBinBuiltinSet<HasBF=1>;
+  defm vfrsub : RVVFloatingBinVFBuiltinSet<HasBF=1>;
 
   // 13.3. Vector Widening Floating-Point Add/Subtract Instructions
   // Widening FP add/subtract, 2*SEW = 2*SEW +/- SEW
@@ -1038,7 +1038,7 @@ let ManualCodegen = [{
   defm vfwsub : RVVFloatingWidenOp0BinBuiltinSet;
 
   // 13.4. Vector Single-Width Floating-Point Multiply/Divide Instructions
-  defm vfmul  : RVVFloatingBinBuiltinSet;
+  defm vfmul  : RVVFloatingBinBuiltinSet<HasBF=1>;
   defm vfdiv  : RVVFloatingBinBuiltinSet;
   defm vfrdiv : RVVFloatingBinVFBuiltinSet;
 }
@@ -1065,6 +1065,10 @@ let ManualCodegen = [{
         defm vfwmul : RVVOutOp0Op1BuiltinSet<"vfwmul", "x",
                                              [["vv", "w", "wvvu"],
                                               ["vf", "w", "wveu"]]>;
+      let RequiredFeatures = ["zvfbfa"] in
+        defm vfwmul : RVVOutOp0Op1BuiltinSet<"vfwmul", "y",
+                                             [["vv", "vw", "wvvu"],
+                                              ["vf", "vw", "wveu"]]>;
     }
   }
   // 13.3. Vector Widening Floating-Point Add/Subtract Instructions
@@ -1081,6 +1085,10 @@ let ManualCodegen = [{
       defm vfwmul : RVVOutOp0Op1BuiltinSet<"vfwmul", "x",
                                            [["vv", "w", "wvv"],
                                             ["vf", "w", "wve"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm vfwmul : RVVOutOp0Op1BuiltinSet<"vfwmul", "y",
+                                           [["vv", "vw", "wvv"],
+                                            ["vf", "vw", "wve"]]>;
   }
 }
 }
@@ -1170,6 +1178,8 @@ let ManualCodegen = [{
     defm vfrec7 : RVVOutBuiltinSet<"vfrec7", "fd", [["v", "v", "vvu"]]>;
     let RequiredFeatures = ["zvfh"] in
       defm vfrec7 : RVVOutBuiltinSet<"vfrec7", "x", [["v", "v", "vvu"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm vfrec7 : RVVOutBuiltinSet<"vfrec7", "y", [["v", "v", "vvu"]]>;
   }
   // 13.8. Vector Floating-Point Square-Root Instruction
   defm vfsqrt : RVVOutBuiltinSet<"vfsqrt", "fd", [["v", "v", "vv"]]>;
@@ -1180,21 +1190,26 @@ let ManualCodegen = [{
   defm vfrec7 : RVVOutBuiltinSet<"vfrec7", "fd", [["v", "v", "vv"]]>;
   let RequiredFeatures = ["zvfh"] in
     defm vfrec7 : RVVOutBuiltinSet<"vfrec7", "x", [["v", "v", "vv"]]>;
+  let RequiredFeatures = ["zvfbfa"] in
+    defm vfrec7 : RVVOutBuiltinSet<"vfrec7", "y", [["v", "v", "vv"]]>;
 }
 
 // 13.9. Vector Floating-Point Reciprocal Square-Root Estimate Instruction
 defm vfrsqrt7 : RVVOutBuiltinSet<"vfrsqrt7", "fd", [["v", "v", "vv"]]>;
 let RequiredFeatures = ["zvfh"] in
   defm vfrsqrt7 : RVVOutBuiltinSet<"vfrsqrt7", "x", [["v", "v", "vv"]]>;
+let RequiredFeatures = ["zvfbfa"] in
+  defm vfrsqrt7 : RVVOutBuiltinSet<"vfrsqrt7", "y", [["v", "v", "vv"]]>;
+
 
 // 13.11. Vector Floating-Point MIN/MAX Instructions
-defm vfmin : RVVFloatingBinBuiltinSet;
-defm vfmax : RVVFloatingBinBuiltinSet;
+defm vfmin : RVVFloatingBinBuiltinSet<HasBF=1>;
+defm vfmax : RVVFloatingBinBuiltinSet<HasBF=1>;
 
 // 13.12. Vector Floating-Point Sign-Injection Instructions
-defm vfsgnj  : RVVFloatingBinBuiltinSet;
-defm vfsgnjn : RVVFloatingBinBuiltinSet;
-defm vfsgnjx : RVVFloatingBinBuiltinSet;
+defm vfsgnj  : RVVFloatingBinBuiltinSet<HasBF=1>;
+defm vfsgnjn : RVVFloatingBinBuiltinSet<HasBF=1>;
+defm vfsgnjx : RVVFloatingBinBuiltinSet<HasBF=1>;
 }
 defm vfneg_v : RVVPseudoVFUnaryBuiltin<"vfsgnjn", "fd">;
 let RequiredFeatures = ["zvfh"] in
@@ -1219,6 +1234,8 @@ let UnMaskedPolicyScheme = HasPassthruOperand in {
 defm vfclass : RVVOp0BuiltinSet<"vfclass", "fd", [["v", "Uv", "Uvv"]]>;
 let RequiredFeatures = ["zvfh"] in
   defm vfclass : RVVOp0BuiltinSet<"vfclass", "x", [["v", "Uv", "Uvv"]]>;
+let RequiredFeatures = ["zvfbfa"] in
+  defm vfclass : RVVOp0BuiltinSet<"vfclass", "y", [["v", "vUv", "Uvv"]]>;
 }
 
 // 13.15. Vector Floating-Point Merge Instruction
@@ -1239,6 +1256,9 @@ let HasMasked = false,
   let RequiredFeatures = ["zvfh"] in
     defm vfmerge : RVVOutOp1BuiltinSet<"vfmerge", "x",
                                        [["vfm", "v", "vvem"]]>;
+  let RequiredFeatures = ["zvfbfa"] in
+    defm vfmerge : RVVOutOp1BuiltinSet<"vfmerge", "y",
+                                       [["vfm", "v", "vvem"]]>;
 }
 
 // 13.16. Vector Floating-Point Move Instruction
@@ -1252,6 +1272,9 @@ let HasMasked = false,
   let RequiredFeatures = ["zvfh"] in
     defm vfmv_v : RVVOutBuiltinSet<"vfmv_v_f", "x",
                                    [["f", "v", "ve"]]>;
+  let RequiredFeatures = ["zvfbfa"] in
+    defm vfmv_v : RVVOutBuiltinSet<"vfmv_v_f", "y",
+                                   [["f", "v", "ve"]]>;
 }
 
 // 13.17. Single-Width Floating-Point/Integer Type-Convert Instructions
@@ -1287,10 +1310,16 @@ let Log2LMUL = [-3, -2, -1, 0, 1, 2] in {
       defm : RVVConvBuiltinSet<"vfwcvt_f_x_v", "c", [["Fw", "Fwv"]]>;
     }
   }
+  let RequiredFeatures = ["zvfbfa"], OverloadedName = "vfwcvt_f_bf16" in {
+    defm : RVVConvBuiltinSet<"vfwcvt_f_xu_v", "c", [["Yw", "YwUv"]]>;
+    defm : RVVConvBuiltinSet<"vfwcvt_f_x_v", "c", [["Yw", "Ywv"]]>;
+  }
   let OverloadedName = "vfwcvt_f" in {
     defm : RVVConvBuiltinSet<"vfwcvt_f_f_v", "f", [["w", "wv"]]>;
     let RequiredFeatures = ["zvfhmin"] in
       defm : RVVConvBuiltinSet<"vfwcvt_f_f_v", "x", [["w", "wv"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm : RVVConvBuiltinSet<"vfwcvt_f_f_v", "y", [["vw", "wv"]]>;
   }
 }
 
@@ -1300,17 +1329,23 @@ let Log2LMUL = [-3, -2, -1, 0, 1, 2] in {
     defm : RVVConvBuiltinSet<"vfncvt_rtz_xu_f_w", "si", [["Uv", "UvFw"]]>;
     let RequiredFeatures = ["zvfh"] in
       defm : RVVConvBuiltinSet<"vfncvt_rtz_xu_f_w", "c", [["Uv", "UvFw"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm : RVVConvBuiltinSet<"vfncvt_rtz_xu_f_w", "c", [["YwUv", "UvYw"]]>;
   }
   let OverloadedName = "vfncvt_rtz_x" in {
     defm : RVVConvBuiltinSet<"vfncvt_rtz_x_f_w", "si", [["Iv", "IvFw"]]>;
     let RequiredFeatures = ["zvfh"] in
       defm : RVVConvBuiltinSet<"vfncvt_rtz_x_f_w", "c", [["Iv", "IvFw"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm : RVVConvBuiltinSet<"vfncvt_rtz_x_f_w", "c", [["YwIv", "IvYw"]]>;
   }
   let OverloadedName = "vfncvt_rod_f" in {
     defm : RVVConvBuiltinSet<"vfncvt_rod_f_f_w", "f", [["v", "vw"]]>;
     let RequiredFeatures = ["zvfh"] in
       defm : RVVConvBuiltinSet<"vfncvt_rod_f_f_w", "x", [["v", "vw"]]>;
   }
+  let RequiredFeatures = ["zvfbfa"], OverloadedName = "vfncvt_rod_f_bf16" in
+    defm : RVVConvBuiltinSet<"vfncvt_rod_f_f_w", "y", [["v", "vw"]]>;
 }
 
 // Zvfbfmin - Vector convert BF16 to FP32
@@ -1363,11 +1398,15 @@ let ManualCodegen = [{
         defm : RVVConvBuiltinSet<"vfncvt_x_f_w", "si", [["Iv", "IvFwu"]]>;
         let RequiredFeatures = ["zvfh"] in
           defm : RVVConvBuiltinSet<"vfncvt_x_f_w", "c", [["Iv", "IvFwu"]]>;
+        let RequiredFeatures = ["zvfbfa"] in
+          defm : RVVConvBuiltinSet<"vfncvt_x_f_w", "c", [["YwIv", "IvYwu"]]>;
       }
       let OverloadedName = "vfncvt_xu" in {
         defm : RVVConvBuiltinSet<"vfncvt_xu_f_w", "si", [["Uv", "UvFwu"]]>;
         let RequiredFeatures = ["zvfh"] in
           defm : RVVConvBuiltinSet<"vfncvt_xu_f_w", "c", [["Uv", "UvFwu"]]>;
+        let RequiredFeatures = ["zvfbfa"] in
+          defm : RVVConvBuiltinSet<"vfncvt_xu_f_w", "c", [["YwUv", "UvYwu"]]>;
       }
       let OverloadedName = "vfncvt_f" in {
         defm : RVVConvBuiltinSet<"vfncvt_f_x_w", "f", [["v", "vIwu"]]>;
@@ -1382,6 +1421,8 @@ let ManualCodegen = [{
         let RequiredFeatures = ["zvfhmin"] in
           defm : RVVConvBuiltinSet<"vfncvt_f_f_w", "x", [["v", "vwu"]]>;
       }
+      let RequiredFeatures = ["zvfbfa"], OverloadedName = "vfncvt_f_bf16" in
+        defm : RVVConvBuiltinSet<"vfncvt_f_f_w", "y", [["v", "vwu"]]>;
     }
 
     // Zvfbfmin - Vector convert FP32 to BF16
@@ -1430,11 +1471,15 @@ let ManualCodegen = [{
       defm : RVVConvBuiltinSet<"vfncvt_x_f_w", "si", [["Iv", "IvFw"]]>;
       let RequiredFeatures = ["zvfh"] in
         defm : RVVConvBuiltinSet<"vfncvt_x_f_w", "c", [["Iv", "IvFw"]]>;
+      let RequiredFeatures = ["zvfbfa"] in
+        defm : RVVConvBuiltinSet<"vfncvt_x_f_w", "c", [["YwIv", "IvYw"]]>;
     }
     let OverloadedName = "vfncvt_xu" in {
       defm : RVVConvBuiltinSet<"vfncvt_xu_f_w", "si", [["Uv", "UvFw"]]>;
       let RequiredFeatures = ["zvfh"] in
         defm : RVVConvBuiltinSet<"vfncvt_xu_f_w", "c", [["Uv", "UvFw"]]>;
+      let RequiredFeatures = ["zvfbfa"] in
+        defm : RVVConvBuiltinSet<"vfncvt_xu_f_w", "c", [["YwUv", "UvYw"]]>;
     }
     let OverloadedName = "vfncvt_f" in {
       defm : RVVConvBuiltinSet<"vfncvt_f_x_w", "f", [["v", "vIw"]]>;
@@ -1449,6 +1494,8 @@ let ManualCodegen = [{
       let RequiredFeatures = ["zvfhmin"] in
         defm : RVVConvBuiltinSet<"vfncvt_f_f_w", "x", [["v", "vw"]]>;
     }
+    let RequiredFeatures = ["zvfbfa"], OverloadedName = "vfncvt_f_bf16" in
+      defm : RVVConvBuiltinSet<"vfncvt_f_f_w", "y", [["v", "vw"]]>;
   }
 
   // Zvfbfmin - Vector convert FP32 to BF16
@@ -1578,6 +1625,9 @@ let HasMasked = false, MaskedPolicyScheme = NonePolicy in {
     let RequiredFeatures = ["zvfh"] in
       defm vfmv_f : RVVOp0BuiltinSet<"vfmv_f_s", "x",
                                      [["s", "ve", "ev"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm vfmv_f : RVVOp0BuiltinSet<"vfmv_f_s", "y",
+                                     [["s", "ve", "ev"]]>;
   }
   let OverloadedName = "vfmv_s",
       UnMaskedPolicyScheme = HasPassthruOperand,
@@ -1589,6 +1639,9 @@ let HasMasked = false, MaskedPolicyScheme = NonePolicy in {
       defm vfmv_s : RVVOutBuiltinSet<"vfmv_s_f", "x",
                                      [["f", "v", "ve"],
                                       ["x", "Uv", "UvUe"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm vfmv_s : RVVOutBuiltinSet<"vfmv_s_f", "y",
+                                     [["f", "v", "ve"]]>;
   }
 }
 
@@ -1601,11 +1654,11 @@ defm vslidedown : RVVSlideDownBuiltinSet;
 // 16.3.3. Vector Slide1up Instructions
 let UnMaskedPolicyScheme = HasPassthruOperand in {
 defm vslide1up : RVVSlideOneBuiltinSet;
-defm vfslide1up : RVVFloatingBinVFBuiltinSet;
+defm vfslide1up : RVVFloatingBinVFBuiltinSet<HasBF=1>;
 
 // 16.3.4. Vector Slide1down Instruction
 defm vslide1down : RVVSlideOneBuiltinSet;
-defm vfslide1down : RVVFloatingBinVFBuiltinSet;
+defm vfslide1down : RVVFloatingBinVFBuiltinSet<HasBF=1>;
 
 // 16.4. Vector Register Gather Instructions
 // signed and floating type
diff --git a/clang/include/clang/Basic/riscv_vector_common.td b/clang/include/clang/Basic/riscv_vector_common.td
index 767bcee..eaa2ba4 100644
--- a/clang/include/clang/Basic/riscv_vector_common.td
+++ b/clang/include/clang/Basic/riscv_vector_common.td
@@ -83,6 +83,8 @@
 //      elements of the same width
 //   F: given a vector type, compute the vector type with floating-point type
 //      elements of the same width
+//   Y: given a vector type, compute the vector type with bfloat16 type elements
+//      of the same width
 //   S: given a vector type, computes its equivalent one for LMUL=1. This is a
 //      no-op if the vector was already LMUL=1
 //   (Log2EEW:Value): Log2EEW value could be 3/4/5/6 (8/16/32/64), given a
@@ -470,6 +472,10 @@ let HasMaskedOffOperand = false in {
       defm "" : RVVOutOp1BuiltinSet<NAME, "x",
                                     [["vv", "v", "vvvv"],
                                      ["vf", "v", "vvev"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm "" : RVVOutOp1BuiltinSet<NAME, "y",
+                                    [["vv", "v", "vvvv"],
+                                     ["vf", "v", "vvev"]]>;
   }
   multiclass RVVFloatingTerBuiltinSetRoundingMode {
     defm "" : RVVOutOp1BuiltinSet<NAME, "fd",
@@ -479,6 +485,10 @@ let HasMaskedOffOperand = false in {
       defm "" : RVVOutOp1BuiltinSet<NAME, "x",
                                     [["vv", "v", "vvvvu"],
                                      ["vf", "v", "vvevu"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm "" : RVVOutOp1BuiltinSet<NAME, "y",
+                                    [["vv", "v", "vvvvu"],
+                                     ["vf", "v", "vvevu"]]>;
   }
 }
 
@@ -491,6 +501,10 @@ let HasMaskedOffOperand = false, Log2LMUL = [-2, -1, 0, 1, 2] in {
       defm ""  : RVVOutOp1Op2BuiltinSet<NAME, "x",
                                         [["vv", "w", "wwvv"],
                                          ["vf", "w", "wwev"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm ""  : RVVOutOp1Op2BuiltinSet<NAME, "y",
+                                        [["vv", "vw", "wwvv"],
+                                         ["vf", "vw", "wwev"]]>;
   }
   multiclass RVVFloatingWidenTerBuiltinSetRoundingMode {
     defm ""  : RVVOutOp1Op2BuiltinSet<NAME, "f",
@@ -500,10 +514,14 @@ let HasMaskedOffOperand = false, Log2LMUL = [-2, -1, 0, 1, 2] in {
       defm ""  : RVVOutOp1Op2BuiltinSet<NAME, "x",
                                         [["vv", "w", "wwvvu"],
                                          ["vf", "w", "wwevu"]]>;
+    let RequiredFeatures = ["zvfbfa"] in
+      defm ""  : RVVOutOp1Op2BuiltinSet<NAME, "y",
+                                        [["vv", "vw", "wwvvu"],
+                                         ["vf", "vw", "wwevu"]]>;
   }
 }
 
-multiclass RVVFloatingBinBuiltinSet {
+multiclass RVVFloatingBinBuiltinSet<bit HasBF = 0> {
   defm "" : RVVOutOp1BuiltinSet<NAME, "fd",
                                 [["vv", "v", "vvv"],
                                  ["vf", "v", "vve"]]>;
@@ -511,9 +529,15 @@ multiclass RVVFloatingBinBuiltinSet {
     defm "" : RVVOutOp1BuiltinSet<NAME, "x",
                                   [["vv", "v", "vvv"],
                                    ["vf", "v", "vve"]]>;
+  if HasBF then {
+    let RequiredFeatures = ["zvfbfa"] in
+      defm "" : RVVOutOp1BuiltinSet<NAME, "y",
+                                    [["vv", "v", "vvv"],
+                                     ["vf", "v", "vve"]]>;
+  }
 }
 
-multiclass RVVFloatingBinBuiltinSetRoundingMode {
+multiclass RVVFloatingBinBuiltinSetRoundingMode<bit HasBF = 0> {
   defm "" : RVVOutOp1BuiltinSet<NAME, "fd",
                                 [["vv", "v", "vvvu"],
                                  ["vf", "v", "vveu"]]>;
@@ -521,22 +545,38 @@ multiclass RVVFloatingBinBuiltinSetRoundingMode {
     defm "" : RVVOutOp1BuiltinSet<NAME, "x",
                                   [["vv", "v", "vvvu"],
                                    ["vf", "v", "vveu"]]>;
+  if HasBF then {
+    let RequiredFeatures = ["zvfbfa"] in
+      defm "" : RVVOutOp1BuiltinSet<NAME, "y",
+                                    [["vv", "v", "vvvu"],
+                                     ["vf", "v", "vveu"]]>;
+  }
 }
 
-multiclass RVVFloatingBinVFBuiltinSet {
+multiclass RVVFloatingBinVFBuiltinSet<bit HasBF = 0> {
   defm "" : RVVOutOp1BuiltinSet<NAME, "fd",
                                 [["vf", "v", "vve"]]>;
   let RequiredFeatures = ["zvfh"] in
     defm "" : RVVOutOp1BuiltinSet<NAME, "x",
                                   [["vf", "v", "vve"]]>;
+  if HasBF then {
+    let RequiredFeatures = ["zvfbfa"] in
+      defm "" : RVVOutOp1BuiltinSet<NAME, "y",
+                                    [["vf", "v", "vve"]]>;
+  }
 }
 
-multiclass RVVFloatingBinVFBuiltinSetRoundingMode {
+multiclass RVVFloatingBinVFBuiltinSetRoundingMode<bit HasBF = 0> {
   defm "" : RVVOutOp1BuiltinSet<NAME, "fd",
                                 [["vf", "v", "vveu"]]>;
   let RequiredFeatures = ["zvfh"] in
     defm "" : RVVOutOp1BuiltinSet<NAME, "x",
                                   [["vf", "v", "vveu"]]>;
+  if HasBF then {
+    let RequiredFeatures = ["zvfbfa"] in
+      defm "" : RVVOutOp1BuiltinSet<NAME, "y",
+                                    [["vf", "v", "vveu"]]>;
+  }
 }
 
 multiclass RVVFloatingMaskOutBuiltinSet {
@@ -547,6 +587,10 @@ multiclass RVVFloatingMaskOutBuiltinSet {
     defm "" : RVVOp0Op1BuiltinSet<NAME, "x",
                                   [["vv", "vm", "mvv"],
                                    ["vf", "vm", "mve"]]>;
+  let RequiredFeatures = ["zvfbfa"] in
+    defm "" : RVVOp0Op1BuiltinSet<NAME, "y",
+                                  [["vv", "vm", "mvv"],
+                                   ["vf", "vm", "mve"]]>;
 }
 
 multiclass RVVFloatingMaskOutVFBuiltinSet
@@ -748,6 +792,10 @@ multiclass RVVFloatingWidenBinBuiltinSet {
     defm "" : RVVWidenBuiltinSet<NAME, "x",
                                  [["vv", "w", "wvv"],
                                   ["vf", "w", "wve"]]>;
+  let RequiredFeatures = ["zvfbfa"] in
+    defm "" : RVVWidenBuiltinSet<NAME, "y",
+                                 [["vv", "vw", "wvv"],
+                                  ["vf", "vw", "wve"]]>;
 }
 
 multiclass RVVFloatingWidenBinBuiltinSetRoundingMode {
@@ -758,6 +806,10 @@ multiclass RVVFloatingWidenBinBuiltinSetRoundingMode {
     defm "" : RVVWidenBuiltinSet<NAME, "x",
                                  [["vv", "w", "wvvu"],
                                   ["vf", "w", "wveu"]]>;
+  let RequiredFeatures = ["zvfbfa"] in
+    defm "" : RVVWidenBuiltinSet<NAME, "y",
+                                 [["vv", "vw", "wvvu"],
+                                  ["vf", "vw", "wveu"]]>;
 }
 
 multiclass RVVFloatingWidenOp0BinBuiltinSet {
@@ -768,6 +820,10 @@ multiclass RVVFloatingWidenOp0BinBuiltinSet {
     defm "" : RVVWidenWOp0BuiltinSet<NAME # "_w", "x",
                                      [["wv", "w", "wwv"],
                                       ["wf", "w", "wwe"]]>;
+  let RequiredFeatures = ["zvfbfa"] in
+    defm "" : RVVWidenWOp0BuiltinSet<NAME # "_w", "y",
+                                     [["wv", "vw", "wwv"],
+                                      ["wf", "ew", "wwe"]]>;
 }
 
 multiclass RVVFloatingWidenOp0BinBuiltinSetRoundingMode {
@@ -778,4 +834,8 @@ multiclass RVVFloatingWidenOp0BinBuiltinSetRoundingMode {
     defm "" : RVVWidenWOp0BuiltinSet<NAME # "_w", "x",
                                      [["wv", "w", "wwvu"],
                                       ["wf", "w", "wweu"]]>;
+  let RequiredFeatures = ["zvfbfa"] in
+    defm "" : RVVWidenWOp0BuiltinSet<NAME # "_w", "y",
+                                     [["wv", "vw", "wwvu"],
+                                      ["wf", "ew", "wweu"]]>;
 }
diff --git a/clang/include/clang/Format/Format.h b/clang/include/clang/Format/Format.h
index 3df5b92..2852c4a 100644
--- a/clang/include/clang/Format/Format.h
+++ b/clang/include/clang/Format/Format.h
@@ -94,7 +94,7 @@ struct FormatStyle {
     ///
     /// \note
     ///  This currently only applies to braced initializer lists (when
-    ///  ``Cpp11BracedListStyle`` is ``true``) and parentheses.
+    ///  ``Cpp11BracedListStyle`` is not ``Block``) and parentheses.
     /// \endnote
     BAS_BlockIndent,
   };
@@ -2555,29 +2555,67 @@ struct FormatStyle {
   /// \version 3.7
   unsigned ContinuationIndentWidth;
 
-  /// If ``true``, format braced lists as best suited for C++11 braced
-  /// lists.
-  ///
-  /// Important differences:
-  ///
-  /// * No spaces inside the braced list.
-  /// * No line break before the closing brace.
-  /// * Indentation with the continuation indent, not with the block indent.
-  ///
-  /// Fundamentally, C++11 braced lists are formatted exactly like function
-  /// calls would be formatted in their place. If the braced list follows a name
-  /// (e.g. a type or variable name), clang-format formats as if the ``{}`` were
-  /// the parentheses of a function call with that name. If there is no name,
-  /// a zero-length name is assumed.
-  /// \code
-  ///    true:                                  false:
-  ///    vector<int> x{1, 2, 3, 4};     vs.     vector<int> x{ 1, 2, 3, 4 };
-  ///    vector<T> x{{}, {}, {}, {}};           vector<T> x{ {}, {}, {}, {} };
-  ///    f(MyMap[{composite, key}]);            f(MyMap[{ composite, key }]);
-  ///    new int[3]{1, 2, 3};                   new int[3]{ 1, 2, 3 };
-  /// \endcode
+  /// Different ways to handle braced lists.
+  enum BracedListStyle : int8_t {
+    /// Best suited for pre C++11 braced lists.
+    ///
+    /// * Spaces inside the braced list.
+    /// * Line break before the closing brace.
+    /// * Indentation with the block indent.
+    ///
+    /// \code
+    ///    vector<int> x{ 1, 2, 3, 4 };
+    ///    vector<T> x{ {}, {}, {}, {} };
+    ///    f(MyMap[{ composite, key }]);
+    ///    new int[3]{ 1, 2, 3 };
+    ///    Type name{ // Comment
+    ///               value
+    ///    };
+    /// \endcode
+    BLS_Block,
+    /// Best suited for C++11 braced lists.
+    ///
+    /// * No spaces inside the braced list.
+    /// * No line break before the closing brace.
+    /// * Indentation with the continuation indent.
+    ///
+    /// Fundamentally, C++11 braced lists are formatted exactly like function
+    /// calls would be formatted in their place. If the braced list follows a
+    /// name (e.g. a type or variable name), clang-format formats as if the
+    /// ``{}`` were the parentheses of a function call with that name. If there
+    /// is no name, a zero-length name is assumed.
+    /// \code
+    ///    vector<int> x{1, 2, 3, 4};
+    ///    vector<T> x{{}, {}, {}, {}};
+    ///    f(MyMap[{composite, key}]);
+    ///    new int[3]{1, 2, 3};
+    ///    Type name{ // Comment
+    ///        value};
+    /// \endcode
+    BLS_FunctionCall,
+    /// Same as ``FunctionCall``, except for the handling of a comment at the
+    /// begin, it then aligns everything following with the comment.
+    ///
+    /// * No spaces inside the braced list. (Even for a comment at the first
+    ///   position.)
+    /// * No line break before the closing brace.
+    /// * Indentation with the continuation indent, except when followed by a
+    ///   line comment, then it uses the block indent.
+    ///
+    /// \code
+    ///    vector<int> x{1, 2, 3, 4};
+    ///    vector<T> x{{}, {}, {}, {}};
+    ///    f(MyMap[{composite, key}]);
+    ///    new int[3]{1, 2, 3};
+    ///    Type name{// Comment
+    ///              value};
+    /// \endcode
+    BLS_AlignFirstComment,
+  };
+
+  /// The style to handle braced lists.
   /// \version 3.4
-  bool Cpp11BracedListStyle;
+  BracedListStyle Cpp11BracedListStyle;
 
   /// This option is **deprecated**. See ``DeriveLF`` and ``DeriveCRLF`` of
   /// ``LineEnding``.
@@ -4933,7 +4971,7 @@ struct FormatStyle {
   /// Specifies when to insert a space in empty braces.
   /// \note
   ///  This option doesn't apply to initializer braces if
-  ///  ``Cpp11BracedListStyle`` is set to ``true``.
+  ///  ``Cpp11BracedListStyle`` is not ``Block``.
   /// \endnote
   /// \version 22
   SpaceInEmptyBracesStyle SpaceInEmptyBraces;
diff --git a/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CallEvent.h b/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CallEvent.h
index c233ca1..4aee165 100644
--- a/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CallEvent.h
+++ b/clang/include/clang/StaticAnalyzer/Core/PathSensitive/CallEvent.h
@@ -211,6 +211,16 @@ protected:
   getExtraInvalidatedValues(ValueList &Values,
                             RegionAndSymbolInvalidationTraits *ETraits) const {}
 
+  /// A state for looking up relevant Environment entries (arguments, return
+  /// value), dynamic type information and similar "stable" things.
+  /// WARNING: During the evaluation of a function call, several state
+  /// transitions happen, so this state can become partially obsolete!
+  ///
+  /// TODO: Instead of storing a complete state object in the CallEvent, only
+  /// store the relevant parts (such as argument/return SVals etc.) that aren't
+  /// allowed to become obsolete until the end of the call evaluation.
+  ProgramStateRef getState() const { return State; }
+
 public:
   CallEvent &operator=(const CallEvent &) = delete;
   virtual ~CallEvent() = default;
@@ -231,8 +241,11 @@ public:
   }
   void setForeign(bool B) const { Foreign = B; }
 
-  /// The state in which the call is being evaluated.
-  const ProgramStateRef &getState() const { return State; }
+  /// NOTE: There are plans for refactoring that would eliminate this method.
+  /// Prefer to use CheckerContext::getASTContext if possible!
+  const ASTContext &getASTContext() const {
+    return getState()->getStateManager().getContext();
+  }
 
   /// The context in which the call is being evaluated.
   const LocationContext *getLocationContext() const { return LCtx; }
diff --git a/clang/lib/AST/ByteCode/InterpBuiltin.cpp b/clang/lib/AST/ByteCode/InterpBuiltin.cpp
index 5838cf8..0cb4910 100644
--- a/clang/lib/AST/ByteCode/InterpBuiltin.cpp
+++ b/clang/lib/AST/ByteCode/InterpBuiltin.cpp
@@ -3621,6 +3621,15 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
           return LHS.isSigned() ? LHS.ssub_sat(RHS) : LHS.usub_sat(RHS);
         });
 
+  case clang::X86::BI__builtin_ia32_pmulhrsw128:
+  case clang::X86::BI__builtin_ia32_pmulhrsw256:
+  case clang::X86::BI__builtin_ia32_pmulhrsw512:
+    return interp__builtin_elementwise_int_binop(
+        S, OpPC, Call, [](const APSInt &LHS, const APSInt &RHS) {
+          return (llvm::APIntOps::mulsExtended(LHS, RHS).ashr(14) + 1)
+              .extractBits(16, 1);
+        });
+
   case clang::X86::BI__builtin_ia32_pavgb128:
   case clang::X86::BI__builtin_ia32_pavgw128:
   case clang::X86::BI__builtin_ia32_pavgb256:
diff --git a/clang/lib/AST/ExprConstant.cpp b/clang/lib/AST/ExprConstant.cpp
index 16141b2..e308c17 100644
--- a/clang/lib/AST/ExprConstant.cpp
+++ b/clang/lib/AST/ExprConstant.cpp
@@ -11819,6 +11819,14 @@ bool VectorExprEvaluator::VisitCallExpr(const CallExpr *E) {
   case clang::X86::BI__builtin_ia32_pavgw512:
     return EvaluateBinOpExpr(llvm::APIntOps::avgCeilU);
 
+  case clang::X86::BI__builtin_ia32_pmulhrsw128:
+  case clang::X86::BI__builtin_ia32_pmulhrsw256:
+  case clang::X86::BI__builtin_ia32_pmulhrsw512:
+    return EvaluateBinOpExpr([](const APSInt &LHS, const APSInt &RHS) {
+      return (llvm::APIntOps::mulsExtended(LHS, RHS).ashr(14) + 1)
+          .extractBits(16, 1);
+    });
+
   case clang::X86::BI__builtin_ia32_pmaddubsw128:
   case clang::X86::BI__builtin_ia32_pmaddubsw256:
   case clang::X86::BI__builtin_ia32_pmaddubsw512:
diff --git a/clang/lib/Format/BreakableToken.cpp b/clang/lib/Format/BreakableToken.cpp
index 29db200..994a427 100644
--- a/clang/lib/Format/BreakableToken.cpp
+++ b/clang/lib/Format/BreakableToken.cpp
@@ -306,8 +306,10 @@ BreakableStringLiteralUsingOperators::BreakableStringLiteralUsingOperators(
     // In Verilog, all strings are quoted by double quotes, joined by commas,
     // and wrapped in braces.  The comma is always before the newline.
     assert(QuoteStyle == DoubleQuotes);
-    LeftBraceQuote = Style.Cpp11BracedListStyle ? "{\"" : "{ \"";
-    RightBraceQuote = Style.Cpp11BracedListStyle ? "\"}" : "\" }";
+    LeftBraceQuote =
+        Style.Cpp11BracedListStyle != FormatStyle::BLS_Block ? "{\"" : "{ \"";
+    RightBraceQuote =
+        Style.Cpp11BracedListStyle != FormatStyle::BLS_Block ? "\"}" : "\" }";
     Postfix = "\",";
     Prefix = "\"";
   } else {
diff --git a/clang/lib/Format/ContinuationIndenter.cpp b/clang/lib/Format/ContinuationIndenter.cpp
index cd4c1aa..26a9542 100644
--- a/clang/lib/Format/ContinuationIndenter.cpp
+++ b/clang/lib/Format/ContinuationIndenter.cpp
@@ -411,7 +411,7 @@ bool ContinuationIndenter::mustBreak(const LineState &State) {
   }
   if (CurrentState.BreakBeforeClosingBrace &&
       (Current.closesBlockOrBlockTypeList(Style) ||
-       (Current.is(tok::r_brace) &&
+       (Current.is(tok::r_brace) && Current.MatchingParen &&
         Current.isBlockIndentedInitRBrace(Style)))) {
     return true;
   }
@@ -833,7 +833,7 @@ void ContinuationIndenter::addTokenOnCurrentLine(LineState &State, bool DryRun,
   auto IsOpeningBracket = [&](const FormatToken &Tok) {
     auto IsStartOfBracedList = [&]() {
       return Tok.is(tok::l_brace) && Tok.isNot(BK_Block) &&
-             Style.Cpp11BracedListStyle;
+             Style.Cpp11BracedListStyle != FormatStyle::BLS_Block;
     };
     if (Tok.isNoneOf(tok::l_paren, TT_TemplateOpener, tok::l_square) &&
         !IsStartOfBracedList()) {
@@ -925,7 +925,12 @@ void ContinuationIndenter::addTokenOnCurrentLine(LineState &State, bool DryRun,
                         TT_TableGenDAGArgOpenerToBreak) &&
       !(Current.MacroParent && Previous.MacroParent) &&
       (Current.isNot(TT_LineComment) ||
-       Previous.isOneOf(BK_BracedInit, TT_VerilogMultiLineListLParen)) &&
+       (Previous.is(BK_BracedInit) &&
+        (Style.Cpp11BracedListStyle != FormatStyle::BLS_FunctionCall ||
+         !Previous.Previous ||
+         Previous.Previous->isNoneOf(tok::identifier, tok::l_paren,
+                                     BK_BracedInit))) ||
+       Previous.is(TT_VerilogMultiLineListLParen)) &&
       !IsInTemplateString(Current)) {
     CurrentState.Indent = State.Column + Spaces;
     CurrentState.IsAligned = true;
diff --git a/clang/lib/Format/Format.cpp b/clang/lib/Format/Format.cpp
index 093e88f..edd126c 100644
--- a/clang/lib/Format/Format.cpp
+++ b/clang/lib/Format/Format.cpp
@@ -304,6 +304,18 @@ struct ScalarEnumerationTraits<FormatStyle::BreakTemplateDeclarationsStyle> {
   }
 };
 
+template <> struct ScalarEnumerationTraits<FormatStyle::BracedListStyle> {
+  static void enumeration(IO &IO, FormatStyle::BracedListStyle &Value) {
+    IO.enumCase(Value, "Block", FormatStyle::BLS_Block);
+    IO.enumCase(Value, "FunctionCall", FormatStyle::BLS_FunctionCall);
+    IO.enumCase(Value, "AlignFirstComment", FormatStyle::BLS_AlignFirstComment);
+
+    // For backward compatibility.
+    IO.enumCase(Value, "false", FormatStyle::BLS_Block);
+    IO.enumCase(Value, "true", FormatStyle::BLS_AlignFirstComment);
+  }
+};
+
 template <> struct ScalarEnumerationTraits<FormatStyle::DAGArgStyle> {
   static void enumeration(IO &IO, FormatStyle::DAGArgStyle &Value) {
     IO.enumCase(Value, "DontBreak", FormatStyle::DAS_DontBreak);
@@ -1628,7 +1640,7 @@ FormatStyle getLLVMStyle(FormatStyle::LanguageKind Language) {
   LLVMStyle.CompactNamespaces = false;
   LLVMStyle.ConstructorInitializerIndentWidth = 4;
   LLVMStyle.ContinuationIndentWidth = 4;
-  LLVMStyle.Cpp11BracedListStyle = true;
+  LLVMStyle.Cpp11BracedListStyle = FormatStyle::BLS_AlignFirstComment;
   LLVMStyle.DerivePointerAlignment = false;
   LLVMStyle.DisableFormat = false;
   LLVMStyle.EmptyLineAfterAccessModifier = FormatStyle::ELAAMS_Never;
@@ -1904,7 +1916,7 @@ FormatStyle getGoogleStyle(FormatStyle::LanguageKind Language) {
     // beneficial there. Investigate turning this on once proper string reflow
     // has been implemented.
     GoogleStyle.BreakStringLiterals = false;
-    GoogleStyle.Cpp11BracedListStyle = false;
+    GoogleStyle.Cpp11BracedListStyle = FormatStyle::BLS_Block;
     GoogleStyle.SpacesInContainerLiterals = false;
   } else if (Language == FormatStyle::LK_ObjC) {
     GoogleStyle.AlwaysBreakBeforeMultilineStrings = false;
@@ -2000,7 +2012,7 @@ FormatStyle getMozillaStyle() {
   MozillaStyle.BreakTemplateDeclarations = FormatStyle::BTDS_Yes;
   MozillaStyle.ConstructorInitializerIndentWidth = 2;
   MozillaStyle.ContinuationIndentWidth = 2;
-  MozillaStyle.Cpp11BracedListStyle = false;
+  MozillaStyle.Cpp11BracedListStyle = FormatStyle::BLS_Block;
   MozillaStyle.FixNamespaceComments = false;
   MozillaStyle.IndentCaseLabels = true;
   MozillaStyle.ObjCSpaceAfterProperty = true;
@@ -2023,7 +2035,7 @@ FormatStyle getWebKitStyle() {
   Style.BreakBeforeBraces = FormatStyle::BS_WebKit;
   Style.BreakConstructorInitializers = FormatStyle::BCIS_BeforeComma;
   Style.ColumnLimit = 0;
-  Style.Cpp11BracedListStyle = false;
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_Block;
   Style.FixNamespaceComments = false;
   Style.IndentWidth = 4;
   Style.NamespaceIndentation = FormatStyle::NI_Inner;
@@ -2043,7 +2055,7 @@ FormatStyle getGNUStyle() {
   Style.BreakBeforeBraces = FormatStyle::BS_GNU;
   Style.BreakBeforeTernaryOperators = true;
   Style.ColumnLimit = 79;
-  Style.Cpp11BracedListStyle = false;
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_Block;
   Style.FixNamespaceComments = false;
   Style.KeepFormFeed = true;
   Style.SpaceBeforeParens = FormatStyle::SBPO_Always;
diff --git a/clang/lib/Format/FormatToken.cpp b/clang/lib/Format/FormatToken.cpp
index cb3fc1c..d1c6264 100644
--- a/clang/lib/Format/FormatToken.cpp
+++ b/clang/lib/Format/FormatToken.cpp
@@ -65,12 +65,13 @@ bool FormatToken::isTypeOrIdentifier(const LangOptions &LangOpts) const {
 
 bool FormatToken::isBlockIndentedInitRBrace(const FormatStyle &Style) const {
   assert(is(tok::r_brace));
-  if (!Style.Cpp11BracedListStyle ||
+  assert(MatchingParen);
+  assert(MatchingParen->is(tok::l_brace));
+  if (Style.Cpp11BracedListStyle == FormatStyle::BLS_Block ||
       Style.AlignAfterOpenBracket != FormatStyle::BAS_BlockIndent) {
     return false;
   }
   const auto *LBrace = MatchingParen;
-  assert(LBrace && LBrace->is(tok::l_brace));
   if (LBrace->is(BK_BracedInit))
     return true;
   if (LBrace->Previous && LBrace->Previous->is(tok::equal))
@@ -87,7 +88,8 @@ bool FormatToken::opensBlockOrBlockTypeList(const FormatStyle &Style) const {
   return is(TT_ArrayInitializerLSquare) || is(TT_ProtoExtensionLSquare) ||
          (is(tok::l_brace) &&
           (getBlockKind() == BK_Block || is(TT_DictLiteral) ||
-           (!Style.Cpp11BracedListStyle && NestingLevel == 0))) ||
+           (Style.Cpp11BracedListStyle == FormatStyle::BLS_Block &&
+            NestingLevel == 0))) ||
          (is(tok::less) && Style.isProto());
 }
 
@@ -183,7 +185,8 @@ void CommaSeparatedList::precomputeFormattingInfos(const FormatToken *Token) {
   // In C++11 braced list style, we should not format in columns unless they
   // have many items (20 or more) or we allow bin-packing of function call
   // arguments.
-  if (Style.Cpp11BracedListStyle && !Style.BinPackArguments &&
+  if (Style.Cpp11BracedListStyle != FormatStyle::BLS_Block &&
+      !Style.BinPackArguments &&
       (Commas.size() < 19 || !Style.BinPackLongBracedList)) {
     return;
   }
@@ -227,7 +230,7 @@ void CommaSeparatedList::precomputeFormattingInfos(const FormatToken *Token) {
       ItemEnd = Token->MatchingParen;
       const FormatToken *NonCommentEnd = ItemEnd->getPreviousNonComment();
       ItemLengths.push_back(CodePointsBetween(ItemBegin, NonCommentEnd));
-      if (Style.Cpp11BracedListStyle &&
+      if (Style.Cpp11BracedListStyle != FormatStyle::BLS_Block &&
           !ItemEnd->Previous->isTrailingComment()) {
         // In Cpp11 braced list style, the } and possibly other subsequent
         // tokens will need to stay on a line with the last element.
diff --git a/clang/lib/Format/TokenAnnotator.cpp b/clang/lib/Format/TokenAnnotator.cpp
index 5b784ed..778d2ca 100644
--- a/clang/lib/Format/TokenAnnotator.cpp
+++ b/clang/lib/Format/TokenAnnotator.cpp
@@ -3794,18 +3794,12 @@ static bool isFunctionDeclarationName(const LangOptions &LangOpts,
   if (Current.is(TT_FunctionDeclarationName))
     return true;
 
-  if (!Current.Tok.getIdentifierInfo())
+  if (Current.isNoneOf(tok::identifier, tok::kw_operator))
     return false;
 
   const auto *Prev = Current.getPreviousNonComment();
   assert(Prev);
 
-  if (Prev->is(tok::coloncolon))
-    Prev = Prev->Previous;
-
-  if (!Prev)
-    return false;
-
   const auto &Previous = *Prev;
 
   if (const auto *PrevPrev = Previous.getPreviousNonComment();
@@ -3854,6 +3848,8 @@ static bool isFunctionDeclarationName(const LangOptions &LangOpts,
 
   // Find parentheses of parameter list.
   if (Current.is(tok::kw_operator)) {
+    if (Line.startsWith(tok::kw_friend))
+      return true;
     if (Previous.Tok.getIdentifierInfo() &&
         Previous.isNoneOf(tok::kw_return, tok::kw_co_return)) {
       return true;
@@ -4098,7 +4094,8 @@ void TokenAnnotator::calculateFormattingInformation(AnnotatedLine &Line) const {
     if (Current->is(TT_LineComment)) {
       if (Prev->is(BK_BracedInit) && Prev->opensScope()) {
         Current->SpacesRequiredBefore =
-            (Style.Cpp11BracedListStyle && !Style.SpacesInParensOptions.Other)
+            (Style.Cpp11BracedListStyle == FormatStyle::BLS_AlignFirstComment &&
+             !Style.SpacesInParensOptions.Other)
                 ? 0
                 : 1;
       } else if (Prev->is(TT_VerilogMultiLineListLParen)) {
@@ -4449,8 +4446,10 @@ unsigned TokenAnnotator::splitPenalty(const AnnotatedLine &Line,
         (Left.ParameterCount <= 1 || Style.AllowAllArgumentsOnNextLine)) {
       return 0;
     }
-    if (Left.is(tok::l_brace) && !Style.Cpp11BracedListStyle)
+    if (Left.is(tok::l_brace) &&
+        Style.Cpp11BracedListStyle == FormatStyle::BLS_Block) {
       return 19;
+    }
     return Left.ParameterCount > 1 ? Style.PenaltyBreakBeforeFirstCallParameter
                                    : 19;
   }
@@ -4616,7 +4615,7 @@ bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine &Line,
       // Format empty list as `<>`.
       if (Left.is(tok::less) && Right.is(tok::greater))
         return false;
-      return !Style.Cpp11BracedListStyle;
+      return Style.Cpp11BracedListStyle == FormatStyle::BLS_Block;
     }
     // Don't attempt to format operator<(), as it is handled later.
     if (Right.isNot(TT_OverloadedOperatorLParen))
@@ -4784,7 +4783,8 @@ bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine &Line,
   const auto SpaceRequiredForArrayInitializerLSquare =
       [](const FormatToken &LSquareTok, const FormatStyle &Style) {
         return Style.SpacesInContainerLiterals ||
-               (Style.isProto() && !Style.Cpp11BracedListStyle &&
+               (Style.isProto() &&
+                Style.Cpp11BracedListStyle == FormatStyle::BLS_Block &&
                 LSquareTok.endsSequence(tok::l_square, tok::colon,
                                         TT_SelectorName));
       };
@@ -4817,7 +4817,8 @@ bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine &Line,
   if ((Left.is(tok::l_brace) && Left.isNot(BK_Block)) ||
       (Right.is(tok::r_brace) && Right.MatchingParen &&
        Right.MatchingParen->isNot(BK_Block))) {
-    return !Style.Cpp11BracedListStyle || Style.SpacesInParensOptions.Other;
+    return Style.Cpp11BracedListStyle == FormatStyle::BLS_Block ||
+           Style.SpacesInParensOptions.Other;
   }
   if (Left.is(TT_BlockComment)) {
     // No whitespace in x(/*foo=*/1), except for JavaScript.
@@ -4999,7 +5000,7 @@ bool TokenAnnotator::spaceRequiredBefore(const AnnotatedLine &Line,
       Left.Children.empty()) {
     if (Left.is(BK_Block))
       return Style.SpaceInEmptyBraces != FormatStyle::SIEB_Never;
-    if (Style.Cpp11BracedListStyle) {
+    if (Style.Cpp11BracedListStyle != FormatStyle::BLS_Block) {
       return Style.SpacesInParens == FormatStyle::SIPO_Custom &&
              Style.SpacesInParensOptions.InEmptyParentheses;
     }
@@ -5081,7 +5082,7 @@ bool TokenAnnotator::spaceRequiredBefore(const AnnotatedLine &Line,
     if (Left.MatchingParen &&
         Left.MatchingParen->is(TT_ProtoExtensionLSquare) &&
         Right.isOneOf(tok::l_brace, tok::less)) {
-      return !Style.Cpp11BracedListStyle;
+      return Style.Cpp11BracedListStyle == FormatStyle::BLS_Block;
     }
     // A percent is probably part of a formatting specification, such as %lld.
     if (Left.is(tok::percent))
@@ -5521,7 +5522,7 @@ bool TokenAnnotator::spaceRequiredBefore(const AnnotatedLine &Line,
   if (Left.is(tok::greater) && Right.is(tok::greater)) {
     if (Style.isTextProto() ||
         (Style.Language == FormatStyle::LK_Proto && Left.is(TT_DictLiteral))) {
-      return !Style.Cpp11BracedListStyle;
+      return Style.Cpp11BracedListStyle == FormatStyle::BLS_Block;
     }
     return Right.is(TT_TemplateCloser) && Left.is(TT_TemplateCloser) &&
            ((Style.Standard < FormatStyle::LS_Cpp11) ||
@@ -6382,7 +6383,7 @@ bool TokenAnnotator::canBreakBefore(const AnnotatedLine &Line,
     return false;
   }
   if (Left.is(tok::equal) && Right.is(tok::l_brace) &&
-      !Style.Cpp11BracedListStyle) {
+      Style.Cpp11BracedListStyle == FormatStyle::BLS_Block) {
     return false;
   }
   if (Left.is(TT_AttributeLParen) ||
diff --git a/clang/lib/Format/WhitespaceManager.cpp b/clang/lib/Format/WhitespaceManager.cpp
index 7348a3a..9261294 100644
--- a/clang/lib/Format/WhitespaceManager.cpp
+++ b/clang/lib/Format/WhitespaceManager.cpp
@@ -1238,7 +1238,8 @@ void WhitespaceManager::alignArrayInitializersRightJustified(
   if (!CellDescs.isRectangular())
     return;
 
-  const int BracePadding = Style.Cpp11BracedListStyle ? 0 : 1;
+  const int BracePadding =
+      Style.Cpp11BracedListStyle != FormatStyle::BLS_Block ? 0 : 1;
   auto &Cells = CellDescs.Cells;
   // Now go through and fixup the spaces.
   auto *CellIter = Cells.begin();
@@ -1314,7 +1315,8 @@ void WhitespaceManager::alignArrayInitializersLeftJustified(
   if (!CellDescs.isRectangular())
     return;
 
-  const int BracePadding = Style.Cpp11BracedListStyle ? 0 : 1;
+  const int BracePadding =
+      Style.Cpp11BracedListStyle != FormatStyle::BLS_Block ? 0 : 1;
   auto &Cells = CellDescs.Cells;
   // Now go through and fixup the spaces.
   auto *CellIter = Cells.begin();
diff --git a/clang/lib/Headers/avx2intrin.h b/clang/lib/Headers/avx2intrin.h
index fa7f4c2..d35bc0e 100644
--- a/clang/lib/Headers/avx2intrin.h
+++ b/clang/lib/Headers/avx2intrin.h
@@ -1650,9 +1650,8 @@ _mm256_mul_epi32(__m256i __a, __m256i __b) {
 /// \param __b
 ///    A 256-bit vector of [16 x i16] containing one of the source operands.
 /// \returns A 256-bit vector of [16 x i16] containing the rounded products.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_mulhrs_epi16(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mulhrs_epi16(__m256i __a, __m256i __b) {
   return (__m256i)__builtin_ia32_pmulhrsw256((__v16hi)__a, (__v16hi)__b);
 }
 
@@ -1670,8 +1669,7 @@ _mm256_mulhrs_epi16(__m256i __a, __m256i __b)
 ///    A 256-bit vector of [16 x i16] containing one of the source operands.
 /// \returns A 256-bit vector of [16 x i16] containing the products.
 static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
-_mm256_mulhi_epu16(__m256i __a, __m256i __b)
-{
+_mm256_mulhi_epu16(__m256i __a, __m256i __b) {
   return (__m256i)__builtin_ia32_pmulhuw256((__v16hu)__a, (__v16hu)__b);
 }
 
diff --git a/clang/lib/Headers/avx512bwintrin.h b/clang/lib/Headers/avx512bwintrin.h
index 23b2d29..ac75b6c 100644
--- a/clang/lib/Headers/avx512bwintrin.h
+++ b/clang/lib/Headers/avx512bwintrin.h
@@ -1003,23 +1003,20 @@ _mm512_maskz_permutex2var_epi16(__mmask32 __U, __m512i __A, __m512i __I,
                               (__v32hi)_mm512_setzero_si512());
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mulhrs_epi16(__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mulhrs_epi16(__m512i __A, __m512i __B) {
   return (__m512i)__builtin_ia32_pmulhrsw512((__v32hi)__A, (__v32hi)__B);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_mask_mulhrs_epi16(__m512i __W, __mmask32 __U, __m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_mask_mulhrs_epi16(__m512i __W, __mmask32 __U, __m512i __A, __m512i __B) {
   return (__m512i)__builtin_ia32_selectw_512((__mmask32)__U,
                                          (__v32hi)_mm512_mulhrs_epi16(__A, __B),
                                          (__v32hi)__W);
 }
 
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_maskz_mulhrs_epi16(__mmask32 __U, __m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_maskz_mulhrs_epi16(__mmask32 __U, __m512i __A, __m512i __B) {
   return (__m512i)__builtin_ia32_selectw_512((__mmask32)__U,
                                          (__v32hi)_mm512_mulhrs_epi16(__A, __B),
                                          (__v32hi)_mm512_setzero_si512());
diff --git a/clang/lib/Headers/avx512vlbwintrin.h b/clang/lib/Headers/avx512vlbwintrin.h
index 639fb60..0fcfe37 100644
--- a/clang/lib/Headers/avx512vlbwintrin.h
+++ b/clang/lib/Headers/avx512vlbwintrin.h
@@ -1510,28 +1510,28 @@ _mm256_mask_cvtusepi16_storeu_epi8 (void * __P, __mmask16 __M, __m256i __A)
   __builtin_ia32_pmovuswb256mem_mask ((__v16qi*) __P, (__v16hi) __A, __M);
 }
 
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
 _mm_mask_mulhrs_epi16(__m128i __W, __mmask8 __U, __m128i __X, __m128i __Y) {
   return (__m128i)__builtin_ia32_selectw_128((__mmask8)__U,
                                              (__v8hi)_mm_mulhrs_epi16(__X, __Y),
                                              (__v8hi)__W);
 }
 
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
 _mm_maskz_mulhrs_epi16(__mmask8 __U, __m128i __X, __m128i __Y) {
   return (__m128i)__builtin_ia32_selectw_128((__mmask8)__U,
                                              (__v8hi)_mm_mulhrs_epi16(__X, __Y),
                                              (__v8hi)_mm_setzero_si128());
 }
 
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_mask_mulhrs_epi16(__m256i __W, __mmask16 __U, __m256i __X, __m256i __Y) {
   return (__m256i)__builtin_ia32_selectw_256((__mmask16)__U,
                                          (__v16hi)_mm256_mulhrs_epi16(__X, __Y),
                                          (__v16hi)__W);
 }
 
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_maskz_mulhrs_epi16(__mmask16 __U, __m256i __X, __m256i __Y) {
   return (__m256i)__builtin_ia32_selectw_256((__mmask16)__U,
                                          (__v16hi)_mm256_mulhrs_epi16(__X, __Y),
diff --git a/clang/lib/Headers/tmmintrin.h b/clang/lib/Headers/tmmintrin.h
index ee96caa..5d0f20f 100644
--- a/clang/lib/Headers/tmmintrin.h
+++ b/clang/lib/Headers/tmmintrin.h
@@ -544,8 +544,8 @@ _mm_maddubs_pi16(__m64 __a, __m64 __b) {
 ///    A 128-bit vector of [8 x i16] containing one of the source operands.
 /// \returns A 128-bit vector of [8 x i16] containing the rounded and scaled
 ///    products of both operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_mulhrs_epi16(__m128i __a,
-                                                              __m128i __b) {
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_mulhrs_epi16(__m128i __a, __m128i __b) {
   return (__m128i)__builtin_ia32_pmulhrsw128((__v8hi)__a, (__v8hi)__b);
 }
 
@@ -563,11 +563,10 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_mulhrs_epi16(__m128i __a,
 ///    A 64-bit vector of [4 x i16] containing one of the source operands.
 /// \returns A 64-bit vector of [4 x i16] containing the rounded and scaled
 ///    products of both operands.
-static __inline__ __m64 __DEFAULT_FN_ATTRS
-_mm_mulhrs_pi16(__m64 __a, __m64 __b)
-{
-    return __trunc64(__builtin_ia32_pmulhrsw128((__v8hi)__anyext128(__a),
-                                                (__v8hi)__anyext128(__b)));
+static __inline__ __m64 __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_mulhrs_pi16(__m64 __a, __m64 __b) {
+  return __trunc64(__builtin_ia32_pmulhrsw128((__v8hi)__zext128(__a),
+                                              (__v8hi)__zext128(__b)));
 }
 
 /// Copies the 8-bit integers from a 128-bit integer vector to the
diff --git a/clang/lib/Sema/SemaRISCV.cpp b/clang/lib/Sema/SemaRISCV.cpp
index 3ba93ff9..c5ef0d5 100644
--- a/clang/lib/Sema/SemaRISCV.cpp
+++ b/clang/lib/Sema/SemaRISCV.cpp
@@ -1464,7 +1464,8 @@ void SemaRISCV::checkRVVTypeSupport(QualType Ty, SourceLocation Loc, Decl *D,
     }
   else if (Info.ElementType->isBFloat16Type() &&
            !FeatureMap.lookup("zvfbfmin") &&
-           !FeatureMap.lookup("xandesvbfhcvt"))
+           !FeatureMap.lookup("xandesvbfhcvt") &&
+           !FeatureMap.lookup("experimental-zvfbfa"))
     if (DeclareAndesVectorBuiltins) {
       Diag(Loc, diag::err_riscv_type_requires_extension, D)
           << Ty << "zvfbfmin or xandesvbfhcvt";
diff --git a/clang/lib/Sema/SemaTemplateInstantiate.cpp b/clang/lib/Sema/SemaTemplateInstantiate.cpp
index ca7e3b2..038f396 100644
--- a/clang/lib/Sema/SemaTemplateInstantiate.cpp
+++ b/clang/lib/Sema/SemaTemplateInstantiate.cpp
@@ -2864,9 +2864,9 @@ TemplateInstantiator::TransformNestedRequirement(
         TemplateArgs, Constraint->getSourceRange(), Satisfaction,
         /*TopLevelConceptId=*/nullptr, &NewConstraint);
 
-    assert(!Success || !Trap.hasErrorOccurred() &&
-                           "Substitution failures must be handled "
-                           "by CheckConstraintSatisfaction.");
+    assert((!Success || !Trap.hasErrorOccurred()) &&
+           "Substitution failures must be handled "
+           "by CheckConstraintSatisfaction.");
   }
 
   if (!Success || Satisfaction.HasSubstitutionFailure())
diff --git a/clang/lib/StaticAnalyzer/Checkers/BlockInCriticalSectionChecker.cpp b/clang/lib/StaticAnalyzer/Checkers/BlockInCriticalSectionChecker.cpp
index bf35bee..3ddd659 100644
--- a/clang/lib/StaticAnalyzer/Checkers/BlockInCriticalSectionChecker.cpp
+++ b/clang/lib/StaticAnalyzer/Checkers/BlockInCriticalSectionChecker.cpp
@@ -104,7 +104,7 @@ class RAIIMutexDescriptor {
       // this function is called instead of early returning it. To avoid this, a
       // bool variable (IdentifierInfoInitialized) is used and the function will
       // be run only once.
-      const auto &ASTCtx = Call.getState()->getStateManager().getContext();
+      const auto &ASTCtx = Call.getASTContext();
       Guard = &ASTCtx.Idents.get(GuardName);
     }
   }
diff --git a/clang/lib/StaticAnalyzer/Checkers/CheckObjCDealloc.cpp b/clang/lib/StaticAnalyzer/Checkers/CheckObjCDealloc.cpp
index 9d3aeff..2420848 100644
--- a/clang/lib/StaticAnalyzer/Checkers/CheckObjCDealloc.cpp
+++ b/clang/lib/StaticAnalyzer/Checkers/CheckObjCDealloc.cpp
@@ -929,7 +929,7 @@ ObjCDeallocChecker::getValueReleasedByNillingOut(const ObjCMethodCall &M,
   SVal Arg = M.getArgSVal(0);
   ProgramStateRef notNilState, nilState;
   std::tie(notNilState, nilState) =
-      M.getState()->assume(Arg.castAs<DefinedOrUnknownSVal>());
+      C.getState()->assume(Arg.castAs<DefinedOrUnknownSVal>());
   if (!(nilState && !notNilState))
     return nullptr;
 
diff --git a/clang/lib/StaticAnalyzer/Checkers/ObjCSuperDeallocChecker.cpp b/clang/lib/StaticAnalyzer/Checkers/ObjCSuperDeallocChecker.cpp
index f984caf..227cbfa 100644
--- a/clang/lib/StaticAnalyzer/Checkers/ObjCSuperDeallocChecker.cpp
+++ b/clang/lib/StaticAnalyzer/Checkers/ObjCSuperDeallocChecker.cpp
@@ -34,7 +34,7 @@ class ObjCSuperDeallocChecker
       this, "[super dealloc] should not be called more than once",
       categories::CoreFoundationObjectiveC};
 
-  void initIdentifierInfoAndSelectors(ASTContext &Ctx) const;
+  void initIdentifierInfoAndSelectors(const ASTContext &Ctx) const;
 
   bool isSuperDeallocMessage(const ObjCMethodCall &M) const;
 
@@ -214,8 +214,8 @@ void ObjCSuperDeallocChecker::diagnoseCallArguments(const CallEvent &CE,
   }
 }
 
-void
-ObjCSuperDeallocChecker::initIdentifierInfoAndSelectors(ASTContext &Ctx) const {
+void ObjCSuperDeallocChecker::initIdentifierInfoAndSelectors(
+    const ASTContext &Ctx) const {
   if (IIdealloc)
     return;
 
@@ -230,7 +230,7 @@ ObjCSuperDeallocChecker::isSuperDeallocMessage(const ObjCMethodCall &M) const {
   if (M.getOriginExpr()->getReceiverKind() != ObjCMessageExpr::SuperInstance)
     return false;
 
-  ASTContext &Ctx = M.getState()->getStateManager().getContext();
+  const ASTContext &Ctx = M.getASTContext();
   initIdentifierInfoAndSelectors(Ctx);
 
   return M.getSelector() == SELdealloc;
diff --git a/clang/lib/StaticAnalyzer/Checkers/StdVariantChecker.cpp b/clang/lib/StaticAnalyzer/Checkers/StdVariantChecker.cpp
index 4fc1c57..db8bbee 100644
--- a/clang/lib/StaticAnalyzer/Checkers/StdVariantChecker.cpp
+++ b/clang/lib/StaticAnalyzer/Checkers/StdVariantChecker.cpp
@@ -211,13 +211,13 @@ private:
     if (!DefaultType)
       return;
 
-    ProgramStateRef State = ConstructorCall->getState();
+    ProgramStateRef State = C.getState();
     State = State->set<VariantHeldTypeMap>(ThisMemRegion, *DefaultType);
     C.addTransition(State);
   }
 
   bool handleStdGetCall(const CallEvent &Call, CheckerContext &C) const {
-    ProgramStateRef State = Call.getState();
+    ProgramStateRef State = C.getState();
 
     const auto &ArgType = Call.getArgSVal(0)
                               .getType(C.getASTContext())
diff --git a/clang/lib/StaticAnalyzer/Checkers/TaggedUnionModeling.h b/clang/lib/StaticAnalyzer/Checkers/TaggedUnionModeling.h
index dec4612..b8fb572 100644
--- a/clang/lib/StaticAnalyzer/Checkers/TaggedUnionModeling.h
+++ b/clang/lib/StaticAnalyzer/Checkers/TaggedUnionModeling.h
@@ -52,7 +52,7 @@ removeInformationStoredForDeadInstances(const CallEvent &Call,
 template <class TypeMap>
 void handleConstructorAndAssignment(const CallEvent &Call, CheckerContext &C,
                                     SVal ThisSVal) {
-  ProgramStateRef State = Call.getState();
+  ProgramStateRef State = C.getState();
 
   if (!State)
     return;
diff --git a/clang/lib/StaticAnalyzer/Core/CheckerManager.cpp b/clang/lib/StaticAnalyzer/Core/CheckerManager.cpp
index 44c6f9f..8ee4832 100644
--- a/clang/lib/StaticAnalyzer/Core/CheckerManager.cpp
+++ b/clang/lib/StaticAnalyzer/Core/CheckerManager.cpp
@@ -731,19 +731,22 @@ void CheckerManager::runCheckersForEvalCall(ExplodedNodeSet &Dst,
     ExplodedNodeSet checkDst;
     NodeBuilder B(Pred, checkDst, Eng.getBuilderContext());
 
+    ProgramStateRef State = Pred->getState();
+    CallEventRef<> UpdatedCall = Call.cloneWithState(State);
+
     // Check if any of the EvalCall callbacks can evaluate the call.
     for (const auto &EvalCallChecker : EvalCallCheckers) {
       // TODO: Support the situation when the call doesn't correspond
       // to any Expr.
       ProgramPoint L = ProgramPoint::getProgramPoint(
-          Call.getOriginExpr(), ProgramPoint::PostStmtKind,
+          UpdatedCall->getOriginExpr(), ProgramPoint::PostStmtKind,
           Pred->getLocationContext(), EvalCallChecker.Checker);
       bool evaluated = false;
-      { // CheckerContext generates transitions(populates checkDest) on
+      { // CheckerContext generates transitions (populates checkDest) on
         // destruction, so introduce the scope to make sure it gets properly
         // populated.
         CheckerContext C(B, Eng, Pred, L);
-        evaluated = EvalCallChecker(Call, C);
+        evaluated = EvalCallChecker(*UpdatedCall, C);
       }
 #ifndef NDEBUG
       if (evaluated && evaluatorChecker) {
@@ -774,7 +777,7 @@ void CheckerManager::runCheckersForEvalCall(ExplodedNodeSet &Dst,
     // If none of the checkers evaluated the call, ask ExprEngine to handle it.
     if (!evaluatorChecker) {
       NodeBuilder B(Pred, Dst, Eng.getBuilderContext());
-      Eng.defaultEvalCall(B, Pred, Call, CallOpts);
+      Eng.defaultEvalCall(B, Pred, *UpdatedCall, CallOpts);
     }
   }
 }
diff --git a/clang/lib/StaticAnalyzer/Core/ExprEngineCallAndReturn.cpp b/clang/lib/StaticAnalyzer/Core/ExprEngineCallAndReturn.cpp
index 0c491b8..ac6c1d7 100644
--- a/clang/lib/StaticAnalyzer/Core/ExprEngineCallAndReturn.cpp
+++ b/clang/lib/StaticAnalyzer/Core/ExprEngineCallAndReturn.cpp
@@ -628,6 +628,8 @@ void ExprEngine::VisitCallExpr(const CallExpr *CE, ExplodedNode *Pred,
 
 ProgramStateRef ExprEngine::finishArgumentConstruction(ProgramStateRef State,
                                                        const CallEvent &Call) {
+  // WARNING: The state attached to 'Call' may be obsolete, do not call any
+  // methods that rely on it!
   const Expr *E = Call.getOriginExpr();
   // FIXME: Constructors to placement arguments of operator new
   // are not supported yet.
@@ -653,6 +655,8 @@ ProgramStateRef ExprEngine::finishArgumentConstruction(ProgramStateRef State,
 void ExprEngine::finishArgumentConstruction(ExplodedNodeSet &Dst,
                                             ExplodedNode *Pred,
                                             const CallEvent &Call) {
+  // WARNING: The state attached to 'Call' may be obsolete, do not call any
+  // methods that rely on it!
   ProgramStateRef State = Pred->getState();
   ProgramStateRef CleanedState = finishArgumentConstruction(State, Call);
   if (CleanedState == State) {
@@ -670,35 +674,33 @@ void ExprEngine::finishArgumentConstruction(ExplodedNodeSet &Dst,
 }
 
 void ExprEngine::evalCall(ExplodedNodeSet &Dst, ExplodedNode *Pred,
-                          const CallEvent &Call) {
-  // WARNING: At this time, the state attached to 'Call' may be older than the
-  // state in 'Pred'. This is a minor optimization since CheckerManager will
-  // use an updated CallEvent instance when calling checkers, but if 'Call' is
-  // ever used directly in this function all callers should be updated to pass
-  // the most recent state. (It is probably not worth doing the work here since
-  // for some callers this will not be necessary.)
+                          const CallEvent &CallTemplate) {
+  // NOTE: CallTemplate is called a "template" because its attached state may
+  // be obsolete (compared to the state of Pred). The state-dependent methods
+  // of CallEvent should be used only after a `cloneWithState` call that
+  // attaches the up-to-date state to this template object.
 
   // Run any pre-call checks using the generic call interface.
   ExplodedNodeSet dstPreVisit;
-  getCheckerManager().runCheckersForPreCall(dstPreVisit, Pred,
-                                            Call, *this);
+  getCheckerManager().runCheckersForPreCall(dstPreVisit, Pred, CallTemplate,
+                                            *this);
 
   // Actually evaluate the function call.  We try each of the checkers
   // to see if the can evaluate the function call, and get a callback at
   // defaultEvalCall if all of them fail.
   ExplodedNodeSet dstCallEvaluated;
-  getCheckerManager().runCheckersForEvalCall(dstCallEvaluated, dstPreVisit,
-                                             Call, *this, EvalCallOptions());
+  getCheckerManager().runCheckersForEvalCall(
+      dstCallEvaluated, dstPreVisit, CallTemplate, *this, EvalCallOptions());
 
   // If there were other constructors called for object-type arguments
   // of this call, clean them up.
   ExplodedNodeSet dstArgumentCleanup;
   for (ExplodedNode *I : dstCallEvaluated)
-    finishArgumentConstruction(dstArgumentCleanup, I, Call);
+    finishArgumentConstruction(dstArgumentCleanup, I, CallTemplate);
 
   ExplodedNodeSet dstPostCall;
   getCheckerManager().runCheckersForPostCall(dstPostCall, dstArgumentCleanup,
-                                             Call, *this);
+                                             CallTemplate, *this);
 
   // Escaping symbols conjured during invalidating the regions above.
   // Note that, for inlined calls the nodes were put back into the worklist,
@@ -708,12 +710,13 @@ void ExprEngine::evalCall(ExplodedNodeSet &Dst, ExplodedNode *Pred,
   // Run pointerEscape callback with the newly conjured symbols.
   SmallVector<std::pair<SVal, SVal>, 8> Escaped;
   for (ExplodedNode *I : dstPostCall) {
-    NodeBuilder B(I, Dst, *currBldrCtx);
     ProgramStateRef State = I->getState();
+    CallEventRef<> Call = CallTemplate.cloneWithState(State);
+    NodeBuilder B(I, Dst, *currBldrCtx);
     Escaped.clear();
     {
       unsigned Arg = -1;
-      for (const ParmVarDecl *PVD : Call.parameters()) {
+      for (const ParmVarDecl *PVD : Call->parameters()) {
         ++Arg;
         QualType ParamTy = PVD->getType();
         if (ParamTy.isNull() ||
@@ -722,13 +725,13 @@ void ExprEngine::evalCall(ExplodedNodeSet &Dst, ExplodedNode *Pred,
         QualType Pointee = ParamTy->getPointeeType();
         if (Pointee.isConstQualified() || Pointee->isVoidType())
           continue;
-        if (const MemRegion *MR = Call.getArgSVal(Arg).getAsRegion())
+        if (const MemRegion *MR = Call->getArgSVal(Arg).getAsRegion())
           Escaped.emplace_back(loc::MemRegionVal(MR), State->getSVal(MR, Pointee));
       }
     }
 
     State = processPointerEscapedOnBind(State, Escaped, I->getLocationContext(),
-                                        PSK_EscapeOutParameters, &Call);
+                                        PSK_EscapeOutParameters, &*Call);
 
     if (State == I->getState())
       Dst.insert(I);
@@ -1212,48 +1215,47 @@ static bool isTrivialObjectAssignment(const CallEvent &Call) {
 }
 
 void ExprEngine::defaultEvalCall(NodeBuilder &Bldr, ExplodedNode *Pred,
-                                 const CallEvent &CallTemplate,
+                                 const CallEvent &Call,
                                  const EvalCallOptions &CallOpts) {
   // Make sure we have the most recent state attached to the call.
   ProgramStateRef State = Pred->getState();
-  CallEventRef<> Call = CallTemplate.cloneWithState(State);
 
   // Special-case trivial assignment operators.
-  if (isTrivialObjectAssignment(*Call)) {
-    performTrivialCopy(Bldr, Pred, *Call);
+  if (isTrivialObjectAssignment(Call)) {
+    performTrivialCopy(Bldr, Pred, Call);
     return;
   }
 
   // Try to inline the call.
   // The origin expression here is just used as a kind of checksum;
   // this should still be safe even for CallEvents that don't come from exprs.
-  const Expr *E = Call->getOriginExpr();
+  const Expr *E = Call.getOriginExpr();
 
   ProgramStateRef InlinedFailedState = getInlineFailedState(State, E);
   if (InlinedFailedState) {
     // If we already tried once and failed, make sure we don't retry later.
     State = InlinedFailedState;
   } else {
-    RuntimeDefinition RD = Call->getRuntimeDefinition();
-    Call->setForeign(RD.isForeign());
+    RuntimeDefinition RD = Call.getRuntimeDefinition();
+    Call.setForeign(RD.isForeign());
     const Decl *D = RD.getDecl();
-    if (shouldInlineCall(*Call, D, Pred, CallOpts)) {
+    if (shouldInlineCall(Call, D, Pred, CallOpts)) {
       if (RD.mayHaveOtherDefinitions()) {
         AnalyzerOptions &Options = getAnalysisManager().options;
 
         // Explore with and without inlining the call.
         if (Options.getIPAMode() == IPAK_DynamicDispatchBifurcate) {
-          BifurcateCall(RD.getDispatchRegion(), *Call, D, Bldr, Pred);
+          BifurcateCall(RD.getDispatchRegion(), Call, D, Bldr, Pred);
           return;
         }
 
         // Don't inline if we're not in any dynamic dispatch mode.
         if (Options.getIPAMode() != IPAK_DynamicDispatch) {
-          conservativeEvalCall(*Call, Bldr, Pred, State);
+          conservativeEvalCall(Call, Bldr, Pred, State);
           return;
         }
       }
-      ctuBifurcate(*Call, D, Bldr, Pred, State);
+      ctuBifurcate(Call, D, Bldr, Pred, State);
       return;
     }
   }
@@ -1261,10 +1263,10 @@ void ExprEngine::defaultEvalCall(NodeBuilder &Bldr, ExplodedNode *Pred,
   // If we can't inline it, clean up the state traits used only if the function
   // is inlined.
   State = removeStateTraitsUsedForArrayEvaluation(
-      State, dyn_cast_or_null<CXXConstructExpr>(E), Call->getLocationContext());
+      State, dyn_cast_or_null<CXXConstructExpr>(E), Call.getLocationContext());
 
   // Also handle the return value and invalidate the regions.
-  conservativeEvalCall(*Call, Bldr, Pred, State);
+  conservativeEvalCall(Call, Bldr, Pred, State);
 }
 
 void ExprEngine::BifurcateCall(const MemRegion *BifurReg,
diff --git a/clang/lib/Support/RISCVVIntrinsicUtils.cpp b/clang/lib/Support/RISCVVIntrinsicUtils.cpp
index 5a4e805..dad3d0da 100644
--- a/clang/lib/Support/RISCVVIntrinsicUtils.cpp
+++ b/clang/lib/Support/RISCVVIntrinsicUtils.cpp
@@ -654,6 +654,9 @@ PrototypeDescriptor::parsePrototypeDescriptor(
     case 'F':
       TM |= TypeModifier::Float;
       break;
+    case 'Y':
+      TM |= TypeModifier::BFloat;
+      break;
     case 'S':
       TM |= TypeModifier::LMUL1;
       break;
@@ -704,6 +707,8 @@ void RVVType::applyModifier(const PrototypeDescriptor &Transformer) {
     ElementBitwidth *= 2;
     LMUL.MulLog2LMUL(1);
     Scale = LMUL.getScale(ElementBitwidth);
+    if (ScalarType == ScalarTypeKind::BFloat)
+      ScalarType = ScalarTypeKind::Float;
     break;
   case VectorTypeModifier::Widening4XVector:
     ElementBitwidth *= 4;
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfadd.c
new file mode 100644
index 0000000..d7734e0
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfadd.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m8_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfclass.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfclass.c
new file mode 100644
index 0000000..68814f4
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfclass.c
@@ -0,0 +1,134 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.nxv1bf16.i64(<vscale x 1 x i16> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16mf4_u16mf4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.nxv2bf16.i64(<vscale x 2 x i16> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16mf2_u16mf2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.nxv4bf16.i64(<vscale x 4 x i16> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m1_u16m1(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.nxv8bf16.i64(<vscale x 8 x i16> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m2_u16m2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.nxv16bf16.i64(<vscale x 16 x i16> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m4_u16m4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.nxv32bf16.i64(<vscale x 32 x i16> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m8_u16m8(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfclass_v_bf16mf4_u16mf4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfclass_v_bf16mf2_u16mf2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                          size_t vl) {
+  return __riscv_vfclass_v_bf16m1_u16m1_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfclass_v_bf16m2_u16m2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                          size_t vl) {
+  return __riscv_vfclass_v_bf16m4_u16m4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                          size_t vl) {
+  return __riscv_vfclass_v_bf16m8_u16m8_m(vm, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmacc.c
new file mode 100644
index 0000000..616455d
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmacc.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m1(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m8(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m1_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m1_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m8_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m8_m(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmadd.c
new file mode 100644
index 0000000..eec662a
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmadd.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m1(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m8(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m1_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m1_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m8_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m8_m(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmax.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmax.c
new file mode 100644
index 0000000..dfdeb4e
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmax.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m8_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmerge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmerge.c
new file mode 100644
index 0000000..96221c5
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmerge.c
@@ -0,0 +1,69 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmerge_vfm_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmerge_vfm_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, vbool64_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16mf4(op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmerge_vfm_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 2 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmerge_vfm_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, vbool32_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16mf2(op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmerge_vfm_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 4 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmerge_vfm_bf16m1(vbfloat16m1_t op1, __bf16 op2, vbool16_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16m1(op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmerge_vfm_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmerge_vfm_bf16m2(vbfloat16m2_t op1, __bf16 op2, vbool8_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16m2(op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmerge_vfm_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 16 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmerge_vfm_bf16m4(vbfloat16m4_t op1, __bf16 op2, vbool4_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16m4(op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmerge_vfm_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmerge_vfm_bf16m8(vbfloat16m8_t op1, __bf16 op2, vbool2_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16m8(op1, op2, mask, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmin.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmin.c
new file mode 100644
index 0000000..8f8d82ba
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmin.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m8_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsac.c
new file mode 100644
index 0000000..f4644df
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsac.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m1(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m8(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m1_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m1_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m8_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m8_m(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsub.c
new file mode 100644
index 0000000..07053afa
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmsub.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m1(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m8(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m1_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m1_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m8_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m8_m(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmul.c
new file mode 100644
index 0000000..88fb329
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmul.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m8_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmv.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmv.c
new file mode 100644
index 0000000..d80ec3d
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfmv.c
@@ -0,0 +1,189 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_v_f_bf16mf4(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.v.f.nxv1bf16.i64(<vscale x 1 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmv_v_f_bf16mf4(__bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16mf4(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_v_f_bf16mf2(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.v.f.nxv2bf16.i64(<vscale x 2 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmv_v_f_bf16mf2(__bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16mf2(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_v_f_bf16m1(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.v.f.nxv4bf16.i64(<vscale x 4 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmv_v_f_bf16m1(__bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16m1(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_v_f_bf16m2(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.v.f.nxv8bf16.i64(<vscale x 8 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmv_v_f_bf16m2(__bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16m2(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_v_f_bf16m4(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.v.f.nxv16bf16.i64(<vscale x 16 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmv_v_f_bf16m4(__bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16m4(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_v_f_bf16m8(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.v.f.nxv32bf16.i64(<vscale x 32 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmv_v_f_bf16m8(__bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16m8(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16mf4_bf16(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv1bf16(<vscale x 1 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16mf4_bf16(vbfloat16mf4_t src) {
+  return __riscv_vfmv_f_s_bf16mf4_bf16(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_s_f_bf16mf4(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.s.f.nxv1bf16.i64(<vscale x 1 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmv_s_f_bf16mf4(__bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16mf4(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16mf2_bf16(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv2bf16(<vscale x 2 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16mf2_bf16(vbfloat16mf2_t src) {
+  return __riscv_vfmv_f_s_bf16mf2_bf16(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_s_f_bf16mf2(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.s.f.nxv2bf16.i64(<vscale x 2 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmv_s_f_bf16mf2(__bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16mf2(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m1_bf16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv4bf16(<vscale x 4 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16m1_bf16(vbfloat16m1_t src) {
+  return __riscv_vfmv_f_s_bf16m1_bf16(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_s_f_bf16m1(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.s.f.nxv4bf16.i64(<vscale x 4 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmv_s_f_bf16m1(__bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16m1(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m2_bf16(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv8bf16(<vscale x 8 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16m2_bf16(vbfloat16m2_t src) {
+  return __riscv_vfmv_f_s_bf16m2_bf16(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_s_f_bf16m2(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.s.f.nxv8bf16.i64(<vscale x 8 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmv_s_f_bf16m2(__bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16m2(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m4_bf16(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv16bf16(<vscale x 16 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16m4_bf16(vbfloat16m4_t src) {
+  return __riscv_vfmv_f_s_bf16m4_bf16(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_s_f_bf16m4(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.s.f.nxv16bf16.i64(<vscale x 16 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmv_s_f_bf16m4(__bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16m4(src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m8_bf16(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv32bf16(<vscale x 32 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16m8_bf16(vbfloat16m8_t src) {
+  return __riscv_vfmv_f_s_bf16m8_bf16(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_s_f_bf16m8(
+// CHECK-RV64-SAME: bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.s.f.nxv32bf16.i64(<vscale x 32 x bfloat> poison, bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmv_s_f_bf16m8(__bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16m8(src, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt.c
new file mode 100644
index 0000000..a5afab9
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt.c
@@ -0,0 +1,724 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4(vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2(vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1(vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2(vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4(vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                           size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_m(vbool32_t vm, vfloat32m1_t vs2,
+                                           size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_m(vbool16_t vm, vfloat32m2_t vs2,
+                                         size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_m(vbool8_t vm, vfloat32m4_t vs2,
+                                         size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_m(vbool4_t vm, vfloat32m8_t vs2,
+                                         size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm(vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm(vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm(vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm(vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm(vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_rm(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_m(vbool64_t vm,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_m(vbool32_t vm,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_m(vbool64_t vm,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_m(vbool32_t vm,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_m(vbool16_t vm,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_m(vbool32_t vm, vfloat32m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_m(vbool16_t vm, vfloat32m2_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_m(vbool8_t vm, vfloat32m4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_m(vbool4_t vm, vfloat32m8_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_rm_m(vm, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rod.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rod.c
new file mode 100644
index 0000000..70c377b
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rod.c
@@ -0,0 +1,113 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4(vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2(vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1(vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m1(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2(vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4(vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_m(vbool32_t vm, vfloat32m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_m(vbool16_t vm, vfloat32m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m1_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_m(vbool8_t vm, vfloat32m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_m(vbool4_t vm, vfloat32m8_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m4_m(vm, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rtz.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rtz.c
new file mode 100644
index 0000000..854e986
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfncvt_rtz.c
@@ -0,0 +1,267 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8(vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4(vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_m(vbool32_t vm,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_m(vbool64_t vm,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_m(vbool32_t vm,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_m(vbool16_t vm,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                                size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4_m(vm, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmacc.c
new file mode 100644
index 0000000..1848488
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmacc.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m1(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m8(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m1_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m1_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m8_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m8_m(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmadd.c
new file mode 100644
index 0000000..e519e5a
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmadd.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m1(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m8(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m1_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m1_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m8_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m8_m(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsac.c
new file mode 100644
index 0000000..47e1f44
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsac.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m1(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m8(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m1_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m1_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m8_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m8_m(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsub.c
new file mode 100644
index 0000000..4b55b64
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfnmsub.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m1(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m2(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m4(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m8(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m1_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m1_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m2_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m2_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m4_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m4_m(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m8_m(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m8_m(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrec7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrec7.c
new file mode 100644
index 0000000..1ffee73
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrec7.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4(vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf4(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2(vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf2(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1(vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m1(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2(vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m2(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4(vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m4(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8(vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m8(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf4_m(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf2_m(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m1_m(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m2_m(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m4_m(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m8_m(mask, op1, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsqrt7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsqrt7.c
new file mode 100644
index 0000000..964c486
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsqrt7.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4(vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf4(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2(vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf2(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1(vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m1(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2(vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m2(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4(vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m4(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8(vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m8(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf4_m(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf2_m(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m1_m(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m2_m(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m4_m(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m8_m(mask, op1, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsub.c
new file mode 100644
index 0000000..c7c3869e
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfrsub.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m8_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnj.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnj.c
new file mode 100644
index 0000000..778b8b83
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnj.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m8_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjn.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjn.c
new file mode 100644
index 0000000..7de3089
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjn.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m8_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjx.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjx.c
new file mode 100644
index 0000000..5fa285cc
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsgnjx.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m8_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1down.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1down.c
new file mode 100644
index 0000000..b94d26b
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1down.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4(vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf4(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2(vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf2(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1(vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m1(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2(vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m2(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4(vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m4(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8(vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m8(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf4_m(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf2_m(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m1_m(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m2_m(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m4_m(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m8_m(mask, src, value, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1up.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1up.c
new file mode 100644
index 0000000..06e8b49
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfslide1up.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4(vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf4(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2(vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf2(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1(vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m1(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2(vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m2(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4(vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m4(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8(vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m8(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf4_m(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf2_m(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m1_m(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m2_m(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m4_m(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m8_m(mask, src, value, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsub.c
new file mode 100644
index 0000000..2423b0b
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfsub.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m1(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m1_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m8_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwadd.c
new file mode 100644
index 0000000..24d34f4
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwadd.c
@@ -0,0 +1,899 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2,
+                                            vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2(vfloat32mf2_t vs2,
+                                            vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2(vfloat32mf2_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2,
+                                          vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1(vfloat32m1_t vs2, vbfloat16mf2_t vs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1(vfloat32m1_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2(vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2(vfloat32m2_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4(vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4(vfloat32m4_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8(vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8(vfloat32m8_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                              vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vs2,
+                                            vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_m(vbool32_t vm, vfloat32m1_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vs2,
+                                           vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_m(vbool16_t vm, vfloat32m2_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vs2,
+                                           vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_m(vbool8_t vm, vfloat32m4_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vs2,
+                                           vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_m(vbool4_t vm, vfloat32m8_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm(vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm(vfloat32mf2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm(vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm(vfloat32m1_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm(vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm(vfloat32m2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm(vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm(vfloat32m4_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm(vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm(vfloat32m8_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vfloat32mf2_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwcvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwcvt.c
new file mode 100644
index 0000000..fb3e003
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwcvt.c
@@ -0,0 +1,366 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4(vint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2(vint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1(vint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m1(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2(vint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4(vint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8(vint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m8(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4(vuint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2(vuint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1(vuint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m1(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2(vuint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4(vuint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8(vuint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m8(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m1_f32m2(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m2_f32m4(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m4_f32m8(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_m(vbool64_t vm, vint8mf8_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_m(vbool32_t vm, vint8mf4_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_m(vbool16_t vm, vint8mf2_t vs2,
+                                         size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m1_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_m(vbool8_t vm, vint8m1_t vs2,
+                                         size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_m(vbool4_t vm, vint8m2_t vs2,
+                                         size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_m(vbool2_t vm, vint8m4_t vs2,
+                                         size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m8_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_m(vbool64_t vm, vuint8mf8_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_m(vbool32_t vm, vuint8mf4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_m(vbool16_t vm, vuint8mf2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m1_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_m(vbool8_t vm, vuint8m1_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_m(vbool4_t vm, vuint8m2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_m(vbool2_t vm, vuint8m4_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m8_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m1_f32m2_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m2_f32m4_m(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m4_f32m8_m(vm, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmacc.c
new file mode 100644
index 0000000..be09003
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmacc.c
@@ -0,0 +1,486 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2(vfloat32mf2_t vd,
+                                             vbfloat16mf4_t vs1,
+                                             vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1,
+                                             vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1,
+                                           vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1,
+                                           vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                          vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1,
+                                          vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                          vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1,
+                                          vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                          vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1,
+                                          vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs1,
+                                               vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                               __bf16 vs1, vbfloat16mf4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs1,
+                                             vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                             __bf16 vs1, vbfloat16mf2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs1,
+                                            vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                            __bf16 vs1, vbfloat16m1_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs1,
+                                            vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                            __bf16 vs1, vbfloat16m2_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs1,
+                                            vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                            __bf16 vs1, vbfloat16m4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmsac.c
new file mode 100644
index 0000000..7490813
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmsac.c
@@ -0,0 +1,486 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2(vfloat32mf2_t vd,
+                                             vbfloat16mf4_t vs1,
+                                             vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1,
+                                             vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1,
+                                           vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1,
+                                           vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                          vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1,
+                                          vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                          vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1,
+                                          vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                          vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1,
+                                          vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs1,
+                                               vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                               __bf16 vs1, vbfloat16mf4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs1,
+                                             vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                             __bf16 vs1, vbfloat16mf2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs1,
+                                            vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                            __bf16 vs1, vbfloat16m1_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs1,
+                                            vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                            __bf16 vs1, vbfloat16m2_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs1,
+                                            vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                            __bf16 vs1, vbfloat16m4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmul.c
new file mode 100644
index 0000000..6783ba4
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwmul.c
@@ -0,0 +1,455 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2,
+                                            vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2,
+                                          vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmacc.c
new file mode 100644
index 0000000..6127a94
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmacc.c
@@ -0,0 +1,494 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2(vfloat32mf2_t vd,
+                                              vbfloat16mf4_t vs1,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1,
+                                            vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1,
+                                            vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                           vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1,
+                                           vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                                __bf16 vs1, vbfloat16mf4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                              __bf16 vs1, vbfloat16mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                             __bf16 vs1, vbfloat16m1_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                             __bf16 vs1, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                             __bf16 vs1, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm(vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm(vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm(vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmsac.c
new file mode 100644
index 0000000..f37dd31
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwnmsac.c
@@ -0,0 +1,494 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2(vfloat32mf2_t vd,
+                                              vbfloat16mf4_t vs1,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1,
+                                            vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1,
+                                            vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                           vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1,
+                                           vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                                __bf16 vs1, vbfloat16mf4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                              __bf16 vs1, vbfloat16mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                             __bf16 vs1, vbfloat16m1_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                             __bf16 vs1, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                             __bf16 vs1, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_m(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm(vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm(vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm(vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm_m(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm_m(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm_m(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwsub.c
new file mode 100644
index 0000000..510ff91
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vfwsub.c
@@ -0,0 +1,899 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2,
+                                            vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2(vfloat32mf2_t vs2,
+                                            vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2(vfloat32mf2_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2,
+                                          vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1(vfloat32m1_t vs2, vbfloat16mf2_t vs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1(vfloat32m1_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2(vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2(vfloat32m2_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4(vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4(vfloat32m4_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8(vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8(vfloat32m8_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                              vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vs2,
+                                            vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_m(vbool32_t vm, vfloat32m1_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vs2,
+                                           vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_m(vbool16_t vm, vfloat32m2_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vs2,
+                                           vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_m(vbool8_t vm, vfloat32m4_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vs2,
+                                           vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_m(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_m(vbool4_t vm, vfloat32m8_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_m(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm(vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm(vfloat32mf2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm(vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm(vfloat32m1_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm(vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm(vfloat32m2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm(vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm(vfloat32m4_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm(vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_rm(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm(vfloat32m8_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_rm(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vfloat32mf2_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_rm_m(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_rm_m(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfeq.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfeq.c
new file mode 100644
index 0000000..669d042
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfeq.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfge.c
new file mode 100644
index 0000000..b169efd
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfge.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfgt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfgt.c
new file mode 100644
index 0000000..9aea7d2
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfgt.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfle.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfle.c
new file mode 100644
index 0000000..40f0c27
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfle.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmflt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmflt.c
new file mode 100644
index 0000000..f64eee3
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmflt.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfne.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfne.c
new file mode 100644
index 0000000..809ea56
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/non-overloaded/vmfne.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16mf4_b64(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16mf2_b32(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m1_b16(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m2_b8(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m4_b4(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m8_b2(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16mf4_b64_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16mf2_b32_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m1_b16_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m2_b8_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m4_b4_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m8_b2_m(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfadd.c
new file mode 100644
index 0000000..9d6b071c
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfadd.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfclass.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfclass.c
new file mode 100644
index 0000000..2760f85
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfclass.c
@@ -0,0 +1,134 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.nxv1bf16.i64(<vscale x 1 x i16> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfclass(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.nxv2bf16.i64(<vscale x 2 x i16> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfclass(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.nxv4bf16.i64(<vscale x 4 x i16> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfclass(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.nxv8bf16.i64(<vscale x 8 x i16> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfclass(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.nxv16bf16.i64(<vscale x 16 x i16> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfclass(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.nxv32bf16.i64(<vscale x 32 x i16> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfclass(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfclass(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfclass(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                          size_t vl) {
+  return __riscv_vfclass(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfclass(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                          size_t vl) {
+  return __riscv_vfclass(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                          size_t vl) {
+  return __riscv_vfclass(vm, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmacc.c
new file mode 100644
index 0000000..ae3f1f2
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmacc.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmadd.c
new file mode 100644
index 0000000..db2184c
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmadd.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmax.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmax.c
new file mode 100644
index 0000000..66497bf
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmax.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmerge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmerge.c
new file mode 100644
index 0000000..1dc290b
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmerge.c
@@ -0,0 +1,69 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmerge_vfm_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmerge_vfm_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, vbool64_t mask, size_t vl) {
+  return __riscv_vfmerge(op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmerge_vfm_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 2 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmerge_vfm_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, vbool32_t mask, size_t vl) {
+  return __riscv_vfmerge(op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmerge_vfm_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 4 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmerge_vfm_bf16m1(vbfloat16m1_t op1, __bf16 op2, vbool16_t mask, size_t vl) {
+  return __riscv_vfmerge(op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmerge_vfm_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmerge_vfm_bf16m2(vbfloat16m2_t op1, __bf16 op2, vbool8_t mask, size_t vl) {
+  return __riscv_vfmerge(op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmerge_vfm_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 16 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmerge_vfm_bf16m4(vbfloat16m4_t op1, __bf16 op2, vbool4_t mask, size_t vl) {
+  return __riscv_vfmerge(op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmerge_vfm_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmerge_vfm_bf16m8(vbfloat16m8_t op1, __bf16 op2, vbool2_t mask, size_t vl) {
+  return __riscv_vfmerge(op1, op2, mask, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmin.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmin.c
new file mode 100644
index 0000000..1564d11
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmin.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsac.c
new file mode 100644
index 0000000..0384e7d
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsac.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsub.c
new file mode 100644
index 0000000..306f189
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmsub.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmul.c
new file mode 100644
index 0000000..fffd83a
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmul.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmv.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmv.c
new file mode 100644
index 0000000..f85378f
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfmv.c
@@ -0,0 +1,69 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16mf4_bf16(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv1bf16(<vscale x 1 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16mf4_bf16(vbfloat16mf4_t src) {
+  return __riscv_vfmv_f(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16mf2_bf16(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv2bf16(<vscale x 2 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16mf2_bf16(vbfloat16mf2_t src) {
+  return __riscv_vfmv_f(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m1_bf16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv4bf16(<vscale x 4 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16m1_bf16(vbfloat16m1_t src) {
+  return __riscv_vfmv_f(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m2_bf16(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv8bf16(<vscale x 8 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16m2_bf16(vbfloat16m2_t src) {
+  return __riscv_vfmv_f(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m4_bf16(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv16bf16(<vscale x 16 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16m4_bf16(vbfloat16m4_t src) {
+  return __riscv_vfmv_f(src);
+}
+
+// CHECK-RV64-LABEL: define dso_local bfloat @test_vfmv_f_s_bf16m8_bf16(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call bfloat @llvm.riscv.vfmv.f.s.nxv32bf16(<vscale x 32 x bfloat> [[SRC]])
+// CHECK-RV64-NEXT:    ret bfloat [[TMP0]]
+//
+__bf16 test_vfmv_f_s_bf16m8_bf16(vbfloat16m8_t src) {
+  return __riscv_vfmv_f(src);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt.c
new file mode 100644
index 0000000..fb635d6
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt.c
@@ -0,0 +1,724 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4(vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2(vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1(vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2(vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4(vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                           size_t vl) {
+  return __riscv_vfncvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_m(vbool32_t vm, vfloat32m1_t vs2,
+                                           size_t vl) {
+  return __riscv_vfncvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_m(vbool16_t vm, vfloat32m2_t vs2,
+                                         size_t vl) {
+  return __riscv_vfncvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_m(vbool8_t vm, vfloat32m4_t vs2,
+                                         size_t vl) {
+  return __riscv_vfncvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_m(vbool4_t vm, vfloat32m8_t vs2,
+                                         size_t vl) {
+  return __riscv_vfncvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm(vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm(vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm(vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm(vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm(vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16(vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_m(vbool64_t vm,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_m(vbool32_t vm,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_m(vbool64_t vm,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_m(vbool32_t vm,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_m(vbool16_t vm,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_xu(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_f_bf16(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_m(vbool32_t vm, vfloat32m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_f_bf16(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_m(vbool16_t vm, vfloat32m2_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_f_bf16(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_m(vbool8_t vm, vfloat32m4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_f_bf16(vm, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_m(vbool4_t vm, vfloat32m8_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_f_bf16(vm, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rod.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rod.c
new file mode 100644
index 0000000..1ad856d
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rod.c
@@ -0,0 +1,113 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4(vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2(vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1(vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2(vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4(vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_m(vbool32_t vm, vfloat32m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_m(vbool16_t vm, vfloat32m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_m(vbool8_t vm, vfloat32m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_m(vbool4_t vm, vfloat32m8_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16(vm, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rtz.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rtz.c
new file mode 100644
index 0000000..12d0893
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfncvt_rtz.c
@@ -0,0 +1,267 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8(vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4(vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4(vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_rtz_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_m(vbool32_t vm,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_rtz_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfncvt_rtz_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_rtz_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_rtz_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_rtz_x(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_m(vbool64_t vm,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_m(vbool32_t vm,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_m(vbool16_t vm,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> poison, <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_m(vbool2_t vm, vbfloat16m8_t vs2,
+                                                size_t vl) {
+  return __riscv_vfncvt_rtz_xu(vm, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmacc.c
new file mode 100644
index 0000000..6f7928b
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmacc.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmadd.c
new file mode 100644
index 0000000..97d2070
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmadd.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsac.c
new file mode 100644
index 0000000..404b4f8
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsac.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsub.c
new file mode 100644
index 0000000..3a520dd
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfnmsub.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrec7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrec7.c
new file mode 100644
index 0000000..462b6ac
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrec7.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4(vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2(vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1(vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2(vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4(vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8(vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7(mask, op1, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsqrt7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsqrt7.c
new file mode 100644
index 0000000..051fde7
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsqrt7.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4(vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2(vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1(vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2(vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4(vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8(vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(mask, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7(mask, op1, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsub.c
new file mode 100644
index 0000000..0494182
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfrsub.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnj.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnj.c
new file mode 100644
index 0000000..615dedd
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnj.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjn.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjn.c
new file mode 100644
index 0000000..a895e5f
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjn.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjx.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjx.c
new file mode 100644
index 0000000..0187516
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsgnjx.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1down.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1down.c
new file mode 100644
index 0000000..4a76894
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1down.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4(vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2(vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1(vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2(vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4(vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8(vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down(mask, src, value, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1up.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1up.c
new file mode 100644
index 0000000..f9f2dc0
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfslide1up.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4(vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2(vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1(vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2(vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4(vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8(vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(mask, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up(mask, src, value, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsub.c
new file mode 100644
index 0000000..ebcf6fa
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfsub.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwadd.c
new file mode 100644
index 0000000..124e7fb
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwadd.c
@@ -0,0 +1,893 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2,
+                                            vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2(vfloat32mf2_t vs2,
+                                            vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2(vfloat32mf2_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_wf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2,
+                                          vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_vf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1(vfloat32m1_t vs2, vbfloat16mf2_t vs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1(vfloat32m1_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwadd_wf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2(vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_wv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2(vfloat32m2_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwadd_wf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4(vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_wv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4(vfloat32m4_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwadd_wf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_vf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8(vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwadd_wv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8(vfloat32m8_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwadd_wf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                              vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vs2,
+                                            vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_m(vbool32_t vm, vfloat32m1_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vs2,
+                                           vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_m(vbool16_t vm, vfloat32m2_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vs2,
+                                           vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_m(vbool8_t vm, vfloat32m4_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vs2,
+                                           vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_m(vbool4_t vm, vfloat32m8_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm(vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm(vfloat32mf2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm(vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm(vfloat32m1_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm(vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm(vfloat32m2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm(vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm(vfloat32m4_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm(vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm(vfloat32m8_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vfloat32mf2_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwcvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwcvt.c
new file mode 100644
index 0000000..0399a63
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwcvt.c
@@ -0,0 +1,366 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4(vint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2(vint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1(vint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2(vint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4(vint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8(vint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4(vuint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2(vuint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1(vuint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2(vuint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4(vuint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8(vuint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2(vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1(vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2(vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4(vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8(vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f(vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_m(vbool64_t vm, vint8mf8_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_m(vbool32_t vm, vint8mf4_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_m(vbool16_t vm, vint8mf2_t vs2,
+                                         size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_m(vbool8_t vm, vint8m1_t vs2,
+                                         size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_m(vbool4_t vm, vint8m2_t vs2,
+                                         size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_m(vbool2_t vm, vint8m4_t vs2,
+                                         size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> poison, <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_m(vbool64_t vm, vuint8mf8_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> poison, <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_m(vbool32_t vm, vuint8mf4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> poison, <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_m(vbool16_t vm, vuint8mf2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> poison, <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_m(vbool8_t vm, vuint8m1_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> poison, <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_m(vbool4_t vm, vuint8m2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> poison, <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_m(vbool2_t vm, vuint8m4_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_bf16(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwcvt_f(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwcvt_f(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwcvt_f(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwcvt_f(vm, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwcvt_f(vm, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmacc.c
new file mode 100644
index 0000000..2eb7fc8
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmacc.c
@@ -0,0 +1,474 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2(vfloat32mf2_t vd,
+                                             vbfloat16mf4_t vs1,
+                                             vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1,
+                                             vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1,
+                                           vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1,
+                                           vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                          vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1,
+                                          vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                          vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1,
+                                          vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                          vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1,
+                                          vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs1,
+                                               vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                               __bf16 vs1, vbfloat16mf4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs1,
+                                             vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                             __bf16 vs1, vbfloat16mf2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs1,
+                                            vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                            __bf16 vs1, vbfloat16m1_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs1,
+                                            vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                            __bf16 vs1, vbfloat16m2_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs1,
+                                            vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                            __bf16 vs1, vbfloat16m4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmsac.c
new file mode 100644
index 0000000..28f5076
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmsac.c
@@ -0,0 +1,474 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2(vfloat32mf2_t vd,
+                                             vbfloat16mf4_t vs1,
+                                             vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1,
+                                             vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1,
+                                           vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1,
+                                           vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                          vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1,
+                                          vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                          vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1,
+                                          vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                          vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1,
+                                          vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs1,
+                                               vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                               __bf16 vs1, vbfloat16mf4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs1,
+                                             vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                             __bf16 vs1, vbfloat16mf2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs1,
+                                            vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                            __bf16 vs1, vbfloat16m1_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs1,
+                                            vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                            __bf16 vs1, vbfloat16m2_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs1,
+                                            vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                            __bf16 vs1, vbfloat16m4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmul.c
new file mode 100644
index 0000000..8de49fa
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwmul.c
@@ -0,0 +1,451 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2,
+                                            vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2,
+                                          vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwmul(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwmul(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwmul(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwmul(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwmul(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwmul(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwmul(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmacc.c
new file mode 100644
index 0000000..7836931
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmacc.c
@@ -0,0 +1,480 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2(vfloat32mf2_t vd,
+                                              vbfloat16mf4_t vs1,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1,
+                                            vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1,
+                                            vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                           vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1,
+                                           vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                                __bf16 vs1, vbfloat16mf4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                              __bf16 vs1, vbfloat16mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                             __bf16 vs1, vbfloat16m1_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                             __bf16 vs1, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                             __bf16 vs1, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm(vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm(vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm(vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmsac.c
new file mode 100644
index 0000000..ca936af
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwnmsac.c
@@ -0,0 +1,480 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2(vfloat32mf2_t vd,
+                                              vbfloat16mf4_t vs1,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2(vfloat32mf2_t vd, __bf16 vs1,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1(vfloat32m1_t vd, vbfloat16mf2_t vs1,
+                                            vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1(vfloat32m1_t vd, __bf16 vs1,
+                                            vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                           vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2(vfloat32m2_t vd, __bf16 vs1,
+                                           vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4(vfloat32m4_t vd, __bf16 vs1,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8(vfloat32m8_t vd, __bf16 vs1,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vd,
+                                                __bf16 vs1, vbfloat16mf4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vd,
+                                              __bf16 vs1, vbfloat16mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vd,
+                                             __bf16 vs1, vbfloat16m1_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vd,
+                                             __bf16 vs1, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vd,
+                                             __bf16 vs1, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm(vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm(vfloat32mf2_t vd, __bf16 vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm(vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm(vfloat32m1_t vd, __bf16 vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm(vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm(vfloat32m2_t vd, __bf16 vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm(vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm(vfloat32m4_t vd, __bf16 vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm(vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm(vfloat32m8_t vd, __bf16 vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwsub.c
new file mode 100644
index 0000000..2e22e22
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vfwsub.c
@@ -0,0 +1,893 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2(vbfloat16mf4_t vs2,
+                                            vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2(vbfloat16mf4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2(vfloat32mf2_t vs2,
+                                            vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2(vfloat32mf2_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_wf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1(vbfloat16mf2_t vs2,
+                                          vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1(vbfloat16mf2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_vf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1(vfloat32m1_t vs2, vbfloat16mf2_t vs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1(vfloat32m1_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwsub_wf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2(vbfloat16m1_t vs2, vbfloat16m1_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2(vbfloat16m1_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2(vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_wv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2(vfloat32m2_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwsub_wf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4(vbfloat16m2_t vs2, vbfloat16m2_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4(vbfloat16m2_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4(vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_wv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4(vfloat32m4_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwsub_wf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8(vbfloat16m4_t vs2, vbfloat16m4_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8(vbfloat16m4_t vs2, __bf16 rs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_vf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8(vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                         size_t vl) {
+  return __riscv_vfwsub_wv(vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8(vfloat32m8_t vs2, __bf16 rs1,
+                                       size_t vl) {
+  return __riscv_vfwsub_wf(vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_m(vbool64_t vm, vbfloat16mf4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                              vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_m(vbool32_t vm, vfloat32m1_t vs2,
+                                            vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_m(vbool32_t vm, vfloat32m1_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_m(vbool16_t vm, vfloat32m2_t vs2,
+                                           vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_m(vbool16_t vm, vfloat32m2_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_m(vbool8_t vm, vfloat32m4_t vs2,
+                                           vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_m(vbool8_t vm, vfloat32m4_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                           __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_m(vbool4_t vm, vfloat32m8_t vs2,
+                                           vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vm, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_m(vbool4_t vm, vfloat32m8_t vs2,
+                                         __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf(vm, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm(vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm(vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm(vfloat32mf2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm(vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm(vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm(vfloat32m1_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm(vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm(vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm(vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm(vfloat32m2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm(vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm(vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm(vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm(vfloat32m4_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm(vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm(vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm(vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv(vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm(vfloat32m8_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf(vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_m(vbool64_t vm,
+                                                 vfloat32mf2_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> poison, <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_m(vbool64_t vm, vfloat32mf2_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_m(vbool32_t vm, vbfloat16mf2_t vs2,
+                                               __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_m(vbool32_t vm, vfloat32m1_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_m(vbool16_t vm, vbfloat16m1_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> poison, <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_m(vbool16_t vm, vfloat32m2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_m(vbool8_t vm, vbfloat16m2_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> poison, <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_m(vbool8_t vm, vfloat32m4_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_m(vbool4_t vm, vbfloat16m4_t vs2,
+                                              __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv(vm, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> poison, <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_m(vbool4_t vm, vfloat32m8_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf(vm, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfeq.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfeq.c
new file mode 100644
index 0000000..29881c9
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfeq.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfge.c
new file mode 100644
index 0000000..b8083c5e
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfge.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfgt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfgt.c
new file mode 100644
index 0000000..b8749b3
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfgt.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfle.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfle.c
new file mode 100644
index 0000000..724608c
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfle.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmflt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmflt.c
new file mode 100644
index 0000000..1b0b898
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmflt.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfne.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfne.c
new file mode 100644
index 0000000..672c150
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/non-policy/overloaded/vmfne.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vv_bf16mf4_b64(vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vf_bf16mf4_b64(vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vv_bf16mf2_b32(vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vf_bf16mf2_b32(vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vv_bf16m1_b16(vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vf_bf16m1_b16(vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vv_bf16m2_b8(vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vf_bf16m2_b8(vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vv_bf16m4_b4(vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vf_bf16m4_b4(vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vv_bf16m8_b2(vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vf_bf16m8_b2(vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vv_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64_m(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> poison, <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vf_bf16mf4_b64_m(vbool64_t mask, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vv_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32_m(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> poison, <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vf_bf16mf2_b32_m(vbool32_t mask, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vv_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16_m(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> poison, <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vf_bf16m1_b16_m(vbool16_t mask, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vv_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8_m(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> poison, <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vf_bf16m2_b8_m(vbool8_t mask, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vv_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4_m(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> poison, <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vf_bf16m4_b4_m(vbool4_t mask, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vv_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2_m(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> poison, <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vf_bf16m8_b2_m(vbool2_t mask, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne(mask, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfadd.c
new file mode 100644
index 0000000..6d55279
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfadd.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfclass.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfclass.c
new file mode 100644
index 0000000..8e6946d
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfclass.c
@@ -0,0 +1,272 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tu(vuint16mf4_t vd,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16mf4_u16mf4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tu(vuint16mf2_t vd,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16mf2_u16mf2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1_tu(vuint16m1_t vd, vbfloat16m1_t vs2,
+                                           size_t vl) {
+  return __riscv_vfclass_v_bf16m1_u16m1_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2_tu(vuint16m2_t vd, vbfloat16m2_t vs2,
+                                           size_t vl) {
+  return __riscv_vfclass_v_bf16m2_u16m2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4_tu(vuint16m4_t vd, vbfloat16m4_t vs2,
+                                           size_t vl) {
+  return __riscv_vfclass_v_bf16m4_u16m4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8_tu(vuint16m8_t vd, vbfloat16m8_t vs2,
+                                           size_t vl) {
+  return __riscv_vfclass_v_bf16m8_u16m8_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tum(vbool64_t vm, vuint16mf4_t vd,
+                                               vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16mf4_u16mf4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tum(vbool32_t vm, vuint16mf2_t vd,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16mf2_u16mf2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1_tum(vbool16_t vm, vuint16m1_t vd,
+                                            vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m1_u16m1_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2_tum(vbool8_t vm, vuint16m2_t vd,
+                                            vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m2_u16m2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4_tum(vbool4_t vm, vuint16m4_t vd,
+                                            vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m4_u16m4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8_tum(vbool2_t vm, vuint16m8_t vd,
+                                            vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m8_u16m8_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tumu(vbool64_t vm, vuint16mf4_t vd,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16mf4_u16mf4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tumu(vbool32_t vm, vuint16mf2_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16mf2_u16mf2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1_tumu(vbool16_t vm, vuint16m1_t vd,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m1_u16m1_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2_tumu(vbool8_t vm, vuint16m2_t vd,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m2_u16m2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4_tumu(vbool4_t vm, vuint16m4_t vd,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m4_u16m4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8_tumu(vbool2_t vm, vuint16m8_t vd,
+                                             vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m8_u16m8_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_mu(vbool64_t vm, vuint16mf4_t vd,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16mf4_u16mf4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_mu(vbool32_t vm, vuint16mf2_t vd,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16mf2_u16mf2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1_mu(vbool16_t vm, vuint16m1_t vd,
+                                           vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m1_u16m1_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2_mu(vbool8_t vm, vuint16m2_t vd,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m2_u16m2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4_mu(vbool4_t vm, vuint16m4_t vd,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m4_u16m4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8_mu(vbool2_t vm, vuint16m8_t vd,
+                                           vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfclass_v_bf16m8_u16m8_mu(vm, vd, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmacc.c
new file mode 100644
index 0000000..2d4e481
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmacc.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m1_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m8_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m1_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m1_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m8_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m8_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m1_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m1_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vv_bf16m8_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_vf_bf16m8_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmadd.c
new file mode 100644
index 0000000..511e073
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmadd.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m1_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m8_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m1_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m1_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m8_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m8_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m1_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m1_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vv_bf16m8_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_vf_bf16m8_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmax.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmax.c
new file mode 100644
index 0000000..f3698d4
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmax.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmerge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmerge.c
new file mode 100644
index 0000000..bcaf2cb
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmerge.c
@@ -0,0 +1,69 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmerge_vfm_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmerge_vfm_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, vbool64_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16mf4_tu(maskedoff, op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmerge_vfm_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 2 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmerge_vfm_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, vbool32_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16mf2_tu(maskedoff, op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmerge_vfm_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 4 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmerge_vfm_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, vbool16_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16m1_tu(maskedoff, op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmerge_vfm_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmerge_vfm_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, vbool8_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16m2_tu(maskedoff, op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmerge_vfm_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 16 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmerge_vfm_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, vbool4_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16m4_tu(maskedoff, op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmerge_vfm_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmerge_vfm_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, vbool2_t mask, size_t vl) {
+  return __riscv_vfmerge_vfm_bf16m8_tu(maskedoff, op1, op2, mask, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmin.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmin.c
new file mode 100644
index 0000000..911f879
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmin.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsac.c
new file mode 100644
index 0000000..9575ad3
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsac.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m1_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m8_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m1_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m1_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m8_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m8_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m1_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m1_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vv_bf16m8_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_vf_bf16m8_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsub.c
new file mode 100644
index 0000000..8e382f71
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmsub.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m1_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m8_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m1_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m1_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m8_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m8_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m1_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m1_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vv_bf16m8_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_vf_bf16m8_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmul.c
new file mode 100644
index 0000000..716f056
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmul.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmv.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmv.c
new file mode 100644
index 0000000..069ee6a
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfmv.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_v_f_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.v.f.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmv_v_f_bf16mf4_tu(vbfloat16mf4_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16mf4_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_v_f_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.v.f.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmv_v_f_bf16mf2_tu(vbfloat16mf2_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16mf2_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_v_f_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.v.f.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmv_v_f_bf16m1_tu(vbfloat16m1_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16m1_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_v_f_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.v.f.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmv_v_f_bf16m2_tu(vbfloat16m2_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16m2_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_v_f_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.v.f.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmv_v_f_bf16m4_tu(vbfloat16m4_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16m4_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_v_f_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.v.f.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmv_v_f_bf16m8_tu(vbfloat16m8_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_f_bf16m8_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_s_f_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.s.f.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmv_s_f_bf16mf4_tu(vbfloat16mf4_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16mf4_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_s_f_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.s.f.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmv_s_f_bf16mf2_tu(vbfloat16mf2_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16mf2_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_s_f_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.s.f.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmv_s_f_bf16m1_tu(vbfloat16m1_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16m1_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_s_f_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.s.f.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmv_s_f_bf16m2_tu(vbfloat16m2_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16m2_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_s_f_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.s.f.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmv_s_f_bf16m4_tu(vbfloat16m4_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16m4_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_s_f_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.s.f.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmv_s_f_bf16m8_tu(vbfloat16m8_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_f_bf16m8_tu(maskedoff, src, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt.c
new file mode 100644
index 0000000..36d4fc3
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt.c
@@ -0,0 +1,1577 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tu(vint8mf8_t vd, vbfloat16mf4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tu(vint8mf4_t vd, vbfloat16mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tu(vint8mf2_t vd, vbfloat16m1_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tu(vint8m1_t vd, vbfloat16m2_t vs2,
+                                           size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tu(vint8m2_t vd, vbfloat16m4_t vs2,
+                                           size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tu(vint8m4_t vd, vbfloat16m8_t vs2,
+                                           size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tu(vuint8mf8_t vd,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tu(vuint8mf4_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tu(vuint8mf2_t vd,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tu(vuint8m1_t vd, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tu(vuint8m2_t vd, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tu(vuint8m4_t vd, vbfloat16m8_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tu(vbfloat16mf4_t vd,
+                                            vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tu(vbfloat16mf2_t vd, vfloat32m1_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tu(vbfloat16m1_t vd, vfloat32m2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tu(vbfloat16m2_t vd, vfloat32m4_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tu(vbfloat16m4_t vd, vfloat32m8_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tum(vbool64_t vm, vint8mf8_t vd,
+                                               vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tum(vbool32_t vm, vint8mf4_t vd,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tum(vbool16_t vm, vint8mf2_t vd,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tum(vbool8_t vm, vint8m1_t vd,
+                                            vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tum(vbool4_t vm, vint8m2_t vd,
+                                            vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tum(vbool2_t vm, vint8m4_t vd,
+                                            vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tum(vbool64_t vm, vuint8mf8_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tum(vbool32_t vm, vuint8mf4_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tum(vbool16_t vm, vuint8mf2_t vd,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tum(vbool8_t vm, vuint8m1_t vd,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tum(vbool4_t vm, vuint8m2_t vd,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tum(vbool2_t vm, vuint8m4_t vd,
+                                              vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd,
+                                             vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd,
+                                             vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd,
+                                           vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd,
+                                           vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd,
+                                           vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tumu(vbool64_t vm, vint8mf8_t vd,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tumu(vbool32_t vm, vint8mf4_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tumu(vbool16_t vm, vint8mf2_t vd,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tumu(vbool8_t vm, vint8m1_t vd,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tumu(vbool4_t vm, vint8m2_t vd,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tumu(vbool2_t vm, vint8m4_t vd,
+                                             vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tumu(vbool64_t vm, vuint8mf8_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tumu(vbool32_t vm, vuint8mf4_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tumu(vbool16_t vm, vuint8mf2_t vd,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tumu(vbool8_t vm, vuint8m1_t vd,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tumu(vbool4_t vm, vuint8m2_t vd,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tumu(vbool2_t vm, vuint8m4_t vd,
+                                               vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd,
+                                              vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd,
+                                              vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd,
+                                            vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd,
+                                            vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd,
+                                            vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_mu(vbool64_t vm, vint8mf8_t vd,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_mu(vbool32_t vm, vint8mf4_t vd,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_mu(vbool16_t vm, vint8mf2_t vd,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_mu(vbool8_t vm, vint8m1_t vd,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_mu(vbool4_t vm, vint8m2_t vd,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_mu(vbool2_t vm, vint8m4_t vd,
+                                           vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_mu(vbool64_t vm, vuint8mf8_t vd,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_mu(vbool32_t vm, vuint8mf4_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_mu(vbool16_t vm, vuint8mf2_t vd,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_mu(vbool8_t vm, vuint8m1_t vd,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_mu(vbool4_t vm, vuint8m2_t vd,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_mu(vbool2_t vm, vuint8m4_t vd,
+                                             vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd,
+                                            vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd,
+                                            vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd,
+                                          vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd,
+                                          vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd,
+                                          vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tu(vint8mf8_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tu(vint8mf4_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tu(vint8mf2_t vd,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tu(vint8m1_t vd, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tu(vint8m2_t vd, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tu(vint8m4_t vd, vbfloat16m8_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tu(vuint8mf8_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tu(vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tu(vuint8mf4_t vd,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tu(vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tu(vuint8mf2_t vd,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tu(vuint8m1_t vd,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tu(vuint8m2_t vd,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tu(vuint8m4_t vd,
+                                                vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tu(vbfloat16mf4_t vd,
+                                               vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tu(vbfloat16mf2_t vd,
+                                               vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tu(vbfloat16m1_t vd, vfloat32m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tu(vbfloat16m2_t vd, vfloat32m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tu(vbfloat16m4_t vd, vfloat32m8_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_rm_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tum(vbool64_t vm, vint8mf8_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tum(vbool32_t vm, vint8mf4_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tum(vbool16_t vm, vint8mf2_t vd,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tum(vbool8_t vm, vint8m1_t vd,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tum(vbool4_t vm, vint8m2_t vd,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tum(vbool2_t vm, vint8m4_t vd,
+                                               vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tum(vbool64_t vm,
+                                                    vuint8mf8_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tum(vm, vd, vs2,
+                                                    __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tum(vbool32_t vm,
+                                                    vuint8mf4_t vd,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tum(vm, vd, vs2,
+                                                    __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tum(vbool16_t vm, vuint8mf2_t vd,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tum(vbool8_t vm, vuint8m1_t vd,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tum(vbool4_t vm, vuint8m2_t vd,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tum(vbool2_t vm, vuint8m4_t vd,
+                                                 vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tum(vbool64_t vm, vbfloat16mf4_t vd,
+                                                vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tum(vbool32_t vm, vbfloat16mf2_t vd,
+                                                vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tum(vbool16_t vm, vbfloat16m1_t vd,
+                                              vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tum(vbool8_t vm, vbfloat16m2_t vd,
+                                              vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tum(vbool4_t vm, vbfloat16m4_t vd,
+                                              vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_rm_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tumu(vbool64_t vm, vint8mf8_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tumu(vm, vd, vs2,
+                                                    __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tumu(vbool32_t vm, vint8mf4_t vd,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tumu(vm, vd, vs2,
+                                                    __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tumu(vbool16_t vm, vint8mf2_t vd,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tumu(vbool8_t vm, vint8m1_t vd,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tumu(vbool4_t vm, vint8m2_t vd,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tumu(vbool2_t vm, vint8m4_t vd,
+                                                vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tumu(vbool64_t vm,
+                                                     vuint8mf8_t vd,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tumu(vm, vd, vs2,
+                                                     __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tumu(vbool32_t vm,
+                                                     vuint8mf4_t vd,
+                                                     vbfloat16mf2_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tumu(vm, vd, vs2,
+                                                     __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tumu(vbool16_t vm,
+                                                    vuint8mf2_t vd,
+                                                    vbfloat16m1_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tumu(vm, vd, vs2,
+                                                    __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tumu(vbool8_t vm, vuint8m1_t vd,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tumu(vbool4_t vm, vuint8m2_t vd,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tumu(vbool2_t vm, vuint8m4_t vd,
+                                                  vbfloat16m8_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tumu(vbool64_t vm,
+                                                 vbfloat16mf4_t vd,
+                                                 vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tumu(vbool32_t vm,
+                                                 vbfloat16mf2_t vd,
+                                                 vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tumu(vbool16_t vm, vbfloat16m1_t vd,
+                                               vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tumu(vbool8_t vm, vbfloat16m2_t vd,
+                                               vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tumu(vbool4_t vm, vbfloat16m4_t vd,
+                                               vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_rm_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_mu(vbool64_t vm, vint8mf8_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf4_i8mf8_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_mu(vbool32_t vm, vint8mf4_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16mf2_i8mf4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_mu(vbool16_t vm, vint8mf2_t vd,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m1_i8mf2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_mu(vbool8_t vm, vint8m1_t vd,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m2_i8m1_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_mu(vbool4_t vm, vint8m2_t vd,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m4_i8m2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_mu(vbool2_t vm, vint8m4_t vd,
+                                              vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_f_w_bf16m8_i8m4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_mu(vbool64_t vm, vuint8mf8_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_mu(vbool32_t vm, vuint8mf4_t vd,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                   vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_mu(vbool16_t vm, vuint8mf2_t vd,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m1_u8mf2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_mu(vbool8_t vm, vuint8m1_t vd,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m2_u8m1_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_mu(vbool4_t vm, vuint8m2_t vd,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m4_u8m2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_mu(vbool2_t vm, vuint8m4_t vd,
+                                                vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_f_w_bf16m8_u8m4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_mu(vbool64_t vm, vbfloat16mf4_t vd,
+                                               vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_mu(vbool32_t vm, vbfloat16mf2_t vd,
+                                               vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16mf2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_mu(vbool16_t vm, vbfloat16m1_t vd,
+                                             vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m1_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_mu(vbool8_t vm, vbfloat16m2_t vd,
+                                             vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m2_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_mu(vbool4_t vm, vbfloat16m4_t vd,
+                                             vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_f_w_bf16m4_rm_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rod.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rod.c
new file mode 100644
index 0000000..8406684
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rod.c
@@ -0,0 +1,233 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tu(vbfloat16mf4_t vd,
+                                                vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tu(vbfloat16mf2_t vd,
+                                                vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tu(vbfloat16m1_t vd,
+                                              vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m1_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tu(vbfloat16m2_t vd,
+                                              vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tu(vbfloat16m4_t vd,
+                                              vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tum(vbool64_t vm,
+                                                 vbfloat16mf4_t vd,
+                                                 vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tum(vbool32_t vm,
+                                                 vbfloat16mf2_t vd,
+                                                 vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd,
+                                               vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m1_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd,
+                                               vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd,
+                                               vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tumu(vbool64_t vm,
+                                                  vbfloat16mf4_t vd,
+                                                  vfloat32mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tumu(vbool32_t vm,
+                                                  vbfloat16mf2_t vd,
+                                                  vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd,
+                                                vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m1_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd,
+                                                vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd,
+                                                vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd,
+                                                vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd,
+                                                vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16mf2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd,
+                                              vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m1_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd,
+                                              vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd,
+                                              vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_f_w_bf16m4_mu(vm, vd, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rtz.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rtz.c
new file mode 100644
index 0000000..4644eff
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfncvt_rtz.c
@@ -0,0 +1,572 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tu(vint8mf8_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tu(vint8mf4_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tu(vint8mf2_t vd,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tu(vint8m1_t vd, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tu(vint8m2_t vd, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tu(vint8m4_t vd, vbfloat16m8_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tu(vuint8mf8_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tu(vuint8mf4_t vd,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tu(vuint8mf2_t vd,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tu(vuint8m1_t vd,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tu(vuint8m2_t vd,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tu(vuint8m4_t vd,
+                                                 vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tum(vbool64_t vm, vint8mf8_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tum(vbool32_t vm, vint8mf4_t vd,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tum(vbool16_t vm, vint8mf2_t vd,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tum(vbool8_t vm, vint8m1_t vd,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tum(vbool4_t vm, vint8m2_t vd,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tum(vbool2_t vm, vint8m4_t vd,
+                                                vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tum(vbool64_t vm,
+                                                     vuint8mf8_t vd,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tum(vbool32_t vm,
+                                                     vuint8mf4_t vd,
+                                                     vbfloat16mf2_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tum(vbool16_t vm,
+                                                    vuint8mf2_t vd,
+                                                    vbfloat16m1_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tum(vbool8_t vm, vuint8m1_t vd,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tum(vbool4_t vm, vuint8m2_t vd,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tum(vbool2_t vm, vuint8m4_t vd,
+                                                  vbfloat16m8_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tumu(vbool64_t vm, vint8mf8_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tumu(vbool32_t vm, vint8mf4_t vd,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tumu(vbool16_t vm, vint8mf2_t vd,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tumu(vbool8_t vm, vint8m1_t vd,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tumu(vbool4_t vm, vint8m2_t vd,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tumu(vbool2_t vm, vint8m4_t vd,
+                                                 vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tumu(vbool64_t vm,
+                                                      vuint8mf8_t vd,
+                                                      vbfloat16mf4_t vs2,
+                                                      size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tumu(vbool32_t vm,
+                                                      vuint8mf4_t vd,
+                                                      vbfloat16mf2_t vs2,
+                                                      size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tumu(vbool16_t vm,
+                                                     vuint8mf2_t vd,
+                                                     vbfloat16m1_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tumu(vbool8_t vm, vuint8m1_t vd,
+                                                   vbfloat16m2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tumu(vbool4_t vm, vuint8m2_t vd,
+                                                   vbfloat16m4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tumu(vbool2_t vm, vuint8m4_t vd,
+                                                   vbfloat16m8_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_mu(vbool64_t vm, vint8mf8_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_mu(vbool32_t vm, vint8mf4_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_mu(vbool16_t vm, vint8mf2_t vd,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m1_i8mf2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_mu(vbool8_t vm, vint8m1_t vd,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m2_i8m1_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_mu(vbool4_t vm, vint8m2_t vd,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m4_i8m2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_mu(vbool2_t vm, vint8m4_t vd,
+                                               vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_f_w_bf16m8_i8m4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_mu(vbool64_t vm,
+                                                    vuint8mf8_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_mu(vbool32_t vm,
+                                                    vuint8mf4_t vd,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_mu(vbool16_t vm, vuint8mf2_t vd,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_mu(vbool8_t vm, vuint8m1_t vd,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m2_u8m1_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_mu(vbool4_t vm, vuint8m2_t vd,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m4_u8m2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_mu(vbool2_t vm, vuint8m4_t vd,
+                                                 vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_f_w_bf16m8_u8m4_mu(vm, vd, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmacc.c
new file mode 100644
index 0000000..93fd6ba
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmacc.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m1_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m8_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m1_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m1_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m8_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m8_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m1_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m1_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vv_bf16m8_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_vf_bf16m8_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmadd.c
new file mode 100644
index 0000000..d7e6b82
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmadd.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m1_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m8_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m1_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m1_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m8_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m8_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m1_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m1_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vv_bf16m8_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_vf_bf16m8_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsac.c
new file mode 100644
index 0000000..e0c289d
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsac.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m1_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m8_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m1_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m1_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m8_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m8_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m1_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m1_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vv_bf16m8_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_vf_bf16m8_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsub.c
new file mode 100644
index 0000000..05ccda3
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfnmsub.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m1_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m2_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m4_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m8_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m1_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m1_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m2_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m2_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m4_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m4_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m8_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m8_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m1_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m1_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m2_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m2_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m4_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m4_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m8_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m8_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16mf2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16mf2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m1_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m1_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m2_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m2_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m4_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m4_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vv_bf16m8_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_vf_bf16m8_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrec7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrec7.c
new file mode 100644
index 0000000..3123692
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrec7.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf4_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf2_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m1_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m2_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m4_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m8_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf4_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf2_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m1_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m2_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m4_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m8_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf4_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf2_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m1_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m2_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m4_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m8_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf4_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16mf2_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m1_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m2_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m4_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7_v_bf16m8_mu(mask, maskedoff, op1, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsqrt7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsqrt7.c
new file mode 100644
index 0000000..8436f0e
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsqrt7.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf4_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf2_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m1_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m2_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m4_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m8_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf4_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf2_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m1_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m2_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m4_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m8_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf4_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf2_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m1_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m2_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m4_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m8_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf4_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16mf2_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m1_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m2_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m4_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_v_bf16m8_mu(mask, maskedoff, op1, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsub.c
new file mode 100644
index 0000000..7dd2bb6
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfrsub.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnj.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnj.c
new file mode 100644
index 0000000..b39a0be
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnj.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjn.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjn.c
new file mode 100644
index 0000000..7542e78
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjn.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjx.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjx.c
new file mode 100644
index 0000000..104149e
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsgnjx.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1down.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1down.c
new file mode 100644
index 0000000..228dc1cd
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1down.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf4_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf2_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m1_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m2_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m4_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m8_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf4_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf2_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m1_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m2_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m4_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m8_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf4_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf2_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m1_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m2_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m4_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m8_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf4_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16mf2_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m1_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m2_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m4_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_vf_bf16m8_mu(mask, maskedoff, src, value, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1up.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1up.c
new file mode 100644
index 0000000..9e6ff2b
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfslide1up.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf4_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf2_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m1_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m2_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m4_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m8_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf4_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf2_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m1_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m2_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m4_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m8_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf4_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf2_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m1_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m2_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m4_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m8_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf4_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16mf2_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m1_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m2_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m4_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_vf_bf16m8_mu(mask, maskedoff, src, value, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsub.c
new file mode 100644
index 0000000..b6fd94e
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfsub.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m1_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m2_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m4_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m8_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m1_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m2_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m4_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m8_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m1_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m2_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m4_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m8_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16mf2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m1_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub_vv_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_vf_bf16m8_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwadd.c
new file mode 100644
index 0000000..4bee376
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwadd.c
@@ -0,0 +1,2007 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tu(vfloat32mf2_t vd, vfloat32mf2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2,
+                                                vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vfloat32mf2_t vs2,
+                                                vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                             vfloat32mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2,
+                                              vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vfloat32m1_t vs2,
+                                              vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                           vfloat32m1_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2,
+                                             vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vfloat32m2_t vs2,
+                                             vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                           vfloat32m2_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2,
+                                             vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vfloat32m4_t vs2,
+                                             vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                           vfloat32m4_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2,
+                                             vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vfloat32m8_t vs2,
+                                             vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                           vfloat32m8_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vfloat32mf2_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                              vfloat32mf2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vfloat32m1_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                            vfloat32m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vfloat32m2_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                            vfloat32m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vfloat32m4_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                            vfloat32m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vfloat32m8_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                            vfloat32m8_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                            vfloat32mf2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                          vfloat32m1_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                          vfloat32m2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                          vfloat32m4_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                          vfloat32m8_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vfloat32mf2_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vfloat32m1_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vfloat32m2_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vfloat32m4_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vfloat32m8_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   vbfloat16mf4_t vs1,
+                                                   size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, vs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, rs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vfloat32mf2_t vs2,
+                                                   vbfloat16mf4_t vs1,
+                                                   size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, vs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vfloat32mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 vbfloat16mf2_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vfloat32m1_t vs2,
+                                                 vbfloat16mf2_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_rm_tum(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vfloat32m1_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                             vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2,
+                                                vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_rm_tum(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vfloat32m2_t vs2,
+                                                vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              vfloat32m2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                             vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2,
+                                                vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_rm_tum(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vfloat32m4_t vs2,
+                                                vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              vfloat32m4_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                             vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2,
+                                                vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_rm_tum(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vfloat32m8_t vs2,
+                                                vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              vfloat32m8_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                             vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    vbfloat16mf4_t vs1,
+                                                    size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, vs1,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, rs1,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vfloat32mf2_t vs2,
+                                                    vbfloat16mf4_t vs1,
+                                                    size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, vs1,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vfloat32mf2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_rm_tumu(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  vbfloat16mf2_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, vs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, rs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vfloat32m1_t vs2,
+                                                  vbfloat16mf2_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, vs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vfloat32m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2,
+                                                 vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vfloat32m2_t vs2,
+                                                 vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               vfloat32m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2,
+                                                 vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vfloat32m4_t vs2,
+                                                 vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               vfloat32m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2,
+                                                 vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vfloat32m8_t vs2,
+                                                 vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               vfloat32m8_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vfloat32mf2_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32mf2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                             vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vfloat32m1_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16mf2_f32m1_rm_mu(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vfloat32m1_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m1_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                            vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m1_f32m2_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_bf16m1_f32m2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vfloat32m2_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m1_f32m2_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             vfloat32m2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                            vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m2_f32m4_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_bf16m2_f32m4_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vfloat32m4_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m2_f32m4_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             vfloat32m4_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m4_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                            vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_bf16m4_f32m8_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_bf16m4_f32m8_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vfloat32m8_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_bf16m4_f32m8_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             vfloat32m8_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_wf_bf16_f32m8_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                            vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwcvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwcvt.c
new file mode 100644
index 0000000..9151319
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwcvt.c
@@ -0,0 +1,765 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tu(vbfloat16mf4_t vd, vint8mf8_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tu(vbfloat16mf2_t vd, vint8mf4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tu(vbfloat16m1_t vd, vint8mf2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m1_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tu(vbfloat16m2_t vd, vint8m1_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tu(vbfloat16m4_t vd, vint8m2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tu(vbfloat16m8_t vd, vint8m4_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m8_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tu(vbfloat16mf4_t vd, vuint8mf8_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tu(vbfloat16mf2_t vd, vuint8mf4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tu(vbfloat16m1_t vd, vuint8mf2_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m1_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tu(vbfloat16m2_t vd, vuint8m1_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tu(vbfloat16m4_t vd, vuint8m2_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tu(vbfloat16m8_t vd, vuint8m4_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m8_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m1_f32m2_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m2_f32m4_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m4_f32m8_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd,
+                                             vint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd,
+                                             vint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd,
+                                           vint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m1_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd,
+                                           vint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd,
+                                           vint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tum(vbool2_t vm, vbfloat16m8_t vd,
+                                           vint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m8_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd,
+                                              vuint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd,
+                                              vuint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd,
+                                            vuint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m1_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd,
+                                            vuint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd,
+                                            vuint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tum(vbool2_t vm, vbfloat16m8_t vd,
+                                            vuint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m8_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m1_f32m2_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m2_f32m4_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m4_f32m8_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd,
+                                              vint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd,
+                                              vint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd,
+                                            vint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m1_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd,
+                                            vint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd,
+                                            vint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tumu(vbool2_t vm, vbfloat16m8_t vd,
+                                            vint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m8_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd,
+                                               vuint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd,
+                                               vuint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd,
+                                             vuint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m1_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd,
+                                             vuint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd,
+                                             vuint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tumu(vbool2_t vm, vbfloat16m8_t vd,
+                                             vuint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m8_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m1_f32m2_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m2_f32m4_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m4_f32m8_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd,
+                                            vint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd,
+                                            vint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16mf2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd,
+                                          vint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m1_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd,
+                                          vint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd,
+                                          vint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_mu(vbool2_t vm, vbfloat16m8_t vd,
+                                          vint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_x_v_bf16m8_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd,
+                                             vuint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd,
+                                             vuint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16mf2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd,
+                                           vuint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m1_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd,
+                                           vuint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd,
+                                           vuint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_mu(vbool2_t vm, vbfloat16m8_t vd,
+                                           vuint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_xu_v_bf16m8_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf4_f32mf2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16mf2_f32m1_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m1_f32m2_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m2_f32m4_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_f_v_bf16m4_f32m8_mu(vm, vd, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmacc.c
new file mode 100644
index 0000000..f67b100
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmacc.c
@@ -0,0 +1,1017 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                 __bf16 vs1, vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                               __bf16 vs1, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              __bf16 vs1, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              __bf16 vs1, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              __bf16 vs1, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                __bf16 vs1, vbfloat16mf4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                              __bf16 vs1, vbfloat16mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             __bf16 vs1, vbfloat16m1_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             __bf16 vs1, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             __bf16 vs1, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tum(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                  __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                 __bf16 vs1, vbfloat16m1_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                 __bf16 vs1, vbfloat16m2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                 __bf16 vs1, vbfloat16m4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                     vfloat32mf2_t vd,
+                                                     vbfloat16mf4_t vs1,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tumu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                   vfloat32m1_t vd,
+                                                   vbfloat16mf2_t vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                   vfloat32m1_t vd, __bf16 vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                  vbfloat16m1_t vs1,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                  __bf16 vs1, vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                  vbfloat16m2_t vs1,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                  __bf16 vs1, vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                  vbfloat16m4_t vs1,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                  __bf16 vs1, vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_vf_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmsac.c
new file mode 100644
index 0000000..6d78c74
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmsac.c
@@ -0,0 +1,1017 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                 __bf16 vs1, vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                               __bf16 vs1, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              __bf16 vs1, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              __bf16 vs1, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              __bf16 vs1, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                __bf16 vs1, vbfloat16mf4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                              __bf16 vs1, vbfloat16mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             __bf16 vs1, vbfloat16m1_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             __bf16 vs1, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             __bf16 vs1, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tum(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                  __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                 __bf16 vs1, vbfloat16m1_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                 __bf16 vs1, vbfloat16m2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                 __bf16 vs1, vbfloat16m4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                     vfloat32mf2_t vd,
+                                                     vbfloat16mf4_t vs1,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tumu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                   vfloat32m1_t vd,
+                                                   vbfloat16mf2_t vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                   vfloat32m1_t vd, __bf16 vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                  vbfloat16m1_t vs1,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                  __bf16 vs1, vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                  vbfloat16m2_t vs1,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                  __bf16 vs1, vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                  vbfloat16m4_t vs1,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                  __bf16 vs1, vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_vf_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2,
+                                               __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmul.c
new file mode 100644
index 0000000..9fcfe81
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwmul.c
@@ -0,0 +1,1015 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2,
+                                                vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2,
+                                              vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2,
+                                             vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2,
+                                             vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2,
+                                             vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   vbfloat16mf4_t vs1,
+                                                   size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, vs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, rs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 vbfloat16mf2_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2,
+                                                vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_rm_tum(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2,
+                                                vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_rm_tum(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2,
+                                                vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_rm_tum(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    vbfloat16mf4_t vs1,
+                                                    size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, vs1,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, rs1,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  vbfloat16mf2_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, vs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, rs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2,
+                                                 vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2,
+                                                 vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2,
+                                                 vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m1_f32m2_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_vf_bf16m1_f32m2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m2_f32m4_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_vf_bf16m2_f32m4_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_vv_bf16m4_f32m8_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_vf_bf16m4_f32m8_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmacc.c
new file mode 100644
index 0000000..73cc822
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmacc.c
@@ -0,0 +1,1034 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tu(vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tu(vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tu(vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tum(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tum(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 __bf16 vs1, vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                               __bf16 vs1, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                              __bf16 vs1, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                              __bf16 vs1, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                              __bf16 vs1, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                    __bf16 vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                     vfloat32mf2_t vd,
+                                                     vbfloat16mf4_t vs1,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tum(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm,
+                                                   vfloat32m1_t vd,
+                                                   vbfloat16mf2_t vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm,
+                                                   vfloat32m1_t vd, __bf16 vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                  vbfloat16m1_t vs1,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                  __bf16 vs1, vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                  vbfloat16m2_t vs1,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                  __bf16 vs1, vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                  vbfloat16m4_t vs1,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                  __bf16 vs1, vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                      vfloat32mf2_t vd,
+                                                      vbfloat16mf4_t vs1,
+                                                      vbfloat16mf4_t vs2,
+                                                      size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2,
+                                                    __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tumu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2,
+                                                    __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                    vfloat32m1_t vd,
+                                                    vbfloat16mf2_t vs1,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                    vfloat32m1_t vd, __bf16 vs1,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm,
+                                                   vfloat32m2_t vd,
+                                                   vbfloat16m1_t vs1,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm,
+                                                   vfloat32m2_t vd, __bf16 vs1,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                   vbfloat16m2_t vs1,
+                                                   vbfloat16m2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                   __bf16 vs1,
+                                                   vbfloat16m2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                   vbfloat16m4_t vs1,
+                                                   vbfloat16m4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                   __bf16 vs1,
+                                                   vbfloat16m4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_mu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                  __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                 __bf16 vs1, vbfloat16m1_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                 __bf16 vs1, vbfloat16m2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                 __bf16 vs1, vbfloat16m4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_vf_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmsac.c
new file mode 100644
index 0000000..6133230
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwnmsac.c
@@ -0,0 +1,1034 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tu(vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tu(vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tu(vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tum(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tum(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 __bf16 vs1, vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                               __bf16 vs1, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                              __bf16 vs1, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                              __bf16 vs1, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                              __bf16 vs1, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                    __bf16 vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                  vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                 vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                     vfloat32mf2_t vd,
+                                                     vbfloat16mf4_t vs1,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tum(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm,
+                                                   vfloat32m1_t vd,
+                                                   vbfloat16mf2_t vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm,
+                                                   vfloat32m1_t vd, __bf16 vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                  vbfloat16m1_t vs1,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                  __bf16 vs1, vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                  vbfloat16m2_t vs1,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                  __bf16 vs1, vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                  vbfloat16m4_t vs1,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                  __bf16 vs1, vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm_tum(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                      vfloat32mf2_t vd,
+                                                      vbfloat16mf4_t vs1,
+                                                      vbfloat16mf4_t vs2,
+                                                      size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2,
+                                                    __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tumu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs1, vs2,
+                                                    __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                    vfloat32m1_t vd,
+                                                    vbfloat16mf2_t vs1,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                    vfloat32m1_t vd, __bf16 vs1,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs1, vs2,
+                                                   __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm,
+                                                   vfloat32m2_t vd,
+                                                   vbfloat16m1_t vs1,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm,
+                                                   vfloat32m2_t vd, __bf16 vs1,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                   vbfloat16m2_t vs1,
+                                                   vbfloat16m2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                   __bf16 vs1,
+                                                   vbfloat16m2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                   vbfloat16m4_t vs1,
+                                                   vbfloat16m4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                   __bf16 vs1,
+                                                   vbfloat16m4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_mu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs1, vs2,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                  __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs1, vs2,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                 __bf16 vs1, vbfloat16m1_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m1_f32m2_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                 __bf16 vs1, vbfloat16m2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m2_f32m4_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                 __bf16 vs1, vbfloat16m4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_vf_bf16m4_f32m8_rm_mu(vm, vd, vs1, vs2,
+                                                __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwsub.c
new file mode 100644
index 0000000..9d9b0b0
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vfwsub.c
@@ -0,0 +1,2007 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tu(vfloat32mf2_t vd, vfloat32mf2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2,
+                                                vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vfloat32mf2_t vs2,
+                                                vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                             vfloat32mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2,
+                                              vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vfloat32m1_t vs2,
+                                              vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                           vfloat32m1_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2,
+                                             vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vfloat32m2_t vs2,
+                                             vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                           vfloat32m2_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2,
+                                             vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vfloat32m4_t vs2,
+                                             vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                           vfloat32m4_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2,
+                                             vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vfloat32m8_t vs2,
+                                             vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                           vfloat32m8_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vfloat32mf2_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                              vfloat32mf2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vfloat32m1_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                            vfloat32m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vfloat32m2_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                            vfloat32m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vfloat32m4_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                            vfloat32m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vfloat32m8_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                            vfloat32m8_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                            vfloat32mf2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                          vfloat32m1_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                          vfloat32m2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                          vfloat32m4_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                          vfloat32m8_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vfloat32mf2_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                                vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vfloat32m1_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                               vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vfloat32m2_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vfloat32m4_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vfloat32m8_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_rm_tu(vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_rm_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   vbfloat16mf4_t vs1,
+                                                   size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, vs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, rs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vfloat32mf2_t vs2,
+                                                   vbfloat16mf4_t vs1,
+                                                   size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm_tum(vm, vd, vs2, vs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vfloat32mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 vbfloat16mf2_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_rm_tum(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_rm_tum(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vfloat32m1_t vs2,
+                                                 vbfloat16mf2_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_rm_tum(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vfloat32m1_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                             vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2,
+                                                vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_rm_tum(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vfloat32m2_t vs2,
+                                                vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              vfloat32m2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                             vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2,
+                                                vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_rm_tum(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vfloat32m4_t vs2,
+                                                vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              vfloat32m4_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                             vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2,
+                                                vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_rm_tum(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vfloat32m8_t vs2,
+                                                vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_rm_tum(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              vfloat32m8_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_rm_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                             vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    vbfloat16mf4_t vs1,
+                                                    size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, vs1,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, rs1,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vfloat32mf2_t vs2,
+                                                    vbfloat16mf4_t vs1,
+                                                    size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm_tumu(vm, vd, vs2, vs1,
+                                                  __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vfloat32mf2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_rm_tumu(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  vbfloat16mf2_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, vs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, rs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vfloat32m1_t vs2,
+                                                  vbfloat16mf2_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_rm_tumu(vm, vd, vs2, vs1,
+                                                 __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vfloat32m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2,
+                                                 vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_rm_tumu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vfloat32m2_t vs2,
+                                                 vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               vfloat32m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2,
+                                                 vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_rm_tumu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vfloat32m4_t vs2,
+                                                 vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               vfloat32m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2,
+                                                 vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_rm_tumu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vfloat32m8_t vs2,
+                                                 vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_rm_tumu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               vfloat32m8_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_rm_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, rs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vfloat32mf2_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf4_f32mf2_rm_mu(vm, vd, vs2, vs1,
+                                                __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32mf2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                             vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16mf2_f32m1_rm_mu(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_bf16mf2_f32m1_rm_mu(vm, vd, vs2, rs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vfloat32m1_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16mf2_f32m1_rm_mu(vm, vd, vs2, vs1,
+                                               __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vfloat32m1_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m1_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                            vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m1_f32m2_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_bf16m1_f32m2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vfloat32m2_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m1_f32m2_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             vfloat32m2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m2_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                            vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m2_f32m4_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_bf16m2_f32m4_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vfloat32m4_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m2_f32m4_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             vfloat32m4_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m4_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                            vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_bf16m4_f32m8_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_bf16m4_f32m8_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vfloat32m8_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_bf16m4_f32m8_rm_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE,
+                                              vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             vfloat32m8_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_wf_bf16_f32m8_rm_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE,
+                                            vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfeq.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfeq.c
new file mode 100644
index 0000000..b96aae5
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfeq.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfeq_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfge.c
new file mode 100644
index 0000000..47d0427
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfge.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfge_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfgt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfgt.c
new file mode 100644
index 0000000..0a0ead2
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfgt.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfgt_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfle.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfle.c
new file mode 100644
index 0000000..27ddefe
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfle.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfle_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmflt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmflt.c
new file mode 100644
index 0000000..d5f4f77
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmflt.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmflt_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfne.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfne.c
new file mode 100644
index 0000000..c2df947
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/non-overloaded/vmfne.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16mf4_b64_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16mf2_b32_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m1_b16_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m2_b8_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m4_b4_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfne_vv_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_vf_bf16m8_b2_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfadd.c
new file mode 100644
index 0000000..2bd3b39
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfadd.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfadd_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfadd_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfadd_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfadd_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfadd_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfadd_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfadd_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfclass.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfclass.c
new file mode 100644
index 0000000..e2a993a
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfclass.c
@@ -0,0 +1,272 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tu(vuint16mf4_t vd,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfclass_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tu(vuint16mf2_t vd,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfclass_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1_tu(vuint16m1_t vd, vbfloat16m1_t vs2,
+                                           size_t vl) {
+  return __riscv_vfclass_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2_tu(vuint16m2_t vd, vbfloat16m2_t vs2,
+                                           size_t vl) {
+  return __riscv_vfclass_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4_tu(vuint16m4_t vd, vbfloat16m4_t vs2,
+                                           size_t vl) {
+  return __riscv_vfclass_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8_tu(vuint16m8_t vd, vbfloat16m8_t vs2,
+                                           size_t vl) {
+  return __riscv_vfclass_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tum(vbool64_t vm, vuint16mf4_t vd,
+                                               vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfclass_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tum(vbool32_t vm, vuint16mf2_t vd,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfclass_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1_tum(vbool16_t vm, vuint16m1_t vd,
+                                            vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfclass_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2_tum(vbool8_t vm, vuint16m2_t vd,
+                                            vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfclass_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4_tum(vbool4_t vm, vuint16m4_t vd,
+                                            vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfclass_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8_tum(vbool2_t vm, vuint16m8_t vd,
+                                            vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfclass_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_tumu(vbool64_t vm, vuint16mf4_t vd,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfclass_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_tumu(vbool32_t vm, vuint16mf2_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfclass_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1_tumu(vbool16_t vm, vuint16m1_t vd,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfclass_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2_tumu(vbool8_t vm, vuint16m2_t vd,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfclass_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4_tumu(vbool4_t vm, vuint16m4_t vd,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfclass_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8_tumu(vbool2_t vm, vuint16m8_t vd,
+                                             vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfclass_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i16> @test_vfclass_v_bf16mf4_u16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i16> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1bf16.i64(<vscale x 1 x i16> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i16> [[TMP0]]
+//
+vuint16mf4_t test_vfclass_v_bf16mf4_u16mf4_mu(vbool64_t vm, vuint16mf4_t vd,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfclass_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i16> @test_vfclass_v_bf16mf2_u16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i16> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2bf16.i64(<vscale x 2 x i16> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i16> [[TMP0]]
+//
+vuint16mf2_t test_vfclass_v_bf16mf2_u16mf2_mu(vbool32_t vm, vuint16mf2_t vd,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfclass_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i16> @test_vfclass_v_bf16m1_u16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i16> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4bf16.i64(<vscale x 4 x i16> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i16> [[TMP0]]
+//
+vuint16m1_t test_vfclass_v_bf16m1_u16m1_mu(vbool16_t vm, vuint16m1_t vd,
+                                           vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfclass_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i16> @test_vfclass_v_bf16m2_u16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i16> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8bf16.i64(<vscale x 8 x i16> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
+//
+vuint16m2_t test_vfclass_v_bf16m2_u16m2_mu(vbool8_t vm, vuint16m2_t vd,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfclass_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i16> @test_vfclass_v_bf16m4_u16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i16> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16bf16.i64(<vscale x 16 x i16> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i16> [[TMP0]]
+//
+vuint16m4_t test_vfclass_v_bf16m4_u16m4_mu(vbool4_t vm, vuint16m4_t vd,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfclass_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i16> @test_vfclass_v_bf16m8_u16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i16> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32bf16.i64(<vscale x 32 x i16> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i16> [[TMP0]]
+//
+vuint16m8_t test_vfclass_v_bf16m8_u16m8_mu(vbool2_t vm, vuint16m8_t vd,
+                                           vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfclass_mu(vm, vd, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmacc.c
new file mode 100644
index 0000000..eb74271
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmacc.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmacc_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmacc_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmacc_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmacc_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmacc_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmacc_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmacc_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmacc_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmacc_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmacc_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmacc_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmacc_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmadd.c
new file mode 100644
index 0000000..68d490d
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmadd.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmadd_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmadd_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmadd_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmadd_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmadd_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmadd_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmax.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmax.c
new file mode 100644
index 0000000..5f682e8
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmax.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmax_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmax_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmax_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmax_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmax_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmax_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmax_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmax_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmax_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmax_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmax_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmax_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmax_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmerge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmerge.c
new file mode 100644
index 0000000..9593ad5
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmerge.c
@@ -0,0 +1,69 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmerge_vfm_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmerge_vfm_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, vbool64_t mask, size_t vl) {
+  return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmerge_vfm_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 2 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmerge_vfm_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, vbool32_t mask, size_t vl) {
+  return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmerge_vfm_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 4 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmerge_vfm_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, vbool16_t mask, size_t vl) {
+  return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmerge_vfm_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmerge_vfm_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, vbool8_t mask, size_t vl) {
+  return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmerge_vfm_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 16 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmerge_vfm_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, vbool4_t mask, size_t vl) {
+  return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmerge_vfm_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmerge_vfm_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, vbool2_t mask, size_t vl) {
+  return __riscv_vfmerge_tu(maskedoff, op1, op2, mask, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmin.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmin.c
new file mode 100644
index 0000000..f3ef3c3
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmin.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmin_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmin_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmin_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmin_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmin_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmin_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmin_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmin_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmin_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmin_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmin_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmin_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmin_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsac.c
new file mode 100644
index 0000000..0587c57
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsac.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsac_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsac_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsac_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsac_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsac_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsac_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsac_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsac_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsac_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsac_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsac_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsac_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsub.c
new file mode 100644
index 0000000..2ad26f8
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmsub.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmsub_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmsub_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmsub_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmsub_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmsub_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmsub_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmul.c
new file mode 100644
index 0000000..d1e726a9
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmul.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmul_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmul_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmul_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmul_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmul_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmul_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmul_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmul_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmul_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmul_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmul_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmul_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfmul_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmv.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmv.c
new file mode 100644
index 0000000..9fd1ffc
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfmv.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_v_f_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.v.f.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmv_v_f_bf16mf4_tu(vbfloat16mf4_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_v_f_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.v.f.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmv_v_f_bf16mf2_tu(vbfloat16mf2_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_v_f_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.v.f.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmv_v_f_bf16m1_tu(vbfloat16m1_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_v_f_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.v.f.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmv_v_f_bf16m2_tu(vbfloat16m2_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_v_f_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.v.f.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmv_v_f_bf16m4_tu(vbfloat16m4_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_v_f_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.v.f.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmv_v_f_bf16m8_tu(vbfloat16m8_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_v_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfmv_s_f_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.s.f.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfmv_s_f_bf16mf4_tu(vbfloat16mf4_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfmv_s_f_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.s.f.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfmv_s_f_bf16mf2_tu(vbfloat16mf2_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfmv_s_f_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.s.f.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfmv_s_f_bf16m1_tu(vbfloat16m1_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfmv_s_f_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.s.f.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfmv_s_f_bf16m2_tu(vbfloat16m2_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfmv_s_f_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.s.f.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfmv_s_f_bf16m4_tu(vbfloat16m4_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_tu(maskedoff, src, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfmv_s_f_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], bfloat noundef [[SRC:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.s.f.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], bfloat [[SRC]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfmv_s_f_bf16m8_tu(vbfloat16m8_t maskedoff, __bf16 src, size_t vl) {
+  return __riscv_vfmv_s_tu(maskedoff, src, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt.c
new file mode 100644
index 0000000..c6cd0a5
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt.c
@@ -0,0 +1,1539 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tu(vint8mf8_t vd, vbfloat16mf4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tu(vint8mf4_t vd, vbfloat16mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tu(vint8mf2_t vd, vbfloat16m1_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tu(vint8m1_t vd, vbfloat16m2_t vs2,
+                                           size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tu(vint8m2_t vd, vbfloat16m4_t vs2,
+                                           size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tu(vint8m4_t vd, vbfloat16m8_t vs2,
+                                           size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tu(vuint8mf8_t vd,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tu(vuint8mf4_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tu(vuint8mf2_t vd,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tu(vuint8m1_t vd, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tu(vuint8m2_t vd, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tu(vuint8m4_t vd, vbfloat16m8_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tu(vbfloat16mf4_t vd,
+                                            vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tu(vbfloat16mf2_t vd, vfloat32m1_t vs2,
+                                            size_t vl) {
+  return __riscv_vfncvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tu(vbfloat16m1_t vd, vfloat32m2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tu(vbfloat16m2_t vd, vfloat32m4_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tu(vbfloat16m4_t vd, vfloat32m8_t vs2,
+                                          size_t vl) {
+  return __riscv_vfncvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tum(vbool64_t vm, vint8mf8_t vd,
+                                               vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tum(vbool32_t vm, vint8mf4_t vd,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tum(vbool16_t vm, vint8mf2_t vd,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tum(vbool8_t vm, vint8m1_t vd,
+                                            vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tum(vbool4_t vm, vint8m2_t vd,
+                                            vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tum(vbool2_t vm, vint8m4_t vd,
+                                            vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tum(vbool64_t vm, vuint8mf8_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tum(vbool32_t vm, vuint8mf4_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tum(vbool16_t vm, vuint8mf2_t vd,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tum(vbool8_t vm, vuint8m1_t vd,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tum(vbool4_t vm, vuint8m2_t vd,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tum(vbool2_t vm, vuint8m4_t vd,
+                                              vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd,
+                                             vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd,
+                                             vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd,
+                                           vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd,
+                                           vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd,
+                                           vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_tumu(vbool64_t vm, vint8mf8_t vd,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_tumu(vbool32_t vm, vint8mf4_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_tumu(vbool16_t vm, vint8mf2_t vd,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_tumu(vbool8_t vm, vint8m1_t vd,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_tumu(vbool4_t vm, vint8m2_t vd,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_tumu(vbool2_t vm, vint8m4_t vd,
+                                             vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_tumu(vbool64_t vm, vuint8mf8_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_tumu(vbool32_t vm, vuint8mf4_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_tumu(vbool16_t vm, vuint8mf2_t vd,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_tumu(vbool8_t vm, vuint8m1_t vd,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_tumu(vbool4_t vm, vuint8m2_t vd,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_tumu(vbool2_t vm, vuint8m4_t vd,
+                                               vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd,
+                                              vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd,
+                                              vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd,
+                                            vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd,
+                                            vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd,
+                                            vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_mu(vbool64_t vm, vint8mf8_t vd,
+                                              vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_mu(vbool32_t vm, vint8mf4_t vd,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_mu(vbool16_t vm, vint8mf2_t vd,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_mu(vbool8_t vm, vint8m1_t vd,
+                                           vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_mu(vbool4_t vm, vint8m2_t vd,
+                                           vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_mu(vbool2_t vm, vint8m4_t vd,
+                                           vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_mu(vbool64_t vm, vuint8mf8_t vd,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_mu(vbool32_t vm, vuint8mf4_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_mu(vbool16_t vm, vuint8mf2_t vd,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_mu(vbool8_t vm, vuint8m1_t vd,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_mu(vbool4_t vm, vuint8m2_t vd,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_mu(vbool2_t vm, vuint8m4_t vd,
+                                             vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd,
+                                            vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd,
+                                            vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd,
+                                          vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd,
+                                          vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd,
+                                          vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tu(vint8mf8_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tu(vint8mf4_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tu(vint8mf2_t vd,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tu(vint8m1_t vd, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tu(vint8m2_t vd, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tu(vint8m4_t vd, vbfloat16m8_t vs2,
+                                              size_t vl) {
+  return __riscv_vfncvt_x_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tu(vuint8mf8_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tu(vuint8mf4_t vd,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tu(vuint8mf2_t vd,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tu(vuint8m1_t vd,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tu(vuint8m2_t vd,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tu(vuint8m4_t vd,
+                                                vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tu(vbfloat16mf4_t vd,
+                                               vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tu(vbfloat16mf2_t vd,
+                                               vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tu(vbfloat16m1_t vd, vfloat32m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_f_bf16_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tu(vbfloat16m2_t vd, vfloat32m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_f_bf16_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tu(vbfloat16m4_t vd, vfloat32m8_t vs2,
+                                             size_t vl) {
+  return __riscv_vfncvt_f_bf16_tu(vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tum(vbool64_t vm, vint8mf8_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tum(vbool32_t vm, vint8mf4_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tum(vbool16_t vm, vint8mf2_t vd,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tum(vbool8_t vm, vint8m1_t vd,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tum(vbool4_t vm, vint8m2_t vd,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tum(vbool2_t vm, vint8m4_t vd,
+                                               vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tum(vbool64_t vm,
+                                                    vuint8mf8_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tum(vbool32_t vm,
+                                                    vuint8mf4_t vd,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tum(vbool16_t vm, vuint8mf2_t vd,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tum(vbool8_t vm, vuint8m1_t vd,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tum(vbool4_t vm, vuint8m2_t vd,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tum(vbool2_t vm, vuint8m4_t vd,
+                                                 vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tum(vbool64_t vm, vbfloat16mf4_t vd,
+                                                vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tum(vbool32_t vm, vbfloat16mf2_t vd,
+                                                vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tum(vbool16_t vm, vbfloat16m1_t vd,
+                                              vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tum(vbool8_t vm, vbfloat16m2_t vd,
+                                              vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tum(vbool4_t vm, vbfloat16m4_t vd,
+                                              vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tum(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_tumu(vbool64_t vm, vint8mf8_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_tumu(vbool32_t vm, vint8mf4_t vd,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_tumu(vbool16_t vm, vint8mf2_t vd,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_tumu(vbool8_t vm, vint8m1_t vd,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_tumu(vbool4_t vm, vint8m2_t vd,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_tumu(vbool2_t vm, vint8m4_t vd,
+                                                vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_tumu(vbool64_t vm,
+                                                     vuint8mf8_t vd,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_tumu(vbool32_t vm,
+                                                     vuint8mf4_t vd,
+                                                     vbfloat16mf2_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_tumu(vbool16_t vm,
+                                                    vuint8mf2_t vd,
+                                                    vbfloat16m1_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_tumu(vbool8_t vm, vuint8m1_t vd,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_tumu(vbool4_t vm, vuint8m2_t vd,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_tumu(vbool2_t vm, vuint8m4_t vd,
+                                                  vbfloat16m8_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_tumu(vbool64_t vm,
+                                                 vbfloat16mf4_t vd,
+                                                 vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_tumu(vbool32_t vm,
+                                                 vbfloat16mf2_t vd,
+                                                 vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_tumu(vbool16_t vm, vbfloat16m1_t vd,
+                                               vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_tumu(vbool8_t vm, vbfloat16m2_t vd,
+                                               vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_tumu(vbool4_t vm, vbfloat16m4_t vd,
+                                               vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_tumu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_x_f_w_bf16mf4_i8mf8_rm_mu(vbool64_t vm, vint8mf8_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_x_f_w_bf16mf2_i8mf4_rm_mu(vbool32_t vm, vint8mf4_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_x_f_w_bf16m1_i8mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_x_f_w_bf16m1_i8mf2_rm_mu(vbool16_t vm, vint8mf2_t vd,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_x_f_w_bf16m2_i8m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_x_f_w_bf16m2_i8m1_rm_mu(vbool8_t vm, vint8m1_t vd,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_x_f_w_bf16m4_i8m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_x_f_w_bf16m4_i8m2_rm_mu(vbool4_t vm, vint8m2_t vd,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_x_f_w_bf16m8_i8m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_x_f_w_bf16m8_i8m4_rm_mu(vbool2_t vm, vint8m4_t vd,
+                                              vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_x_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_xu_f_w_bf16mf4_u8mf8_rm_mu(vbool64_t vm, vuint8mf8_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_xu_f_w_bf16mf2_u8mf4_rm_mu(vbool32_t vm, vuint8mf4_t vd,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_xu_f_w_bf16m1_u8mf2_rm_mu(vbool16_t vm, vuint8mf2_t vd,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_xu_f_w_bf16m2_u8m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_xu_f_w_bf16m2_u8m1_rm_mu(vbool8_t vm, vuint8m1_t vd,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_xu_f_w_bf16m4_u8m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_xu_f_w_bf16m4_u8m2_rm_mu(vbool4_t vm, vuint8m2_t vd,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_xu_f_w_bf16m8_u8m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_xu_f_w_bf16m8_u8m4_rm_mu(vbool2_t vm, vuint8m4_t vd,
+                                                vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_xu_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_f_f_w_bf16mf4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_f_f_w_bf16mf4_rm_mu(vbool64_t vm, vbfloat16mf4_t vd,
+                                               vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_f_f_w_bf16mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_f_f_w_bf16mf2_rm_mu(vbool32_t vm, vbfloat16mf2_t vd,
+                                               vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_f_f_w_bf16m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_f_f_w_bf16m1_rm_mu(vbool16_t vm, vbfloat16m1_t vd,
+                                             vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_f_f_w_bf16m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_f_f_w_bf16m2_rm_mu(vbool8_t vm, vbfloat16m2_t vd,
+                                             vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_f_f_w_bf16m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_f_f_w_bf16m4_rm_mu(vbool4_t vm, vbfloat16m4_t vd,
+                                             vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_f_bf16_mu(vm, vd, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rod.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rod.c
new file mode 100644
index 0000000..0745633
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rod.c
@@ -0,0 +1,233 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tu(vbfloat16mf4_t vd,
+                                                vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tu(vbfloat16mf2_t vd,
+                                                vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tu(vbfloat16m1_t vd,
+                                              vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tu(vbfloat16m2_t vd,
+                                              vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tu(vbfloat16m4_t vd,
+                                              vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tum(vbool64_t vm,
+                                                 vbfloat16mf4_t vd,
+                                                 vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tum(vbool32_t vm,
+                                                 vbfloat16mf2_t vd,
+                                                 vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd,
+                                               vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd,
+                                               vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd,
+                                               vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_tumu(vbool64_t vm,
+                                                  vbfloat16mf4_t vd,
+                                                  vfloat32mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_tumu(vbool32_t vm,
+                                                  vbfloat16mf2_t vd,
+                                                  vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd,
+                                                vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd,
+                                                vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd,
+                                                vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfncvt_rod_f_f_w_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd,
+                                                vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfncvt_rod_f_f_w_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfncvt_rod_f_f_w_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd,
+                                                vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfncvt_rod_f_f_w_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfncvt_rod_f_f_w_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd,
+                                              vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfncvt_rod_f_f_w_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfncvt_rod_f_f_w_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd,
+                                              vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfncvt_rod_f_f_w_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfncvt_rod_f_f_w_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd,
+                                              vfloat32m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rod_f_bf16_mu(vm, vd, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rtz.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rtz.c
new file mode 100644
index 0000000..b906c5f
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfncvt_rtz.c
@@ -0,0 +1,572 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tu(vint8mf8_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tu(vint8mf4_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tu(vint8mf2_t vd,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tu(vint8m1_t vd, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tu(vint8m2_t vd, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tu(vint8m4_t vd, vbfloat16m8_t vs2,
+                                               size_t vl) {
+  return __riscv_vfncvt_rtz_x_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tu(
+// CHECK-RV64-SAME: <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tu(vuint8mf8_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tu(
+// CHECK-RV64-SAME: <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tu(vuint8mf4_t vd,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tu(vuint8mf2_t vd,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tu(
+// CHECK-RV64-SAME: <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tu(vuint8m1_t vd,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tu(
+// CHECK-RV64-SAME: <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tu(vuint8m2_t vd,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tu(
+// CHECK-RV64-SAME: <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tu(vuint8m4_t vd,
+                                                 vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tum(vbool64_t vm, vint8mf8_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tum(vbool32_t vm, vint8mf4_t vd,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tum(vbool16_t vm, vint8mf2_t vd,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tum(vbool8_t vm, vint8m1_t vd,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tum(vbool4_t vm, vint8m2_t vd,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tum(vbool2_t vm, vint8m4_t vd,
+                                                vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tum(vbool64_t vm,
+                                                     vuint8mf8_t vd,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tum(vbool32_t vm,
+                                                     vuint8mf4_t vd,
+                                                     vbfloat16mf2_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tum(vbool16_t vm,
+                                                    vuint8mf2_t vd,
+                                                    vbfloat16m1_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tum(vbool8_t vm, vuint8m1_t vd,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tum(vbool4_t vm, vuint8m2_t vd,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tum(vbool2_t vm, vuint8m4_t vd,
+                                                  vbfloat16m8_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_tumu(vbool64_t vm, vint8mf8_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_tumu(vbool32_t vm, vint8mf4_t vd,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_tumu(vbool16_t vm, vint8mf2_t vd,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_tumu(vbool8_t vm, vint8m1_t vd,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_tumu(vbool4_t vm, vint8m2_t vd,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_tumu(vbool2_t vm, vint8m4_t vd,
+                                                 vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_tumu(vbool64_t vm,
+                                                      vuint8mf8_t vd,
+                                                      vbfloat16mf4_t vs2,
+                                                      size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_tumu(vbool32_t vm,
+                                                      vuint8mf4_t vd,
+                                                      vbfloat16mf2_t vs2,
+                                                      size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_tumu(vbool16_t vm,
+                                                     vuint8mf2_t vd,
+                                                     vbfloat16m1_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_tumu(vbool8_t vm, vuint8m1_t vd,
+                                                   vbfloat16m2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_tumu(vbool4_t vm, vuint8m2_t vd,
+                                                   vbfloat16m4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_tumu(vbool2_t vm, vuint8m4_t vd,
+                                                   vbfloat16m8_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vint8mf8_t test_vfncvt_rtz_x_f_w_bf16mf4_i8mf8_mu(vbool64_t vm, vint8mf8_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vint8mf4_t test_vfncvt_rtz_x_f_w_bf16mf2_i8mf4_mu(vbool32_t vm, vint8mf4_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vint8mf2_t test_vfncvt_rtz_x_f_w_bf16m1_i8mf2_mu(vbool16_t vm, vint8mf2_t vd,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_x_f_w_bf16m2_i8m1_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vint8m1_t test_vfncvt_rtz_x_f_w_bf16m2_i8m1_mu(vbool8_t vm, vint8m1_t vd,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_x_f_w_bf16m4_i8m2_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vint8m2_t test_vfncvt_rtz_x_f_w_bf16m4_i8m2_mu(vbool4_t vm, vint8m2_t vd,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_x_f_w_bf16m8_i8m4_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vint8m4_t test_vfncvt_rtz_x_f_w_bf16m8_i8m4_mu(vbool2_t vm, vint8m4_t vd,
+                                               vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_x_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x i8> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16.i64(<vscale x 1 x i8> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i8> [[TMP0]]
+//
+vuint8mf8_t test_vfncvt_rtz_xu_f_w_bf16mf4_u8mf8_mu(vbool64_t vm,
+                                                    vuint8mf8_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i8> @test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x i8> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16.i64(<vscale x 2 x i8> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i8> [[TMP0]]
+//
+vuint8mf4_t test_vfncvt_rtz_xu_f_w_bf16mf2_u8mf4_mu(vbool32_t vm,
+                                                    vuint8mf4_t vd,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i8> @test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x i8> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16.i64(<vscale x 4 x i8> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i8> [[TMP0]]
+//
+vuint8mf2_t test_vfncvt_rtz_xu_f_w_bf16m1_u8mf2_mu(vbool16_t vm, vuint8mf2_t vd,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i8> @test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x i8> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16.i64(<vscale x 8 x i8> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i8> [[TMP0]]
+//
+vuint8m1_t test_vfncvt_rtz_xu_f_w_bf16m2_u8m1_mu(vbool8_t vm, vuint8m1_t vd,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i8> @test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x i8> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16.i64(<vscale x 16 x i8> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
+//
+vuint8m2_t test_vfncvt_rtz_xu_f_w_bf16m4_u8m2_mu(vbool4_t vm, vuint8m2_t vd,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i8> @test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x i8> [[VD:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16.i64(<vscale x 32 x i8> [[VD]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i8> [[TMP0]]
+//
+vuint8m4_t test_vfncvt_rtz_xu_f_w_bf16m8_u8m4_mu(vbool2_t vm, vuint8m4_t vd,
+                                                 vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfncvt_rtz_xu_mu(vm, vd, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmacc.c
new file mode 100644
index 0000000..cc487b4
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmacc.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmacc_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmacc_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmacc_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmacc_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmacc_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmacc_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmacc_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmacc_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmacc_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmacc_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmacc_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmacc.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmacc_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmacc_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmadd.c
new file mode 100644
index 0000000..f9c348b
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmadd.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmadd_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmadd_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmadd_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmadd_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmadd_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmadd_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmadd_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmadd_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmadd_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmadd_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmadd_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmadd.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmadd_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmadd_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsac.c
new file mode 100644
index 0000000..83d35e8
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsac.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsac_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsac_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsac_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsac_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsac_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsac_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsac_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsac_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsac_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsac_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsac_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsac.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsac_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsac_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsub.c
new file mode 100644
index 0000000..f5282a1
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfnmsub.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tu(vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tu(vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tu(vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tu(vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1_tu(vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1_tu(vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2_tu(vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2_tu(vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4_tu(vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4_tu(vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8_tu(vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8_tu(vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tu(vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tum(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_tumu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, vbfloat16mf4_t vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfnmsub_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfnmsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t vd, __bf16 rs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, vbfloat16mf2_t vs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfnmsub_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfnmsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t vd, __bf16 rs1, vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, vbfloat16m1_t vs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfnmsub_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfnmsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t vd, __bf16 rs1, vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, vbfloat16m2_t vs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfnmsub_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfnmsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t vd, __bf16 rs1, vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, vbfloat16m4_t vs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfnmsub_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfnmsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t vd, __bf16 rs1, vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x bfloat> [[VS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x bfloat> [[VS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, vbfloat16m8_t vs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfnmsub_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], bfloat noundef [[RS1:%.*]], <vscale x 32 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfnmsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[VD]], bfloat [[RS1]], <vscale x 32 x bfloat> [[VS2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfnmsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t vd, __bf16 rs1, vbfloat16m8_t vs2, size_t vl) {
+  return __riscv_vfnmsub_mu(mask, vd, rs1, vs2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrec7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrec7.c
new file mode 100644
index 0000000..f8e5a33
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrec7.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrec7_v_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrec7_v_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrec7_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrec7_v_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrec7_v_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrec7_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrec7_v_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrec7_v_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrec7_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrec7_v_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrec7_v_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrec7_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrec7_v_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrec7_v_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrec7_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrec7_v_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrec7_v_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrec7_mu(mask, maskedoff, op1, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsqrt7.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsqrt7.c
new file mode 100644
index 0000000..7c6c926
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsqrt7.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tu(maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tum(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_tumu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsqrt7_v_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsqrt7_v_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsqrt7_v_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsqrt7_v_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsqrt7_v_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsqrt7_v_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsqrt7_v_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsqrt7_v_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsqrt7_v_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsqrt7_v_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsqrt7_v_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsqrt7_v_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, size_t vl) {
+  return __riscv_vfrsqrt7_mu(mask, maskedoff, op1, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsub.c
new file mode 100644
index 0000000..c09caeb
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfrsub.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfrsub_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfrsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfrsub_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfrsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfrsub_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfrsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfrsub_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfrsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfrsub_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfrsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfrsub_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfrsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfrsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnj.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnj.c
new file mode 100644
index 0000000..c1f69932
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnj.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnj_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnj_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnj_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnj_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnj_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnj_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnj_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnj_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnj_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnj_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnj_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnj_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnj_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjn.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjn.c
new file mode 100644
index 0000000..1b799d8
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjn.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjn_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjn_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjn_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjn_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjn_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjn_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjn_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjn_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjn_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjn_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjn_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjn_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjn_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjx.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjx.c
new file mode 100644
index 0000000..9c5f2af
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsgnjx.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsgnjx_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsgnjx_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsgnjx_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsgnjx_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsgnjx_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsgnjx_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsgnjx_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsgnjx_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsgnjx_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsgnjx_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsgnjx_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsgnjx_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsgnjx_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1down.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1down.c
new file mode 100644
index 0000000..691302e
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1down.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1down_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1down_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1down_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1down_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1down_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1down_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1down_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1down_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1down_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1down_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1down_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1down_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1down_mu(mask, maskedoff, src, value, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1up.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1up.c
new file mode 100644
index 0000000..1238d22
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfslide1up.c
@@ -0,0 +1,249 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tu(maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tum(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_tumu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfslide1up_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfslide1up_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfslide1up_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfslide1up_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfslide1up_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfslide1up_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfslide1up_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfslide1up_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfslide1up_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfslide1up_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfslide1up_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[SRC:%.*]], bfloat noundef [[VALUE:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[SRC]], bfloat [[VALUE]], <vscale x 32 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfslide1up_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t src, __bf16 value, size_t vl) {
+  return __riscv_vfslide1up_mu(mask, maskedoff, src, value, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsub.c
new file mode 100644
index 0000000..ea4f8f0
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfsub.c
@@ -0,0 +1,489 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4_tu(vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2_tu(vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1_tu(vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2_tu(vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4_tu(vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8_tu(vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tu(maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4_tum(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2_tum(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1_tum(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2_tum(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4_tum(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8_tum(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tum(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4_tumu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2_tumu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1_tumu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2_tumu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4_tumu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8_tumu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_tumu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vv_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vv_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfsub_vf_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x bfloat> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16.i64(<vscale x 1 x bfloat> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfsub_vf_bf16mf4_mu(vbool64_t mask, vbfloat16mf4_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vv_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vv_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfsub_vf_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x bfloat> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16.i64(<vscale x 2 x bfloat> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfsub_vf_bf16mf2_mu(vbool32_t mask, vbfloat16mf2_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vv_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vv_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfsub_vf_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x bfloat> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16.i64(<vscale x 4 x bfloat> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfsub_vf_bf16m1_mu(vbool16_t mask, vbfloat16m1_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vv_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vv_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfsub_vf_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x bfloat> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16.i64(<vscale x 8 x bfloat> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfsub_vf_bf16m2_mu(vbool8_t mask, vbfloat16m2_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vv_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vv_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfsub_vf_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x bfloat> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16.i64(<vscale x 16 x bfloat> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfsub_vf_bf16m4_mu(vbool4_t mask, vbfloat16m4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vv_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vv_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfsub_vf_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x bfloat> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16.i64(<vscale x 32 x bfloat> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfsub_vf_bf16m8_mu(vbool2_t mask, vbfloat16m8_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vfsub_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwadd.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwadd.c
new file mode 100644
index 0000000..e5b7b8d
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwadd.c
@@ -0,0 +1,1932 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tu(vfloat32mf2_t vd, vfloat32mf2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2,
+                                                vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vfloat32mf2_t vs2,
+                                                vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                             vfloat32mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2,
+                                              vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vfloat32m1_t vs2,
+                                              vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                           vfloat32m1_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2,
+                                             vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vfloat32m2_t vs2,
+                                             vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                           vfloat32m2_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2,
+                                             vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vfloat32m4_t vs2,
+                                             vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                           vfloat32m4_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2,
+                                             vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vfloat32m8_t vs2,
+                                             vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                           vfloat32m8_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vfloat32mf2_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                              vfloat32mf2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vfloat32m1_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                            vfloat32m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vfloat32m2_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                            vfloat32m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vfloat32m4_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                            vfloat32m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vfloat32m8_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                            vfloat32m8_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                            vfloat32mf2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                          vfloat32m1_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                          vfloat32m2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                          vfloat32m4_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                          vfloat32m8_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vfloat32mf2_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vfloat32m1_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vfloat32m2_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vfloat32m4_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vfloat32m8_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   vbfloat16mf4_t vs1,
+                                                   size_t vl) {
+  return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vfloat32mf2_t vs2,
+                                                   vbfloat16mf4_t vs1,
+                                                   size_t vl) {
+  return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vfloat32mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 vbfloat16mf2_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vfloat32m1_t vs2,
+                                                 vbfloat16mf2_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vfloat32m1_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2,
+                                                vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vfloat32m2_t vs2,
+                                                vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              vfloat32m2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2,
+                                                vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vfloat32m4_t vs2,
+                                                vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              vfloat32m4_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2,
+                                                vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vfloat32m8_t vs2,
+                                                vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              vfloat32m8_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwadd_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    vbfloat16mf4_t vs1,
+                                                    size_t vl) {
+  return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vfloat32mf2_t vs2,
+                                                    vbfloat16mf4_t vs1,
+                                                    size_t vl) {
+  return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vfloat32mf2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  vbfloat16mf2_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vfloat32m1_t vs2,
+                                                  vbfloat16mf2_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vfloat32m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2,
+                                                 vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vfloat32m2_t vs2,
+                                                 vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               vfloat32m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2,
+                                                 vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vfloat32m4_t vs2,
+                                                 vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               vfloat32m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2,
+                                                 vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwadd_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vfloat32m8_t vs2,
+                                                 vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               vfloat32m8_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vfloat32mf2_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwadd_wf_bf16_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwadd_wf_bf16_f32mf2_rm_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vfloat32m1_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwadd_wf_bf16_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwadd_wf_bf16_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vfloat32m1_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vfloat32m2_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwadd_wf_bf16_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwadd_wf_bf16_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             vfloat32m2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vfloat32m4_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwadd_wf_bf16_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwadd_wf_bf16_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             vfloat32m4_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwadd_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vfloat32m8_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwadd_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwadd_wf_bf16_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwadd_wf_bf16_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             vfloat32m8_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwadd_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwcvt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwcvt.c
new file mode 100644
index 0000000..7300104
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwcvt.c
@@ -0,0 +1,765 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tu(vbfloat16mf4_t vd, vint8mf8_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tu(vbfloat16mf2_t vd, vint8mf4_t vs2,
+                                            size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tu(vbfloat16m1_t vd, vint8mf2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tu(vbfloat16m2_t vd, vint8m1_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tu(vbfloat16m4_t vd, vint8m2_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tu(vbfloat16m8_t vd, vint8m4_t vs2,
+                                          size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tu(
+// CHECK-RV64-SAME: <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tu(vbfloat16mf4_t vd, vuint8mf8_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tu(
+// CHECK-RV64-SAME: <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tu(vbfloat16mf2_t vd, vuint8mf4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tu(
+// CHECK-RV64-SAME: <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tu(vbfloat16m1_t vd, vuint8mf2_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tu(
+// CHECK-RV64-SAME: <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tu(vbfloat16m2_t vd, vuint8m1_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tu(
+// CHECK-RV64-SAME: <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tu(vbfloat16m4_t vd, vuint8m2_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tu(
+// CHECK-RV64-SAME: <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tu(vbfloat16m8_t vd, vuint8m4_t vs2,
+                                           size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwcvt_f_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_tu(vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd,
+                                             vint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd,
+                                             vint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd,
+                                           vint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd,
+                                           vint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd,
+                                           vint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tum(vbool2_t vm, vbfloat16m8_t vd,
+                                           vint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tum(vbool64_t vm, vbfloat16mf4_t vd,
+                                              vuint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tum(vbool32_t vm, vbfloat16mf2_t vd,
+                                              vuint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tum(vbool16_t vm, vbfloat16m1_t vd,
+                                            vuint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tum(vbool8_t vm, vbfloat16m2_t vd,
+                                            vuint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tum(vbool4_t vm, vbfloat16m4_t vd,
+                                            vuint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tum(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tum(vbool2_t vm, vbfloat16m8_t vd,
+                                            vuint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwcvt_f_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwcvt_f_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_tum(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd,
+                                              vint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd,
+                                              vint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd,
+                                            vint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd,
+                                            vint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd,
+                                            vint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_tumu(vbool2_t vm, vbfloat16m8_t vd,
+                                            vint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_tumu(vbool64_t vm, vbfloat16mf4_t vd,
+                                               vuint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_tumu(vbool32_t vm, vbfloat16mf2_t vd,
+                                               vuint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_tumu(vbool16_t vm, vbfloat16m1_t vd,
+                                             vuint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_tumu(vbool8_t vm, vbfloat16m2_t vd,
+                                             vuint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_tumu(vbool4_t vm, vbfloat16m4_t vd,
+                                             vuint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_tumu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_tumu(vbool2_t vm, vbfloat16m8_t vd,
+                                             vuint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwcvt_f_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwcvt_f_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_tumu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_x_v_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_x_v_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd,
+                                            vint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_x_v_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_x_v_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd,
+                                            vint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_x_v_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_x_v_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd,
+                                          vint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_x_v_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_x_v_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd,
+                                          vint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_x_v_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_x_v_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd,
+                                          vint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_x_v_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_x_v_bf16m8_mu(vbool2_t vm, vbfloat16m8_t vd,
+                                          vint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x bfloat> @test_vfwcvt_f_xu_v_bf16mf4_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x bfloat> [[VD:%.*]], <vscale x 1 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8.i64(<vscale x 1 x bfloat> [[VD]], <vscale x 1 x i8> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x bfloat> [[TMP0]]
+//
+vbfloat16mf4_t test_vfwcvt_f_xu_v_bf16mf4_mu(vbool64_t vm, vbfloat16mf4_t vd,
+                                             vuint8mf8_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x bfloat> @test_vfwcvt_f_xu_v_bf16mf2_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x bfloat> [[VD:%.*]], <vscale x 2 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8.i64(<vscale x 2 x bfloat> [[VD]], <vscale x 2 x i8> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x bfloat> [[TMP0]]
+//
+vbfloat16mf2_t test_vfwcvt_f_xu_v_bf16mf2_mu(vbool32_t vm, vbfloat16mf2_t vd,
+                                             vuint8mf4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x bfloat> @test_vfwcvt_f_xu_v_bf16m1_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x bfloat> [[VD:%.*]], <vscale x 4 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8.i64(<vscale x 4 x bfloat> [[VD]], <vscale x 4 x i8> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x bfloat> [[TMP0]]
+//
+vbfloat16m1_t test_vfwcvt_f_xu_v_bf16m1_mu(vbool16_t vm, vbfloat16m1_t vd,
+                                           vuint8mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x bfloat> @test_vfwcvt_f_xu_v_bf16m2_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x bfloat> [[VD:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8.i64(<vscale x 8 x bfloat> [[VD]], <vscale x 8 x i8> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
+//
+vbfloat16m2_t test_vfwcvt_f_xu_v_bf16m2_mu(vbool8_t vm, vbfloat16m2_t vd,
+                                           vuint8m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x bfloat> @test_vfwcvt_f_xu_v_bf16m4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x bfloat> [[VD:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8.i64(<vscale x 16 x bfloat> [[VD]], <vscale x 16 x i8> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x bfloat> [[TMP0]]
+//
+vbfloat16m4_t test_vfwcvt_f_xu_v_bf16m4_mu(vbool4_t vm, vbfloat16m4_t vd,
+                                           vuint8m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x bfloat> @test_vfwcvt_f_xu_v_bf16m8_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[VM:%.*]], <vscale x 32 x bfloat> [[VD:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8.i64(<vscale x 32 x bfloat> [[VD]], <vscale x 32 x i8> [[VS2]], <vscale x 32 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 32 x bfloat> [[TMP0]]
+//
+vbfloat16m8_t test_vfwcvt_f_xu_v_bf16m8_mu(vbool2_t vm, vbfloat16m8_t vd,
+                                           vuint8m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_bf16_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwcvt_f_f_v_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwcvt_f_f_v_bf16mf4_f32mf2_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwcvt_f_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwcvt_f_f_v_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwcvt_f_f_v_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwcvt_f_f_v_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwcvt_f_f_v_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwcvt_f_f_v_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwcvt_f_f_v_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_mu(vm, vd, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwcvt_f_f_v_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwcvt.f.f.v.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwcvt_f_f_v_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwcvt_f_mu(vm, vd, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmacc.c
new file mode 100644
index 0000000..b05f8802
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmacc.c
@@ -0,0 +1,977 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                 __bf16 vs1, vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                               __bf16 vs1, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              __bf16 vs1, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              __bf16 vs1, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              __bf16 vs1, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                __bf16 vs1, vbfloat16mf4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                              __bf16 vs1, vbfloat16mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             __bf16 vs1, vbfloat16m1_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             __bf16 vs1, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             __bf16 vs1, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tum(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                  __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                 __bf16 vs1, vbfloat16m1_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                 __bf16 vs1, vbfloat16m2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                 __bf16 vs1, vbfloat16m4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                     vfloat32mf2_t vd,
+                                                     vbfloat16mf4_t vs1,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_tumu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                   vfloat32m1_t vd,
+                                                   vbfloat16mf2_t vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                   vfloat32m1_t vd, __bf16 vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                  vbfloat16m1_t vs1,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                  __bf16 vs1, vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                  vbfloat16m2_t vs1,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                  __bf16 vs1, vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                  vbfloat16m4_t vs1,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                  __bf16 vs1, vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmsac.c
new file mode 100644
index 0000000..93721f6
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmsac.c
@@ -0,0 +1,977 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                 __bf16 vs1, vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                               __bf16 vs1, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              __bf16 vs1, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              __bf16 vs1, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              __bf16 vs1, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs1,
+                                                vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                __bf16 vs1, vbfloat16mf4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs1,
+                                              vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                              __bf16 vs1, vbfloat16mf2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs1,
+                                             vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             __bf16 vs1, vbfloat16m1_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs1,
+                                             vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             __bf16 vs1, vbfloat16m2_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs1,
+                                             vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             __bf16 vs1, vbfloat16m4_t vs2,
+                                             size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tum(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                  __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                 __bf16 vs1, vbfloat16m1_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                 __bf16 vs1, vbfloat16m2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                 __bf16 vs1, vbfloat16m4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                     vfloat32mf2_t vd,
+                                                     vbfloat16mf4_t vs1,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_tumu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                   vfloat32m1_t vd,
+                                                   vbfloat16mf2_t vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                   vfloat32m1_t vd, __bf16 vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                  vbfloat16m1_t vs1,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                  __bf16 vs1, vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                  vbfloat16m2_t vs1,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                  __bf16 vs1, vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                  vbfloat16m4_t vs1,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                  __bf16 vs1, vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmul.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmul.c
new file mode 100644
index 0000000..4a2b5e3
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwmul.c
@@ -0,0 +1,975 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2,
+                                                vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2,
+                                              vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2,
+                                             vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2,
+                                             vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2,
+                                             vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   vbfloat16mf4_t vs1,
+                                                   size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 vbfloat16mf2_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2,
+                                                vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2,
+                                                vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2,
+                                                vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    vbfloat16mf4_t vs1,
+                                                    size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  vbfloat16mf2_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2,
+                                                 vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2,
+                                                 vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2,
+                                                 vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwmul_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmul_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmul_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmul_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmul_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmul_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmul_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmul_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmul_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmul_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmul_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwmul_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmacc.c
new file mode 100644
index 0000000..57e43344
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmacc.c
@@ -0,0 +1,994 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tu(vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tu(vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tu(vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tum(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tum(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 __bf16 vs1, vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                               __bf16 vs1, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                              __bf16 vs1, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                              __bf16 vs1, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                              __bf16 vs1, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                    __bf16 vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                     vfloat32mf2_t vd,
+                                                     vbfloat16mf4_t vs1,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tum(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm,
+                                                   vfloat32m1_t vd,
+                                                   vbfloat16mf2_t vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm,
+                                                   vfloat32m1_t vd, __bf16 vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                  vbfloat16m1_t vs1,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                  __bf16 vs1, vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                  vbfloat16m2_t vs1,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                  __bf16 vs1, vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                  vbfloat16m4_t vs1,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                  __bf16 vs1, vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                      vfloat32mf2_t vd,
+                                                      vbfloat16mf4_t vs1,
+                                                      vbfloat16mf4_t vs2,
+                                                      size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_tumu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                    vfloat32m1_t vd,
+                                                    vbfloat16mf2_t vs1,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                    vfloat32m1_t vd, __bf16 vs1,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm,
+                                                   vfloat32m2_t vd,
+                                                   vbfloat16m1_t vs1,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm,
+                                                   vfloat32m2_t vd, __bf16 vs1,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                   vbfloat16m2_t vs1,
+                                                   vbfloat16m2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                   __bf16 vs1,
+                                                   vbfloat16m2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                   vbfloat16m4_t vs1,
+                                                   vbfloat16m4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                   __bf16 vs1,
+                                                   vbfloat16m4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmacc_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_bf16mf4_f32mf2_rm_mu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                  __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                 __bf16 vs1, vbfloat16m1_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                 __bf16 vs1, vbfloat16m2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                 __bf16 vs1, vbfloat16m4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmacc_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmsac.c
new file mode 100644
index 0000000..42da060
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwnmsac.c
@@ -0,0 +1,994 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd, __bf16 vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd, __bf16 vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tu(vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, __bf16 vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tu(vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, __bf16 vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tu(vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, __bf16 vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tum(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tum(vbool64_t vm,
+                                                  vfloat32mf2_t vd, __bf16 vs1,
+                                                  vbfloat16mf4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs1,
+                                                vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                __bf16 vs1, vbfloat16mf2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs1,
+                                               vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                               __bf16 vs1, vbfloat16m1_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs1,
+                                               vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                               __bf16 vs1, vbfloat16m2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs1,
+                                               vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                               __bf16 vs1, vbfloat16m4_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_tumu(vbool64_t vm,
+                                                   vfloat32mf2_t vd, __bf16 vs1,
+                                                   vbfloat16mf4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs1,
+                                                 vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                 __bf16 vs1, vbfloat16mf2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs1,
+                                                vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                __bf16 vs1, vbfloat16m1_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs1,
+                                                vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                __bf16 vs1, vbfloat16m2_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs1,
+                                                vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                __bf16 vs1, vbfloat16m4_t vs2,
+                                                size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs1,
+                                                 vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 __bf16 vs1, vbfloat16mf4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs1,
+                                               vbfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                               __bf16 vs1, vbfloat16mf2_t vs2,
+                                               size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs1,
+                                              vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                              __bf16 vs1, vbfloat16m1_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs1,
+                                              vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                              __bf16 vs1, vbfloat16m2_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs1,
+                                              vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                              __bf16 vs1, vbfloat16m4_t vs2,
+                                              size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                    __bf16 vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd, __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd, __bf16 vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd, __bf16 vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd, __bf16 vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                     vfloat32mf2_t vd,
+                                                     vbfloat16mf4_t vs1,
+                                                     vbfloat16mf4_t vs2,
+                                                     size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tum(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm,
+                                                   vfloat32m1_t vd,
+                                                   vbfloat16mf2_t vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm,
+                                                   vfloat32m1_t vd, __bf16 vs1,
+                                                   vbfloat16mf2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                  vbfloat16m1_t vs1,
+                                                  vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                  __bf16 vs1, vbfloat16m1_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                  vbfloat16m2_t vs1,
+                                                  vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                  __bf16 vs1, vbfloat16m2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                  vbfloat16m4_t vs1,
+                                                  vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                  __bf16 vs1, vbfloat16m4_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_tum(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                      vfloat32mf2_t vd,
+                                                      vbfloat16mf4_t vs1,
+                                                      vbfloat16mf4_t vs2,
+                                                      size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_tumu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                    vfloat32m1_t vd,
+                                                    vbfloat16mf2_t vs1,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm,
+                                                    vfloat32m1_t vd, __bf16 vs1,
+                                                    vbfloat16mf2_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm,
+                                                   vfloat32m2_t vd,
+                                                   vbfloat16m1_t vs1,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm,
+                                                   vfloat32m2_t vd, __bf16 vs1,
+                                                   vbfloat16m1_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                   vbfloat16m2_t vs1,
+                                                   vbfloat16m2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                   __bf16 vs1,
+                                                   vbfloat16m2_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                   vbfloat16m4_t vs1,
+                                                   vbfloat16m4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                   __bf16 vs1,
+                                                   vbfloat16m4_t vs2,
+                                                   size_t vl) {
+  return __riscv_vfwnmsac_tumu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs1,
+                                                    vbfloat16mf4_t vs2,
+                                                    size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], bfloat [[VS1]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_bf16mf4_f32mf2_rm_mu(
+    vbool64_t vm, vfloat32mf2_t vd, __bf16 vs1, vbfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], bfloat [[VS1]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                  __bf16 vs1,
+                                                  vbfloat16mf2_t vs2,
+                                                  size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs1,
+                                                 vbfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], bfloat [[VS1]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                                 __bf16 vs1, vbfloat16m1_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs1,
+                                                 vbfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], bfloat [[VS1]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                                 __bf16 vs1, vbfloat16m2_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs1,
+                                                 vbfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], bfloat noundef [[VS1:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], bfloat [[VS1]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                                 __bf16 vs1, vbfloat16m4_t vs2,
+                                                 size_t vl) {
+  return __riscv_vfwnmsac_mu(vm, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwsub.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwsub.c
new file mode 100644
index 0000000..1378bc9
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vfwsub.c
@@ -0,0 +1,1932 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tu(vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tu(vfloat32mf2_t vd, vfloat32mf2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tu(vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tu(vfloat32m2_t vd, vbfloat16m1_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tu(vfloat32m4_t vd, vbfloat16m2_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tu(vfloat32m8_t vd, vbfloat16m4_t vs2,
+                                            __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tu(vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 7, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                          __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_tu(vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2,
+                                                vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vbfloat16mf4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vfloat32mf2_t vs2,
+                                                vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                             vfloat32mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2,
+                                              vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vbfloat16mf2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vfloat32m1_t vs2,
+                                              vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tum(vbool32_t vm, vfloat32m1_t vd,
+                                           vfloat32m1_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2,
+                                             vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vbfloat16m1_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                             vfloat32m2_t vs2,
+                                             vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tum(vbool16_t vm, vfloat32m2_t vd,
+                                           vfloat32m2_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2,
+                                             vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vbfloat16m2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                             vfloat32m4_t vs2,
+                                             vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tum(vbool8_t vm, vfloat32m4_t vd,
+                                           vfloat32m4_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2,
+                                             vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vbfloat16m4_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                             vfloat32m8_t vs2,
+                                             vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tum(vbool4_t vm, vfloat32m8_t vd,
+                                           vfloat32m8_t vs2, __bf16 rs1,
+                                           size_t vl) {
+  return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vbfloat16mf4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vfloat32mf2_t vs2,
+                                                 vbfloat16mf4_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                              vfloat32mf2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vbfloat16mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vfloat32m1_t vs2,
+                                               vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                            vfloat32m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vbfloat16m1_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                              vfloat32m2_t vs2,
+                                              vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                            vfloat32m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vbfloat16m2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                              vfloat32m4_t vs2,
+                                              vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                            vfloat32m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vbfloat16m4_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                              vfloat32m8_t vs2,
+                                              vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                            vfloat32m8_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vbfloat16mf4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2,
+                                               vbfloat16mf4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                            vfloat32mf2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vbfloat16mf2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vfloat32m1_t vs2,
+                                             vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_mu(vbool32_t vm, vfloat32m1_t vd,
+                                          vfloat32m1_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2,
+                                            vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vbfloat16m1_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                            vfloat32m2_t vs2, vbfloat16m1_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_mu(vbool16_t vm, vfloat32m2_t vd,
+                                          vfloat32m2_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2,
+                                            vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vbfloat16m2_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                            vfloat32m4_t vs2, vbfloat16m2_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_mu(vbool8_t vm, vfloat32m4_t vd,
+                                          vfloat32m4_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2,
+                                            vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vbfloat16m4_t vs2, __bf16 rs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                            vfloat32m8_t vs2, vbfloat16m4_t vs1,
+                                            size_t vl) {
+  return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 7, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_mu(vbool4_t vm, vfloat32m8_t vd,
+                                          vfloat32m8_t vs2, __bf16 rs1,
+                                          size_t vl) {
+  return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                                  vfloat32mf2_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tu(vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tu(vfloat32m1_t vd,
+                                                vfloat32m1_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tu(
+// CHECK-RV64-SAME: <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tu(vfloat32m1_t vd, vfloat32m1_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tu(vfloat32m2_t vd,
+                                               vfloat32m2_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tu(
+// CHECK-RV64-SAME: <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tu(vfloat32m2_t vd, vfloat32m2_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tu(vfloat32m4_t vd,
+                                               vfloat32m4_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tu(
+// CHECK-RV64-SAME: <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tu(vfloat32m4_t vd, vfloat32m4_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tu(vfloat32m8_t vd,
+                                               vfloat32m8_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tu(vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tu(
+// CHECK-RV64-SAME: <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tu(vfloat32m8_t vd, vfloat32m8_t vs2,
+                                             __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_wf_tu(vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   vbfloat16mf4_t vs1,
+                                                   size_t vl) {
+  return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vbfloat16mf4_t vs2,
+                                                   __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tum(vbool64_t vm,
+                                                   vfloat32mf2_t vd,
+                                                   vfloat32mf2_t vs2,
+                                                   vbfloat16mf4_t vs1,
+                                                   size_t vl) {
+  return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tum(vbool64_t vm, vfloat32mf2_t vd,
+                                                vfloat32mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2,
+                                                 vbfloat16mf2_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vbfloat16mf2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                                 vfloat32m1_t vs2,
+                                                 vbfloat16mf2_t vs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tum(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tum(vbool32_t vm, vfloat32m1_t vd,
+                                              vfloat32m1_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2,
+                                                vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vbfloat16m1_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                                vfloat32m2_t vs2,
+                                                vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tum(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tum(vbool16_t vm, vfloat32m2_t vd,
+                                              vfloat32m2_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2,
+                                                vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vbfloat16m2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                                vfloat32m4_t vs2,
+                                                vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tum(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tum(vbool8_t vm, vfloat32m4_t vd,
+                                              vfloat32m4_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2,
+                                                vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vbfloat16m4_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                                vfloat32m8_t vs2,
+                                                vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tum(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tum(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tum(vbool4_t vm, vfloat32m8_t vd,
+                                              vfloat32m8_t vs2, __bf16 rs1,
+                                              size_t vl) {
+  return __riscv_vfwsub_wf_tum(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    vbfloat16mf4_t vs1,
+                                                    size_t vl) {
+  return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vbfloat16mf4_t vs2,
+                                                    __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_tumu(vbool64_t vm,
+                                                    vfloat32mf2_t vd,
+                                                    vfloat32mf2_t vs2,
+                                                    vbfloat16mf4_t vs1,
+                                                    size_t vl) {
+  return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_tumu(vbool64_t vm, vfloat32mf2_t vd,
+                                                 vfloat32mf2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  vbfloat16mf2_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vbfloat16mf2_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                                  vfloat32m1_t vs2,
+                                                  vbfloat16mf2_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_tumu(vbool32_t vm, vfloat32m1_t vd,
+                                               vfloat32m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2,
+                                                 vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vbfloat16m1_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                                 vfloat32m2_t vs2,
+                                                 vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_tumu(vbool16_t vm, vfloat32m2_t vd,
+                                               vfloat32m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2,
+                                                 vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vbfloat16m2_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                                 vfloat32m4_t vs2,
+                                                 vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_tumu(vbool8_t vm, vfloat32m4_t vd,
+                                               vfloat32m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2,
+                                                 vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vbfloat16m4_t vs2, __bf16 rs1,
+                                                 size_t vl) {
+  return __riscv_vfwsub_vf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                                 vfloat32m8_t vs2,
+                                                 vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_tumu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_tumu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_tumu(vbool4_t vm, vfloat32m8_t vd,
+                                               vfloat32m8_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_tumu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_vf_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_vf_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vbfloat16mf4_t vs2,
+                                                  __bf16 rs1, size_t vl) {
+  return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wv_bf16mf4_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], <vscale x 1 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], <vscale x 1 x bfloat> [[VS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wv_bf16mf4_f32mf2_rm_mu(vbool64_t vm,
+                                                  vfloat32mf2_t vd,
+                                                  vfloat32mf2_t vs2,
+                                                  vbfloat16mf4_t vs1,
+                                                  size_t vl) {
+  return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwsub_wf_bf16_f32mf2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[VM:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x float> [[VS2]], bfloat [[RS1]], <vscale x 1 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwsub_wf_bf16_f32mf2_rm_mu(vbool64_t vm, vfloat32mf2_t vd,
+                                               vfloat32mf2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_vf_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_vf_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vbfloat16mf2_t vs2, __bf16 rs1,
+                                                size_t vl) {
+  return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wv_bf16mf2_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], <vscale x 2 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], <vscale x 2 x bfloat> [[VS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wv_bf16mf2_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                                vfloat32m1_t vs2,
+                                                vbfloat16mf2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwsub_wf_bf16_f32m1_rm_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[VM:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x float> [[VS2]], bfloat [[RS1]], <vscale x 2 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwsub_wf_bf16_f32m1_rm_mu(vbool32_t vm, vfloat32m1_t vd,
+                                             vfloat32m1_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_vf_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_vf_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vbfloat16m1_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wv_bf16m1_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], <vscale x 4 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], <vscale x 4 x bfloat> [[VS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wv_bf16m1_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                               vfloat32m2_t vs2,
+                                               vbfloat16m1_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwsub_wf_bf16_f32m2_rm_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[VM:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x float> [[VS2]], bfloat [[RS1]], <vscale x 4 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwsub_wf_bf16_f32m2_rm_mu(vbool16_t vm, vfloat32m2_t vd,
+                                             vfloat32m2_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_vf_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_vf_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vbfloat16m2_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wv_bf16m2_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], <vscale x 8 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], <vscale x 8 x bfloat> [[VS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wv_bf16m2_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                               vfloat32m4_t vs2,
+                                               vbfloat16m2_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwsub_wf_bf16_f32m4_rm_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[VM:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x float> [[VS2]], bfloat [[RS1]], <vscale x 8 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwsub_wf_bf16_f32m4_rm_mu(vbool8_t vm, vfloat32m4_t vd,
+                                             vfloat32m4_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_vv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_vf_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x bfloat> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x bfloat> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_vf_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vbfloat16m4_t vs2, __bf16 rs1,
+                                               size_t vl) {
+  return __riscv_vfwsub_vf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wv_bf16m4_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], <vscale x 16 x bfloat> [[VS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], <vscale x 16 x bfloat> [[VS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wv_bf16m4_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                               vfloat32m8_t vs2,
+                                               vbfloat16m4_t vs1, size_t vl) {
+  return __riscv_vfwsub_wv_mu(vm, vd, vs2, vs1, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwsub_wf_bf16_f32m8_rm_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[VM:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x float> [[VS2:%.*]], bfloat noundef [[RS1:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x float> [[VS2]], bfloat [[RS1]], <vscale x 16 x i1> [[VM]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwsub_wf_bf16_f32m8_rm_mu(vbool4_t vm, vfloat32m8_t vd,
+                                             vfloat32m8_t vs2, __bf16 rs1,
+                                             size_t vl) {
+  return __riscv_vfwsub_wf_mu(vm, vd, vs2, rs1, __RISCV_FRM_RNE, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfeq.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfeq.c
new file mode 100644
index 0000000..3945f82
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfeq.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfeq_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfeq_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfeq_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfeq_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfeq_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfeq_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfeq_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfeq_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfeq_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfeq_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfeq_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfeq.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfeq_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfeq_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfge.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfge.c
new file mode 100644
index 0000000..82586da
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfge.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfge_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfge_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfge_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfge_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfge_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfge_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfge_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfge_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfge_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfge_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfge_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfge.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfge_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfge_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfgt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfgt.c
new file mode 100644
index 0000000..75ccbbc
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfgt.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfgt_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfgt_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfgt_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfgt_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfgt_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfgt_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfgt_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfgt_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfgt_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfgt_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfgt_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfgt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfgt_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfgt_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfle.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfle.c
new file mode 100644
index 0000000..49ff1c9
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfle.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfle_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfle_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfle_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfle_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfle_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfle_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfle_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfle_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfle_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfle_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfle_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfle.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfle_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfle_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmflt.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmflt.c
new file mode 100644
index 0000000..24b3f9c
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmflt.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmflt_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmflt_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmflt_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmflt_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmflt_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmflt_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmflt_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmflt_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmflt_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmflt_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmflt_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmflt.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmflt_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmflt_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfne.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfne.c
new file mode 100644
index 0000000..ca3e134
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/zvfbfa/policy/overloaded/vmfne.c
@@ -0,0 +1,129 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v \
+// RUN:   -target-feature +experimental-zvfbfa -disable-O0-optnone \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vv_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], <vscale x 1 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.nxv1bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], <vscale x 1 x bfloat> [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vv_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, vbfloat16mf4_t op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x i1> @test_vmfne_vf_bf16mf4_b64_mu(
+// CHECK-RV64-SAME: <vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x i1> [[MASKEDOFF:%.*]], <vscale x 1 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.bf16.i64(<vscale x 1 x i1> [[MASKEDOFF]], <vscale x 1 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i1> [[TMP0]]
+//
+vbool64_t test_vmfne_vf_bf16mf4_b64_mu(vbool64_t mask, vbool64_t maskedoff, vbfloat16mf4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vv_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], <vscale x 2 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.nxv2bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], <vscale x 2 x bfloat> [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vv_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, vbfloat16mf2_t op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i1> @test_vmfne_vf_bf16mf2_b32_mu(
+// CHECK-RV64-SAME: <vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x i1> [[MASKEDOFF:%.*]], <vscale x 2 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.bf16.i64(<vscale x 2 x i1> [[MASKEDOFF]], <vscale x 2 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i1> [[TMP0]]
+//
+vbool32_t test_vmfne_vf_bf16mf2_b32_mu(vbool32_t mask, vbool32_t maskedoff, vbfloat16mf2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vv_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], <vscale x 4 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.nxv4bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], <vscale x 4 x bfloat> [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vv_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, vbfloat16m1_t op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i1> @test_vmfne_vf_bf16m1_b16_mu(
+// CHECK-RV64-SAME: <vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x i1> [[MASKEDOFF:%.*]], <vscale x 4 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.bf16.i64(<vscale x 4 x i1> [[MASKEDOFF]], <vscale x 4 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x i1> [[TMP0]]
+//
+vbool16_t test_vmfne_vf_bf16m1_b16_mu(vbool16_t mask, vbool16_t maskedoff, vbfloat16m1_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vv_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], <vscale x 8 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.nxv8bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], <vscale x 8 x bfloat> [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vv_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, vbfloat16m2_t op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i1> @test_vmfne_vf_bf16m2_b8_mu(
+// CHECK-RV64-SAME: <vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x i1> [[MASKEDOFF:%.*]], <vscale x 8 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.bf16.i64(<vscale x 8 x i1> [[MASKEDOFF]], <vscale x 8 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 8 x i1> [[TMP0]]
+//
+vbool8_t test_vmfne_vf_bf16m2_b8_mu(vbool8_t mask, vbool8_t maskedoff, vbfloat16m2_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vv_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], <vscale x 16 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.nxv16bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], <vscale x 16 x bfloat> [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vv_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, vbfloat16m4_t op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i1> @test_vmfne_vf_bf16m4_b4_mu(
+// CHECK-RV64-SAME: <vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x i1> [[MASKEDOFF:%.*]], <vscale x 16 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.bf16.i64(<vscale x 16 x i1> [[MASKEDOFF]], <vscale x 16 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
+//
+vbool4_t test_vmfne_vf_bf16m4_b4_mu(vbool4_t mask, vbool4_t maskedoff, vbfloat16m4_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vv_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], <vscale x 32 x bfloat> [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.nxv32bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], <vscale x 32 x bfloat> [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vv_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, vbfloat16m8_t op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 32 x i1> @test_vmfne_vf_bf16m8_b2_mu(
+// CHECK-RV64-SAME: <vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x i1> [[MASKEDOFF:%.*]], <vscale x 32 x bfloat> [[OP1:%.*]], bfloat noundef [[OP2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  [[ENTRY:.*:]]
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x i1> @llvm.riscv.vmfne.mask.nxv32bf16.bf16.i64(<vscale x 32 x i1> [[MASKEDOFF]], <vscale x 32 x bfloat> [[OP1]], bfloat [[OP2]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 32 x i1> [[TMP0]]
+//
+vbool2_t test_vmfne_vf_bf16m8_b2_mu(vbool2_t mask, vbool2_t maskedoff, vbfloat16m8_t op1, __bf16 op2, size_t vl) {
+  return __riscv_vmfne_mu(mask, maskedoff, op1, op2, vl);
+}
+
diff --git a/clang/test/CodeGen/X86/avx2-builtins.c b/clang/test/CodeGen/X86/avx2-builtins.c
index b798618..a505d70 100644
--- a/clang/test/CodeGen/X86/avx2-builtins.c
+++ b/clang/test/CodeGen/X86/avx2-builtins.c
@@ -1038,6 +1038,7 @@ __m256i test_mm256_mulhrs_epi16(__m256i a, __m256i b) {
   // CHECK: call <16 x i16> @llvm.x86.avx2.pmul.hr.sw(<16 x i16> %{{.*}}, <16 x i16> %{{.*}})
   return _mm256_mulhrs_epi16(a, b);
 }
+TEST_CONSTEXPR(match_v16hi(_mm256_mulhrs_epi16((__m256i)(__v16hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600}, (__m256i)(__v16hi){+1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +5, -9, -13, +16, +18, -20, -21, -22, -22, +21, +20, -18, -16, +13, +9, -5));
 
 __m256i test_mm256_mullo_epi16(__m256i a, __m256i b) {
   // CHECK-LABEL: test_mm256_mullo_epi16
diff --git a/clang/test/CodeGen/X86/avx512bw-builtins.c b/clang/test/CodeGen/X86/avx512bw-builtins.c
index fddf17d..55bf482 100644
--- a/clang/test/CodeGen/X86/avx512bw-builtins.c
+++ b/clang/test/CodeGen/X86/avx512bw-builtins.c
@@ -1596,18 +1596,24 @@ __m512i test_mm512_mulhrs_epi16(__m512i __A, __m512i __B) {
   // CHECK: @llvm.x86.avx512.pmul.hr.sw.512
   return _mm512_mulhrs_epi16(__A,__B); 
 }
+TEST_CONSTEXPR(match_v32hi(_mm512_mulhrs_epi16((__m512i)(__v32hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600, -1700, -1800, +1900, +2000, -2100, -2200, +2300, +2400, -2500, -2600, +2700, +2800, -2900, -3000, +3100, +3200}, (__m512i)(__v32hi){+3200, -3100, +3000, -2900, +2800, -2700, +2600, -2500, +2400, -2300, +2200, -2100, +2000, -1900, +1800, -1700, +1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +10, -19, -27, +35, +43, -49, -56, -61, -66, +70, +74, -77, -79, +81, +82, -83, -83, +82, +81, -79, -77, +74, +70, -66, -61, +56, +49, -43, -35, +27, +19, -10));
+
 __m512i test_mm512_mask_mulhrs_epi16(__m512i __W, __mmask32 __U, __m512i __A, __m512i __B) {
   // CHECK-LABEL: test_mm512_mask_mulhrs_epi16
   // CHECK: @llvm.x86.avx512.pmul.hr.sw.512
   // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_mask_mulhrs_epi16(__W,__U,__A,__B); 
 }
+TEST_CONSTEXPR(match_v32hi(_mm512_mask_mulhrs_epi16(_mm512_set1_epi16(1), 0x0000FFFF, (__m512i)(__v32hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600, -1700, -1800, +1900, +2000, -2100, -2200, +2300, +2400, -2500, -2600, +2700, +2800, -2900, -3000, +3100, +3200}, (__m512i)(__v32hi){+3200, -3100, +3000, -2900, +2800, -2700, +2600, -2500, +2400, -2300, +2200, -2100, +2000, -1900, +1800, -1700, +1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +10, -19, -27, +35, +43, -49, -56, -61, -66, +70, +74, -77, -79, +81, +82, -83, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1, +1));
+
 __m512i test_mm512_maskz_mulhrs_epi16(__mmask32 __U, __m512i __A, __m512i __B) {
   // CHECK-LABEL: test_mm512_maskz_mulhrs_epi16
   // CHECK: @llvm.x86.avx512.pmul.hr.sw.512
   // CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
   return _mm512_maskz_mulhrs_epi16(__U,__A,__B); 
 }
+TEST_CONSTEXPR(match_v32hi(_mm512_maskz_mulhrs_epi16(0x0000FFFF, (__m512i)(__v32hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600, -1700, -1800, +1900, +2000, -2100, -2200, +2300, +2400, -2500, -2600, +2700, +2800, -2900, -3000, +3100, +3200}, (__m512i)(__v32hi){+3200, -3100, +3000, -2900, +2800, -2700, +2600, -2500, +2400, -2300, +2200, -2100, +2000, -1900, +1800, -1700, +1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +10, -19, -27, +35, +43, -49, -56, -61, -66, +70, +74, -77, -79, +81, +82, -83, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0));
+
 __m512i test_mm512_mulhi_epi16(__m512i __A, __m512i __B) {
   // CHECK-LABEL: test_mm512_mulhi_epi16
   // CHECK: @llvm.x86.avx512.pmulh.w.512
diff --git a/clang/test/CodeGen/X86/avx512vlbw-builtins.c b/clang/test/CodeGen/X86/avx512vlbw-builtins.c
index d569283..95e4d40 100644
--- a/clang/test/CodeGen/X86/avx512vlbw-builtins.c
+++ b/clang/test/CodeGen/X86/avx512vlbw-builtins.c
@@ -2061,6 +2061,7 @@ __m128i test_mm_mask_mulhrs_epi16(__m128i __W, __mmask8 __U, __m128i __X, __m128
   // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_mask_mulhrs_epi16(__W, __U, __X, __Y); 
 }
+TEST_CONSTEXPR(match_v8hi(_mm_mask_mulhrs_epi16(_mm_set1_epi16(1), 0x0F, (__m128i)(__v8hi){+100, +200, -300, -400, +500, +600, -700, +800}, (__m128i)(__v8hi){+8000, -7000, +6000, -5000, +4000, -3000, +2000, -1000}), +24, -43, -55, +61, +1, +1, +1, +1));
 
 __m128i test_mm_maskz_mulhrs_epi16(__mmask8 __U, __m128i __X, __m128i __Y) {
   // CHECK-LABEL: test_mm_maskz_mulhrs_epi16
@@ -2068,6 +2069,7 @@ __m128i test_mm_maskz_mulhrs_epi16(__mmask8 __U, __m128i __X, __m128i __Y) {
   // CHECK: select <8 x i1> %{{.*}}, <8 x i16> %{{.*}}, <8 x i16> %{{.*}}
   return _mm_maskz_mulhrs_epi16(__U, __X, __Y); 
 }
+TEST_CONSTEXPR(match_v8hi(_mm_maskz_mulhrs_epi16(0x0F, (__m128i)(__v8hi){+100, +200, -300, -400, +500, +600, -700, +800}, (__m128i)(__v8hi){+8000, -7000, +6000, -5000, +4000, -3000, +2000, -1000}), +24, -43, -55, +61, 0, 0, 0, 0));
 
 __m256i test_mm256_mask_mulhrs_epi16(__m256i __W, __mmask16 __U, __m256i __X, __m256i __Y) {
   // CHECK-LABEL: test_mm256_mask_mulhrs_epi16
@@ -2075,6 +2077,7 @@ __m256i test_mm256_mask_mulhrs_epi16(__m256i __W, __mmask16 __U, __m256i __X, __
   // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_mask_mulhrs_epi16(__W, __U, __X, __Y); 
 }
+TEST_CONSTEXPR(match_v16hi(_mm256_mask_mulhrs_epi16(_mm256_set1_epi16(1), 0xF00F, (__m256i)(__v16hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600}, (__m256i)(__v16hi){+1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +5, -9, -13, +16, +1, +1, +1, +1, +1, +1, +1, +1, -16, +13, +9, -5));
 
 __m256i test_mm256_maskz_mulhrs_epi16(__mmask16 __U, __m256i __X, __m256i __Y) {
   // CHECK-LABEL: test_mm256_maskz_mulhrs_epi16
@@ -2082,6 +2085,7 @@ __m256i test_mm256_maskz_mulhrs_epi16(__mmask16 __U, __m256i __X, __m256i __Y) {
   // CHECK: select <16 x i1> %{{.*}}, <16 x i16> %{{.*}}, <16 x i16> %{{.*}}
   return _mm256_maskz_mulhrs_epi16(__U, __X, __Y); 
 }
+TEST_CONSTEXPR(match_v16hi(_mm256_maskz_mulhrs_epi16(0xF00F, (__m256i)(__v16hi){+100, +200, -300, -400, +500, +600, -700, +800, -900, -1000, +1100, +1200, -1300, -1400, +1500, +1600}, (__m256i)(__v16hi){+1600, -1500, +1400, -1300, +1200, -1100, +1000, -900, +800, -700, +600, -500, +400, -300, +200, -100}), +5, -9, -13, +16, 0, 0, 0, 0, 0, 0, 0, 0, -16, +13, +9, -5));
 
 __m128i test_mm_mask_mulhi_epu16(__m128i __W, __mmask8 __U, __m128i __A, __m128i __B) {
   // CHECK-LABEL: test_mm_mask_mulhi_epu16
diff --git a/clang/test/CodeGen/X86/mmx-builtins.c b/clang/test/CodeGen/X86/mmx-builtins.c
index d9041d4..c1ac57b 100644
--- a/clang/test/CodeGen/X86/mmx-builtins.c
+++ b/clang/test/CodeGen/X86/mmx-builtins.c
@@ -438,6 +438,7 @@ __m64 test_mm_mulhrs_pi16(__m64 a, __m64 b) {
   // CHECK: call <8 x i16> @llvm.x86.ssse3.pmul.hr.sw.128(
   return _mm_mulhrs_pi16(a, b);
 }
+TEST_CONSTEXPR(match_v4hi(_mm_mulhrs_pi16((__m64)(__v4hi){+100, +200, -300, -400}, (__m64)(__v4hi){+30000, -20000, +10000, -5000}), +92, -122, -92, +61));
 
 __m64 test_mm_mullo_pi16(__m64 a, __m64 b) {
   // CHECK-LABEL: test_mm_mullo_pi16
diff --git a/clang/test/CodeGen/X86/ssse3-builtins.c b/clang/test/CodeGen/X86/ssse3-builtins.c
index 32abd9d..f70afc0 100644
--- a/clang/test/CodeGen/X86/ssse3-builtins.c
+++ b/clang/test/CodeGen/X86/ssse3-builtins.c
@@ -110,6 +110,7 @@ __m128i test_mm_mulhrs_epi16(__m128i a, __m128i b) {
   // CHECK: call <8 x i16> @llvm.x86.ssse3.pmul.hr.sw.128(<8 x i16> %{{.*}}, <8 x i16> %{{.*}})
   return _mm_mulhrs_epi16(a, b);
 }
+TEST_CONSTEXPR(match_v8hi(_mm_mulhrs_epi16((__m128i)(__v8hi){+100, +200, -300, -400, +500, +600, -700, +800}, (__m128i)(__v8hi){+8000, -7000, +6000, -5000, +4000, -3000, +2000, -1000}), +24, -43, -55, +61, +61, -55, -43, -24));
 
 __m128i test_mm_shuffle_epi8(__m128i a, __m128i b) {
   // CHECK-LABEL: test_mm_shuffle_epi8
diff --git a/clang/unittests/Format/AlignBracketsTest.cpp b/clang/unittests/Format/AlignBracketsTest.cpp
index c4380ae..ea8db51 100644
--- a/clang/unittests/Format/AlignBracketsTest.cpp
+++ b/clang/unittests/Format/AlignBracketsTest.cpp
@@ -778,6 +778,19 @@ TEST_F(AlignBracketsTest, ParenthesesAndOperandAlignment) {
                Style);
 }
 
+TEST_F(AlignBracketsTest, BlockIndentAndNamespace) {
+  auto Style = getLLVMStyleWithColumns(120);
+  Style.AllowShortNamespacesOnASingleLine = true;
+  Style.AlignAfterOpenBracket = FormatStyle::BAS_BlockIndent;
+
+  verifyNoCrash(
+      "namespace {\n"
+      "void xxxxxxxxxxxxxxxxxxxxx(nnnnn::TTTTTTTTTTTTT const *mmmm,\n"
+      "                           YYYYYYYYYYYYYYYYY &yyyyyyyyyyyyyy);\n"
+      "} //",
+      Style);
+}
+
 } // namespace
 } // namespace test
 } // namespace format
diff --git a/clang/unittests/Format/ConfigParseTest.cpp b/clang/unittests/Format/ConfigParseTest.cpp
index 52f02c3..6488e38 100644
--- a/clang/unittests/Format/ConfigParseTest.cpp
+++ b/clang/unittests/Format/ConfigParseTest.cpp
@@ -176,7 +176,6 @@ TEST(ConfigParseTest, ParsesConfigurationBools) {
   CHECK_PARSE_BOOL(BreakBeforeTernaryOperators);
   CHECK_PARSE_BOOL(BreakStringLiterals);
   CHECK_PARSE_BOOL(CompactNamespaces);
-  CHECK_PARSE_BOOL(Cpp11BracedListStyle);
   CHECK_PARSE_BOOL(DerivePointerAlignment);
   CHECK_PARSE_BOOL_FIELD(DerivePointerAlignment, "DerivePointerBinding");
   CHECK_PARSE_BOOL(DisableFormat);
@@ -1139,6 +1138,18 @@ TEST(ConfigParseTest, ParsesConfiguration) {
               FormatStyle::SDS_Leave);
   CHECK_PARSE("SeparateDefinitionBlocks: Never", SeparateDefinitionBlocks,
               FormatStyle::SDS_Never);
+
+  CHECK_PARSE("Cpp11BracedListStyle: Block", Cpp11BracedListStyle,
+              FormatStyle::BLS_Block);
+  CHECK_PARSE("Cpp11BracedListStyle: FunctionCall", Cpp11BracedListStyle,
+              FormatStyle::BLS_FunctionCall);
+  CHECK_PARSE("Cpp11BracedListStyle: AlignFirstComment", Cpp11BracedListStyle,
+              FormatStyle::BLS_AlignFirstComment);
+  // For backward compatibility:
+  CHECK_PARSE("Cpp11BracedListStyle: false", Cpp11BracedListStyle,
+              FormatStyle::BLS_Block);
+  CHECK_PARSE("Cpp11BracedListStyle: true", Cpp11BracedListStyle,
+              FormatStyle::BLS_AlignFirstComment);
 }
 
 TEST(ConfigParseTest, ParsesConfigurationWithLanguages) {
diff --git a/clang/unittests/Format/FormatTest.cpp b/clang/unittests/Format/FormatTest.cpp
index b9ad930..0fb8139 100644
--- a/clang/unittests/Format/FormatTest.cpp
+++ b/clang/unittests/Format/FormatTest.cpp
@@ -14363,7 +14363,7 @@ TEST_F(FormatTest, LayoutCxx11BraceInitializers) {
       BreakBeforeLambdaBody);
 
   FormatStyle ExtraSpaces = getLLVMStyle();
-  ExtraSpaces.Cpp11BracedListStyle = false;
+  ExtraSpaces.Cpp11BracedListStyle = FormatStyle::BLS_Block;
   ExtraSpaces.ColumnLimit = 75;
   verifyFormat("vector<int> x{ 1, 2, 3, 4 };", ExtraSpaces);
   verifyFormat("vector<T> x{ {}, {}, {}, {} };", ExtraSpaces);
@@ -20346,7 +20346,7 @@ TEST_F(FormatTest, AlignConsecutiveDeclarations) {
                "  return 0;\n"
                "}()};",
                BracedAlign);
-  BracedAlign.Cpp11BracedListStyle = false;
+  BracedAlign.Cpp11BracedListStyle = FormatStyle::BLS_Block;
   verifyFormat("const auto result{ []() {\n"
                "  const auto something = 1;\n"
                "  return 2;\n"
@@ -21953,14 +21953,14 @@ TEST_F(FormatTest, CatchAlignArrayOfStructuresRightAlignment) {
                "});",
                Style);
 
-  Style.Cpp11BracedListStyle = false;
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_Block;
   verifyFormat("struct test demo[] = {\n"
                "  { 56,    23, \"hello\" },\n"
                "  { -1, 93463, \"world\" },\n"
                "  {  7,     5,    \"!!\" }\n"
                "};",
                Style);
-  Style.Cpp11BracedListStyle = true;
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_AlignFirstComment;
 
   Style.ColumnLimit = 0;
   verifyFormat(
@@ -22220,14 +22220,14 @@ TEST_F(FormatTest, CatchAlignArrayOfStructuresLeftAlignment) {
                "  };",
                Style);
 
-  Style.Cpp11BracedListStyle = false;
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_Block;
   verifyFormat("struct test demo[] = {\n"
                "  { 56, 23,    \"hello\" },\n"
                "  { -1, 93463, \"world\" },\n"
                "  { 7,  5,     \"!!\"    }\n"
                "};",
                Style);
-  Style.Cpp11BracedListStyle = true;
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_AlignFirstComment;
 
   Style.ColumnLimit = 0;
   verifyFormat(
diff --git a/clang/unittests/Format/FormatTestCSharp.cpp b/clang/unittests/Format/FormatTestCSharp.cpp
index ea85ed6..d7fb15d 100644
--- a/clang/unittests/Format/FormatTestCSharp.cpp
+++ b/clang/unittests/Format/FormatTestCSharp.cpp
@@ -1194,7 +1194,7 @@ TEST_F(FormatTestCSharp, CSharpSpaces) {
   Style.SpaceBeforeSquareBrackets = false;
   Style.SpacesInSquareBrackets = false;
   Style.SpaceBeforeCpp11BracedList = true;
-  Style.Cpp11BracedListStyle = false;
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_Block;
   Style.SpacesInContainerLiterals = false;
   Style.SpaceAfterCStyleCast = false;
 
diff --git a/clang/unittests/Format/FormatTestComments.cpp b/clang/unittests/Format/FormatTestComments.cpp
index 69026bc..fc80bf4 100644
--- a/clang/unittests/Format/FormatTestComments.cpp
+++ b/clang/unittests/Format/FormatTestComments.cpp
@@ -4699,6 +4699,58 @@ TEST_F(FormatTestComments, SplitCommentIntroducers) {
                    getLLVMStyleWithColumns(10)));
 }
 
+TEST_F(FormatTestComments, LineCommentsOnStartOfFunctionCall) {
+  auto Style = getLLVMStyle();
+
+  EXPECT_EQ(Style.Cpp11BracedListStyle, FormatStyle::BLS_AlignFirstComment);
+  verifyFormat("Type name{// Comment\n"
+               "          value};",
+               Style);
+
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_Block;
+  verifyFormat("Type name{ // Comment\n"
+               "           value\n"
+               "};",
+               Style);
+
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_FunctionCall;
+  verifyFormat("Type name{ // Comment\n"
+               "    value};",
+               Style);
+
+  verifyFormat("T foo( // Comment\n"
+               "    arg);",
+               Style);
+
+  verifyFormat("T bar{ // Comment\n"
+               "    arg};",
+               Style);
+
+  verifyFormat("T baz({ // Comment\n"
+               "    arg});",
+               Style);
+
+  verifyFormat("T baz{{ // Comment\n"
+               "    arg}};",
+               Style);
+
+  verifyFormat("T b0z(f( // Comment\n"
+               "    arg));",
+               Style);
+
+  verifyFormat("T b0z(F{ // Comment\n"
+               "    arg});",
+               Style);
+
+  verifyFormat("func( // Comment\n"
+               "    arg);",
+               Style);
+
+  verifyFormat("func({ // Comment\n"
+               "    arg});",
+               Style);
+}
+
 } // end namespace
 } // namespace test
 } // end namespace format
diff --git a/clang/unittests/Format/FormatTestJava.cpp b/clang/unittests/Format/FormatTestJava.cpp
index 1275564..1416614b 100644
--- a/clang/unittests/Format/FormatTestJava.cpp
+++ b/clang/unittests/Format/FormatTestJava.cpp
@@ -236,7 +236,7 @@ TEST_F(FormatTestJava, ArrayInitializers) {
                "};");
 
   FormatStyle Style = getStyleWithColumns(65);
-  Style.Cpp11BracedListStyle = false;
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_Block;
   verifyFormat(
       "expected = new int[] { 10, 10, 10, 10, 10, 10, 10, 10, 10, 10,\n"
       "  100, 100, 100, 100, 100, 100, 100, 100, 100, 100 };",
diff --git a/clang/unittests/Format/FormatTestTextProto.cpp b/clang/unittests/Format/FormatTestTextProto.cpp
index fd65c9a..6cddb838 100644
--- a/clang/unittests/Format/FormatTestTextProto.cpp
+++ b/clang/unittests/Format/FormatTestTextProto.cpp
@@ -514,7 +514,7 @@ TEST_F(FormatTestTextProto, FormatsRepeatedListInitializers) {
                "key: value");
 
   auto Style = getDefaultStyle();
-  Style.Cpp11BracedListStyle = true;
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_AlignFirstComment;
   verifyFormat("keys: [1]", Style);
 }
 
diff --git a/clang/unittests/Format/FormatTestVerilog.cpp b/clang/unittests/Format/FormatTestVerilog.cpp
index 5c50ae6..63e2cadf 100644
--- a/clang/unittests/Format/FormatTestVerilog.cpp
+++ b/clang/unittests/Format/FormatTestVerilog.cpp
@@ -1287,7 +1287,7 @@ TEST_F(FormatTestVerilog, StringLiteral) {
                getStyleWithColumns(getDefaultStyle(), 32));
   // Space around braces should be correct.
   auto Style = getStyleWithColumns(getDefaultStyle(), 24);
-  Style.Cpp11BracedListStyle = false;
+  Style.Cpp11BracedListStyle = FormatStyle::BLS_Block;
   verifyFormat(R"(x({ "xxxxxxxxxxxxxxxx ",
     "xxxx" });)",
                R"(x("xxxxxxxxxxxxxxxx xxxx");)", Style);
diff --git a/clang/unittests/Format/TokenAnnotatorTest.cpp b/clang/unittests/Format/TokenAnnotatorTest.cpp
index 1152466..1002515 100644
--- a/clang/unittests/Format/TokenAnnotatorTest.cpp
+++ b/clang/unittests/Format/TokenAnnotatorTest.cpp
@@ -1129,6 +1129,11 @@ TEST_F(TokenAnnotatorTest, UnderstandsOverloadedOperators) {
   ASSERT_EQ(Tokens.size(), 7u) << Tokens;
   // Not TT_FunctionDeclarationName.
   EXPECT_TOKEN(Tokens[3], tok::kw_operator, TT_Unknown);
+
+  Tokens = annotate("SomeAPI::operator()();");
+  ASSERT_EQ(Tokens.size(), 9u) << Tokens;
+  // Not TT_FunctionDeclarationName.
+  EXPECT_TOKEN(Tokens[2], tok::kw_operator, TT_Unknown);
 }
 
 TEST_F(TokenAnnotatorTest, OverloadedOperatorInTemplate) {
diff --git a/compiler-rt/test/asan/TestCases/Windows/basic_exception_handling.cpp b/compiler-rt/test/asan/TestCases/Windows/basic_exception_handling.cpp
index 6f02814..8d1b9ef1 100644
--- a/compiler-rt/test/asan/TestCases/Windows/basic_exception_handling.cpp
+++ b/compiler-rt/test/asan/TestCases/Windows/basic_exception_handling.cpp
@@ -7,13 +7,14 @@
 // This code is based on the repro in https://github.com/google/sanitizers/issues/749
 #include <cstdio>
 #include <exception>
+#include <stdexcept>
 
-void throwInFunction() { throw std::exception("test2"); }
+void throwInFunction() { throw std::runtime_error("test2"); }
 
 int main() {
   // case 1: direct throw
   try {
-    throw std::exception("test1");
+    throw std::runtime_error("test1");
   } catch (const std::exception &ex) {
     puts(ex.what());
     // CHECK: test1
diff --git a/llvm/benchmarks/SpecialCaseListBM.cpp b/llvm/benchmarks/SpecialCaseListBM.cpp
index 00aa3cd..b5d8268 100644
--- a/llvm/benchmarks/SpecialCaseListBM.cpp
+++ b/llvm/benchmarks/SpecialCaseListBM.cpp
@@ -110,6 +110,26 @@ std::string genGlobAtBothSides(const std::vector<std::string> &Files) {
   return S;
 }
 
+std::string genGlobAtBothSidesAndMid(const std::vector<std::string> &Files) {
+  std::string S;
+  std::minstd_rand Rng(RNG_SEED);
+  for (std::string F : Files) {
+    std::uniform_int_distribution<> PosDistrib(0, F.size() - 1);
+    F[PosDistrib(Rng)] = '*';
+
+    std::uniform_int_distribution<> Ends(0, 1);
+    if (Ends(Rng)) {
+      F.back() = '*';
+      F.front() = '*';
+    }
+
+    S += "src:";
+    S += F;
+    S += "\n";
+  }
+  return S;
+}
+
 void BM_Make_(
     benchmark::State &state,
     std::string (*GenerateCaseList)(const std::vector<std::string> &Files)) {
@@ -171,6 +191,9 @@ BENCHMARK_CAPTURE(BM_Make_, Mid__, genGlobInMid)
 BENCHMARK_CAPTURE(BM_Make_, Both_, genGlobAtBothSides)
     ->RangeMultiplier(MAX_LIST_MUL)
     ->Range(MAX_LIST_MIN, MAX_LIST_MAX);
+BENCHMARK_CAPTURE(BM_Make_, Mix__, genGlobAtBothSidesAndMid)
+    ->RangeMultiplier(MAX_LIST_MUL)
+    ->Range(MAX_LIST_MIN, MAX_LIST_MAX);
 
 BENCHMARK_CAPTURE(BM_True_, None_, genGlobNone)
     ->RangeMultiplier(MAX_LIST_MUL)
@@ -187,6 +210,9 @@ BENCHMARK_CAPTURE(BM_True_, Mid__, genGlobInMid)
 BENCHMARK_CAPTURE(BM_True_, Both_, genGlobAtBothSides)
     ->RangeMultiplier(MAX_LIST_MUL)
     ->Range(MAX_LIST_MIN, MAX_LIST_MAX);
+BENCHMARK_CAPTURE(BM_True_, Mix__, genGlobAtBothSidesAndMid)
+    ->RangeMultiplier(MAX_LIST_MUL)
+    ->Range(MAX_LIST_MIN, MAX_LIST_MAX);
 
 BENCHMARK_CAPTURE(BM_False, None_, genGlobNone)
     ->RangeMultiplier(MAX_LIST_MUL)
@@ -203,5 +229,8 @@ BENCHMARK_CAPTURE(BM_False, Mid__, genGlobInMid)
 BENCHMARK_CAPTURE(BM_False, Both_, genGlobAtBothSides)
     ->RangeMultiplier(MAX_LIST_MUL)
     ->Range(MAX_LIST_MIN, MAX_LIST_MAX);
+BENCHMARK_CAPTURE(BM_False, Mix__, genGlobAtBothSidesAndMid)
+    ->RangeMultiplier(MAX_LIST_MUL)
+    ->Range(MAX_LIST_MIN, MAX_LIST_MAX);
 
 BENCHMARK_MAIN();
diff --git a/llvm/include/llvm/ADT/APFloat.h b/llvm/include/llvm/ADT/APFloat.h
index 3eb3d6f..bccdb89 100644
--- a/llvm/include/llvm/ADT/APFloat.h
+++ b/llvm/include/llvm/ADT/APFloat.h
@@ -967,69 +967,11 @@ class APFloat : public APFloatBase {
       llvm_unreachable("Unexpected semantics");
     }
 
-    ~Storage() {
-      if (usesLayout<IEEEFloat>(*semantics)) {
-        IEEE.~IEEEFloat();
-        return;
-      }
-      if (usesLayout<DoubleAPFloat>(*semantics)) {
-        Double.~DoubleAPFloat();
-        return;
-      }
-      llvm_unreachable("Unexpected semantics");
-    }
-
-    Storage(const Storage &RHS) {
-      if (usesLayout<IEEEFloat>(*RHS.semantics)) {
-        new (this) IEEEFloat(RHS.IEEE);
-        return;
-      }
-      if (usesLayout<DoubleAPFloat>(*RHS.semantics)) {
-        new (this) DoubleAPFloat(RHS.Double);
-        return;
-      }
-      llvm_unreachable("Unexpected semantics");
-    }
-
-    Storage(Storage &&RHS) {
-      if (usesLayout<IEEEFloat>(*RHS.semantics)) {
-        new (this) IEEEFloat(std::move(RHS.IEEE));
-        return;
-      }
-      if (usesLayout<DoubleAPFloat>(*RHS.semantics)) {
-        new (this) DoubleAPFloat(std::move(RHS.Double));
-        return;
-      }
-      llvm_unreachable("Unexpected semantics");
-    }
-
-    Storage &operator=(const Storage &RHS) {
-      if (usesLayout<IEEEFloat>(*semantics) &&
-          usesLayout<IEEEFloat>(*RHS.semantics)) {
-        IEEE = RHS.IEEE;
-      } else if (usesLayout<DoubleAPFloat>(*semantics) &&
-                 usesLayout<DoubleAPFloat>(*RHS.semantics)) {
-        Double = RHS.Double;
-      } else if (this != &RHS) {
-        this->~Storage();
-        new (this) Storage(RHS);
-      }
-      return *this;
-    }
-
-    Storage &operator=(Storage &&RHS) {
-      if (usesLayout<IEEEFloat>(*semantics) &&
-          usesLayout<IEEEFloat>(*RHS.semantics)) {
-        IEEE = std::move(RHS.IEEE);
-      } else if (usesLayout<DoubleAPFloat>(*semantics) &&
-                 usesLayout<DoubleAPFloat>(*RHS.semantics)) {
-        Double = std::move(RHS.Double);
-      } else if (this != &RHS) {
-        this->~Storage();
-        new (this) Storage(std::move(RHS));
-      }
-      return *this;
-    }
+    LLVM_ABI ~Storage();
+    LLVM_ABI Storage(const Storage &RHS);
+    LLVM_ABI Storage(Storage &&RHS);
+    LLVM_ABI Storage &operator=(const Storage &RHS);
+    LLVM_ABI Storage &operator=(Storage &&RHS);
   } U;
 
   template <typename T> static bool usesLayout(const fltSemantics &Semantics) {
diff --git a/llvm/include/llvm/ADT/Bitfields.h b/llvm/include/llvm/ADT/Bitfields.h
index 1af2761..1fbc41c 100644
--- a/llvm/include/llvm/ADT/Bitfields.h
+++ b/llvm/include/llvm/ADT/Bitfields.h
@@ -154,12 +154,9 @@ struct ResolveUnderlyingType {
   using type = std::underlying_type_t<T>;
 };
 template <typename T> struct ResolveUnderlyingType<T, false> {
-  using type = T;
-};
-template <> struct ResolveUnderlyingType<bool, false> {
-  /// In case sizeof(bool) != 1, replace `void` by an additionnal
-  /// std::conditional.
-  using type = std::conditional_t<sizeof(bool) == 1, uint8_t, void>;
+  static_assert(!std::is_same_v<T, bool> || sizeof(bool) == 1,
+                "T being bool requires sizeof(bool) == 1.");
+  using type = std::conditional_t<std::is_same_v<T, bool>, uint8_t, T>;
 };
 
 } // namespace bitfields_details
diff --git a/llvm/include/llvm/ADT/DenseMap.h b/llvm/include/llvm/ADT/DenseMap.h
index 4bda50f..25b5262 100644
--- a/llvm/include/llvm/ADT/DenseMap.h
+++ b/llvm/include/llvm/ADT/DenseMap.h
@@ -42,7 +42,7 @@ namespace detail {
 // We extend a pair to allow users to override the bucket type with their own
 // implementation without requiring two members.
 template <typename KeyT, typename ValueT>
-struct DenseMapPair : public std::pair<KeyT, ValueT> {
+struct DenseMapPair : std::pair<KeyT, ValueT> {
   using std::pair<KeyT, ValueT>::pair;
 
   KeyT &getFirst() { return std::pair<KeyT, ValueT>::first; }
diff --git a/llvm/include/llvm/ADT/DepthFirstIterator.h b/llvm/include/llvm/ADT/DepthFirstIterator.h
index 4ced758..3c54f32 100644
--- a/llvm/include/llvm/ADT/DepthFirstIterator.h
+++ b/llvm/include/llvm/ADT/DepthFirstIterator.h
@@ -66,8 +66,8 @@ public:
 // one more method, completed, which is invoked when all children of a
 // node have been processed. It is intended to distinguish of back and
 // cross edges in the spanning tree but is not used in the common case.
-template <typename NodeRef, unsigned SmallSize=8>
-struct df_iterator_default_set : public SmallPtrSet<NodeRef, SmallSize> {
+template <typename NodeRef, unsigned SmallSize = 8>
+struct df_iterator_default_set : SmallPtrSet<NodeRef, SmallSize> {
   using BaseSet = SmallPtrSet<NodeRef, SmallSize>;
   using iterator = typename BaseSet::iterator;
 
@@ -235,8 +235,10 @@ iterator_range<df_iterator<T>> depth_first(const T& G) {
 }
 
 // Provide global definitions of external depth first iterators...
-template <class T, class SetTy = df_iterator_default_set<typename GraphTraits<T>::NodeRef>>
-struct df_ext_iterator : public df_iterator<T, SetTy, true> {
+template <class T,
+          class SetTy =
+              df_iterator_default_set<typename GraphTraits<T>::NodeRef>>
+struct df_ext_iterator : df_iterator<T, SetTy, true> {
   df_ext_iterator(const df_iterator<T, SetTy, true> &V)
     : df_iterator<T, SetTy, true>(V) {}
 };
@@ -262,7 +264,7 @@ template <class T,
           class SetTy =
               df_iterator_default_set<typename GraphTraits<T>::NodeRef>,
           bool External = false>
-struct idf_iterator : public df_iterator<Inverse<T>, SetTy, External> {
+struct idf_iterator : df_iterator<Inverse<T>, SetTy, External> {
   idf_iterator(const df_iterator<Inverse<T>, SetTy, External> &V)
     : df_iterator<Inverse<T>, SetTy, External>(V) {}
 };
@@ -284,8 +286,10 @@ iterator_range<idf_iterator<T>> inverse_depth_first(const T& G) {
 }
 
 // Provide global definitions of external inverse depth first iterators...
-template <class T, class SetTy = df_iterator_default_set<typename GraphTraits<T>::NodeRef>>
-struct idf_ext_iterator : public idf_iterator<T, SetTy, true> {
+template <class T,
+          class SetTy =
+              df_iterator_default_set<typename GraphTraits<T>::NodeRef>>
+struct idf_ext_iterator : idf_iterator<T, SetTy, true> {
   idf_ext_iterator(const idf_iterator<T, SetTy, true> &V)
     : idf_iterator<T, SetTy, true>(V) {}
   idf_ext_iterator(const df_iterator<Inverse<T>, SetTy, true> &V)
diff --git a/llvm/include/llvm/ADT/ImmutableSet.h b/llvm/include/llvm/ADT/ImmutableSet.h
index 310539f..8b2425e 100644
--- a/llvm/include/llvm/ADT/ImmutableSet.h
+++ b/llvm/include/llvm/ADT/ImmutableSet.h
@@ -931,8 +931,7 @@ struct ImutProfileInfo<T*> {
 /// ImutContainerInfo - Generic definition of comparison operations for
 ///   elements of immutable containers that defaults to using
 ///   std::equal_to<> and std::less<> to perform comparison of elements.
-template <typename T>
-struct ImutContainerInfo : public ImutProfileInfo<T> {
+template <typename T> struct ImutContainerInfo : ImutProfileInfo<T> {
   using value_type = typename ImutProfileInfo<T>::value_type;
   using value_type_ref = typename ImutProfileInfo<T>::value_type_ref;
   using key_type = value_type;
@@ -957,8 +956,7 @@ struct ImutContainerInfo : public ImutProfileInfo<T> {
 /// ImutContainerInfo - Specialization for pointer values to treat pointers
 ///  as references to unique objects.  Pointers are thus compared by
 ///  their addresses.
-template <typename T>
-struct ImutContainerInfo<T*> : public ImutProfileInfo<T*> {
+template <typename T> struct ImutContainerInfo<T *> : ImutProfileInfo<T *> {
   using value_type = typename ImutProfileInfo<T*>::value_type;
   using value_type_ref = typename ImutProfileInfo<T*>::value_type_ref;
   using key_type = value_type;
diff --git a/llvm/include/llvm/ADT/PostOrderIterator.h b/llvm/include/llvm/ADT/PostOrderIterator.h
index 1cbd3c1..d9aa452 100644
--- a/llvm/include/llvm/ADT/PostOrderIterator.h
+++ b/llvm/include/llvm/ADT/PostOrderIterator.h
@@ -200,7 +200,7 @@ template <class T> iterator_range<po_iterator<T>> post_order(const T &G) {
 
 // Provide global definitions of external postorder iterators...
 template <class T, class SetType = std::set<typename GraphTraits<T>::NodeRef>>
-struct po_ext_iterator : public po_iterator<T, SetType, true> {
+struct po_ext_iterator : po_iterator<T, SetType, true> {
   po_ext_iterator(const po_iterator<T, SetType, true> &V) :
   po_iterator<T, SetType, true>(V) {}
 };
@@ -223,7 +223,7 @@ iterator_range<po_ext_iterator<T, SetType>> post_order_ext(const T &G, SetType &
 // Provide global definitions of inverse post order iterators...
 template <class T, class SetType = std::set<typename GraphTraits<T>::NodeRef>,
           bool External = false>
-struct ipo_iterator : public po_iterator<Inverse<T>, SetType, External> {
+struct ipo_iterator : po_iterator<Inverse<T>, SetType, External> {
   ipo_iterator(const po_iterator<Inverse<T>, SetType, External> &V) :
      po_iterator<Inverse<T>, SetType, External> (V) {}
 };
@@ -245,7 +245,7 @@ iterator_range<ipo_iterator<T>> inverse_post_order(const T &G) {
 
 // Provide global definitions of external inverse postorder iterators...
 template <class T, class SetType = std::set<typename GraphTraits<T>::NodeRef>>
-struct ipo_ext_iterator : public ipo_iterator<T, SetType, true> {
+struct ipo_ext_iterator : ipo_iterator<T, SetType, true> {
   ipo_ext_iterator(const ipo_iterator<T, SetType, true> &V) :
     ipo_iterator<T, SetType, true>(V) {}
   ipo_ext_iterator(const po_iterator<Inverse<T>, SetType, true> &V) :
diff --git a/llvm/include/llvm/ADT/STLExtras.h b/llvm/include/llvm/ADT/STLExtras.h
index 658f262..a9841c6 100644
--- a/llvm/include/llvm/ADT/STLExtras.h
+++ b/llvm/include/llvm/ADT/STLExtras.h
@@ -674,7 +674,7 @@ using zip_traits = iterator_facade_base<
     ReferenceTupleType *, ReferenceTupleType>;
 
 template <typename ZipType, typename ReferenceTupleType, typename... Iters>
-struct zip_common : public zip_traits<ZipType, ReferenceTupleType, Iters...> {
+struct zip_common : zip_traits<ZipType, ReferenceTupleType, Iters...> {
   using Base = zip_traits<ZipType, ReferenceTupleType, Iters...>;
   using IndexSequence = std::index_sequence_for<Iters...>;
   using value_type = typename Base::value_type;
diff --git a/llvm/include/llvm/ADT/STLForwardCompat.h b/llvm/include/llvm/ADT/STLForwardCompat.h
index da9d3ab0..273a5cf 100644
--- a/llvm/include/llvm/ADT/STLForwardCompat.h
+++ b/llvm/include/llvm/ADT/STLForwardCompat.h
@@ -26,6 +26,54 @@ namespace llvm {
 //     Features from C++20
 //===----------------------------------------------------------------------===//
 
+namespace numbers {
+// clang-format off
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T e_v          = T(0x1.5bf0a8b145769P+1); // (2.7182818284590452354) https://oeis.org/A001113
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T egamma_v     = T(0x1.2788cfc6fb619P-1); // (.57721566490153286061) https://oeis.org/A001620
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T ln2_v        = T(0x1.62e42fefa39efP-1); // (.69314718055994530942) https://oeis.org/A002162
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T ln10_v       = T(0x1.26bb1bbb55516P+1); // (2.3025850929940456840) https://oeis.org/A002392
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T log2e_v      = T(0x1.71547652b82feP+0); // (1.4426950408889634074)
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T log10e_v     = T(0x1.bcb7b1526e50eP-2); // (.43429448190325182765)
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T pi_v         = T(0x1.921fb54442d18P+1); // (3.1415926535897932385) https://oeis.org/A000796
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T inv_pi_v     = T(0x1.45f306dc9c883P-2); // (.31830988618379067154) https://oeis.org/A049541
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T inv_sqrtpi_v = T(0x1.20dd750429b6dP-1); // (.56418958354775628695) https://oeis.org/A087197
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T sqrt2_v      = T(0x1.6a09e667f3bcdP+0); // (1.4142135623730950488) https://oeis.org/A00219
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T inv_sqrt2_v  = T(0x1.6a09e667f3bcdP-1); // (.70710678118654752440)
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T sqrt3_v      = T(0x1.bb67ae8584caaP+0); // (1.7320508075688772935) https://oeis.org/A002194
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T inv_sqrt3_v  = T(0x1.279a74590331cP-1); // (.57735026918962576451)
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T phi_v        = T(0x1.9e3779b97f4a8P+0); // (1.6180339887498948482) https://oeis.org/A001622
+
+inline constexpr double e          = e_v<double>;
+inline constexpr double egamma     = egamma_v<double>;
+inline constexpr double ln2        = ln2_v<double>;
+inline constexpr double ln10       = ln10_v<double>;
+inline constexpr double log2e      = log2e_v<double>;
+inline constexpr double log10e     = log10e_v<double>;
+inline constexpr double pi         = pi_v<double>;
+inline constexpr double inv_pi     = inv_pi_v<double>;
+inline constexpr double inv_sqrtpi = inv_sqrtpi_v<double>;
+inline constexpr double sqrt2      = sqrt2_v<double>;
+inline constexpr double inv_sqrt2  = inv_sqrt2_v<double>;
+inline constexpr double sqrt3      = sqrt3_v<double>;
+inline constexpr double inv_sqrt3  = inv_sqrt3_v<double>;
+inline constexpr double phi        = phi_v<double>;
+// clang-format on
+} // namespace numbers
+
 template <typename T>
 struct remove_cvref // NOLINT(readability-identifier-naming)
 {
diff --git a/llvm/include/llvm/ADT/SmallPtrSet.h b/llvm/include/llvm/ADT/SmallPtrSet.h
index f588a77..8e7c8b3 100644
--- a/llvm/include/llvm/ADT/SmallPtrSet.h
+++ b/llvm/include/llvm/ADT/SmallPtrSet.h
@@ -532,18 +532,8 @@ class SmallPtrSet : public SmallPtrSetImpl<PtrType> {
 
   using BaseT = SmallPtrSetImpl<PtrType>;
 
-  // A constexpr version of llvm::bit_ceil.
-  // TODO: Replace this with std::bit_ceil once C++20 is available.
-  static constexpr size_t RoundUpToPowerOfTwo(size_t X) {
-    size_t C = 1;
-    size_t CMax = C << (std::numeric_limits<size_t>::digits - 1);
-    while (C < X && C < CMax)
-      C <<= 1;
-    return C;
-  }
-
   // Make sure that SmallSize is a power of two, round up if not.
-  static constexpr size_t SmallSizePowTwo = RoundUpToPowerOfTwo(SmallSize);
+  static constexpr size_t SmallSizePowTwo = llvm::bit_ceil_constexpr(SmallSize);
   /// SmallStorage - Fixed size storage used in 'small mode'.
   const void *SmallStorage[SmallSizePowTwo];
 
diff --git a/llvm/include/llvm/ADT/bit.h b/llvm/include/llvm/ADT/bit.h
index 66c4f94..8b60b69 100644
--- a/llvm/include/llvm/ADT/bit.h
+++ b/llvm/include/llvm/ADT/bit.h
@@ -336,34 +336,44 @@ template <typename T> [[nodiscard]] T bit_ceil(T Value) {
   return T(1) << llvm::bit_width<T>(Value - 1u);
 }
 
-// Forward-declare rotr so that rotl can use it.
-template <typename T, typename = std::enable_if_t<std::is_unsigned_v<T>>>
-[[nodiscard]] constexpr T rotr(T V, int R);
+/// Returns the smallest integral power of two no smaller than Value if Value is
+/// nonzero.  Returns 1 otherwise.
+///
+/// Ex. bit_ceil(5) == 8.
+///
+/// The return value is undefined if the input is larger than the largest power
+/// of two representable in T.
+template <typename T> [[nodiscard]] constexpr T bit_ceil_constexpr(T Value) {
+  static_assert(std::is_unsigned_v<T>,
+                "Only unsigned integral types are allowed.");
+  if (Value < 2)
+    return 1;
+  return T(1) << llvm::bit_width_constexpr<T>(Value - 1u);
+}
 
 template <typename T, typename = std::enable_if_t<std::is_unsigned_v<T>>>
 [[nodiscard]] constexpr T rotl(T V, int R) {
-  unsigned N = std::numeric_limits<T>::digits;
+  constexpr unsigned N = std::numeric_limits<T>::digits;
 
-  R = R % N;
-  if (!R)
-    return V;
+  static_assert(has_single_bit(N), "& (N - 1) is only valid for powers of two");
+  R = R & (N - 1);
 
-  if (R < 0)
-    return llvm::rotr(V, -R);
+  if (R == 0)
+    return V;
 
   return (V << R) | (V >> (N - R));
 }
 
-template <typename T, typename> [[nodiscard]] constexpr T rotr(T V, int R) {
-  unsigned N = std::numeric_limits<T>::digits;
+template <typename T, typename = std::enable_if_t<std::is_unsigned_v<T>>>
+[[nodiscard]] constexpr T rotr(T V, int R) {
+  constexpr unsigned N = std::numeric_limits<T>::digits;
+
+  static_assert(has_single_bit(N), "& (N - 1) is only valid for powers of two");
+  R = R & (N - 1);
 
-  R = R % N;
-  if (!R)
+  if (R == 0)
     return V;
 
-  if (R < 0)
-    return llvm::rotl(V, -R);
-
   return (V >> R) | (V << (N - R));
 }
 
diff --git a/llvm/include/llvm/Support/Alignment.h b/llvm/include/llvm/Support/Alignment.h
index a4ca54e..f9d7c76 100644
--- a/llvm/include/llvm/Support/Alignment.h
+++ b/llvm/include/llvm/Support/Alignment.h
@@ -103,7 +103,7 @@ inline Align assumeAligned(uint64_t Value) {
 
 /// This struct is a compact representation of a valid (power of two) or
 /// undefined (0) alignment.
-struct MaybeAlign : public std::optional<Align> {
+struct MaybeAlign : std::optional<Align> {
 private:
   using UP = std::optional<Align>;
 
diff --git a/llvm/include/llvm/Support/Casting.h b/llvm/include/llvm/Support/Casting.h
index 2a9a149..6f6df2e 100644
--- a/llvm/include/llvm/Support/Casting.h
+++ b/llvm/include/llvm/Support/Casting.h
@@ -340,7 +340,7 @@ struct ValueFromPointerCast
 /// during the cast. It's also a good example of how to implement a move-only
 /// cast.
 template <typename To, typename From, typename Derived = void>
-struct UniquePtrCast : public CastIsPossible<To, From *> {
+struct UniquePtrCast : CastIsPossible<To, From *> {
   using Self = detail::SelfType<Derived, UniquePtrCast<To, From>>;
   using CastResultType = std::unique_ptr<
       std::remove_reference_t<typename cast_retty<To, From>::ret_type>>;
@@ -473,7 +473,7 @@ struct ForwardToPointerCast {
 // take advantage of the cast traits whenever possible!
 
 template <typename To, typename From, typename Enable = void>
-struct CastInfo : public CastIsPossible<To, From> {
+struct CastInfo : CastIsPossible<To, From> {
   using Self = CastInfo<To, From, Enable>;
 
   using CastReturnType = typename cast_retty<To, From>::ret_type;
@@ -536,8 +536,7 @@ struct CastInfo<To, std::unique_ptr<From>> : public UniquePtrCast<To, From> {};
 /// the input is std::optional<From> that the output can be std::optional<To>.
 /// If that's not the case, specialize CastInfo for your use case.
 template <typename To, typename From>
-struct CastInfo<To, std::optional<From>> : public OptionalValueCast<To, From> {
-};
+struct CastInfo<To, std::optional<From>> : OptionalValueCast<To, From> {};
 
 /// isa<X> - Return true if the parameter to the template is an instance of one
 /// of the template type arguments.  Used like this:
diff --git a/llvm/include/llvm/Support/CommandLine.h b/llvm/include/llvm/Support/CommandLine.h
index dd05c53..5a5f00e 100644
--- a/llvm/include/llvm/Support/CommandLine.h
+++ b/llvm/include/llvm/Support/CommandLine.h
@@ -549,7 +549,7 @@ template <class DataType> struct OptionValue;
 // The default value safely does nothing. Option value printing is only
 // best-effort.
 template <class DataType, bool isClass>
-struct OptionValueBase : public GenericOptionValue {
+struct OptionValueBase : GenericOptionValue {
   // Temporary storage for argument passing.
   using WrapperType = OptionValue<DataType>;
 
diff --git a/llvm/include/llvm/Support/DOTGraphTraits.h b/llvm/include/llvm/Support/DOTGraphTraits.h
index ffa9abe..3b9fe00 100644
--- a/llvm/include/llvm/Support/DOTGraphTraits.h
+++ b/llvm/include/llvm/Support/DOTGraphTraits.h
@@ -162,9 +162,8 @@ public:
 /// graphs are converted to 'dot' graphs.  When specializing, you may inherit
 /// from DefaultDOTGraphTraits if you don't need to override everything.
 ///
-template <typename Ty>
-struct DOTGraphTraits : public DefaultDOTGraphTraits {
-  DOTGraphTraits (bool simple=false) : DefaultDOTGraphTraits (simple) {}
+template <typename Ty> struct DOTGraphTraits : DefaultDOTGraphTraits {
+  using DefaultDOTGraphTraits::DefaultDOTGraphTraits;
 };
 
 } // End llvm namespace
diff --git a/llvm/include/llvm/Support/ELFAttributes.h b/llvm/include/llvm/Support/ELFAttributes.h
index 270246f..5771a84 100644
--- a/llvm/include/llvm/Support/ELFAttributes.h
+++ b/llvm/include/llvm/Support/ELFAttributes.h
@@ -48,8 +48,6 @@ struct SubsectionAndTagToTagName {
   StringRef SubsectionName;
   unsigned Tag;
   StringRef TagName;
-  SubsectionAndTagToTagName(StringRef SN, unsigned Tg, StringRef TN)
-      : SubsectionName(SN), Tag(Tg), TagName(TN) {}
 };
 
 namespace ELFAttrs {
diff --git a/llvm/include/llvm/Support/LSP/Protocol.h b/llvm/include/llvm/Support/LSP/Protocol.h
index 93b82f1..e38203a 100644
--- a/llvm/include/llvm/Support/LSP/Protocol.h
+++ b/llvm/include/llvm/Support/LSP/Protocol.h
@@ -449,7 +449,7 @@ struct ReferenceContext {
 bool fromJSON(const llvm::json::Value &value, ReferenceContext &result,
               llvm::json::Path path);
 
-struct ReferenceParams : public TextDocumentPositionParams {
+struct ReferenceParams : TextDocumentPositionParams {
   ReferenceContext context;
 };
 
diff --git a/llvm/include/llvm/Support/MD5.h b/llvm/include/llvm/Support/MD5.h
index ed29826..4ba3867 100644
--- a/llvm/include/llvm/Support/MD5.h
+++ b/llvm/include/llvm/Support/MD5.h
@@ -41,7 +41,7 @@ template <typename T> class ArrayRef;
 
 class MD5 {
 public:
-  struct MD5Result : public std::array<uint8_t, 16> {
+  struct MD5Result : std::array<uint8_t, 16> {
     LLVM_ABI SmallString<32> digest() const;
 
     uint64_t low() const {
diff --git a/llvm/include/llvm/Support/MathExtras.h b/llvm/include/llvm/Support/MathExtras.h
index c2716a9..41232335 100644
--- a/llvm/include/llvm/Support/MathExtras.h
+++ b/llvm/include/llvm/Support/MathExtras.h
@@ -13,6 +13,7 @@
 #ifndef LLVM_SUPPORT_MATHEXTRAS_H
 #define LLVM_SUPPORT_MATHEXTRAS_H
 
+#include "llvm/ADT/STLForwardCompat.h"
 #include "llvm/ADT/bit.h"
 #include "llvm/Support/Compiler.h"
 #include <cassert>
@@ -42,38 +43,28 @@ using common_sint =
 
 /// Mathematical constants.
 namespace numbers {
-// TODO: Track C++20 std::numbers.
 // clang-format off
-constexpr double e          = 0x1.5bf0a8b145769P+1, // (2.7182818284590452354) https://oeis.org/A001113
-                 egamma     = 0x1.2788cfc6fb619P-1, // (.57721566490153286061) https://oeis.org/A001620
-                 ln2        = 0x1.62e42fefa39efP-1, // (.69314718055994530942) https://oeis.org/A002162
-                 ln10       = 0x1.26bb1bbb55516P+1, // (2.3025850929940456840) https://oeis.org/A002392
-                 log2e      = 0x1.71547652b82feP+0, // (1.4426950408889634074)
-                 log10e     = 0x1.bcb7b1526e50eP-2, // (.43429448190325182765)
-                 pi         = 0x1.921fb54442d18P+1, // (3.1415926535897932385) https://oeis.org/A000796
-                 inv_pi     = 0x1.45f306dc9c883P-2, // (.31830988618379067154) https://oeis.org/A049541
-                 sqrtpi     = 0x1.c5bf891b4ef6bP+0, // (1.7724538509055160273) https://oeis.org/A002161
-                 inv_sqrtpi = 0x1.20dd750429b6dP-1, // (.56418958354775628695) https://oeis.org/A087197
-                 sqrt2      = 0x1.6a09e667f3bcdP+0, // (1.4142135623730950488) https://oeis.org/A00219
-                 inv_sqrt2  = 0x1.6a09e667f3bcdP-1, // (.70710678118654752440)
-                 sqrt3      = 0x1.bb67ae8584caaP+0, // (1.7320508075688772935) https://oeis.org/A002194
-                 inv_sqrt3  = 0x1.279a74590331cP-1, // (.57735026918962576451)
-                 phi        = 0x1.9e3779b97f4a8P+0; // (1.6180339887498948482) https://oeis.org/A001622
-constexpr float ef          = 0x1.5bf0a8P+1F, // (2.71828183) https://oeis.org/A001113
-                egammaf     = 0x1.2788d0P-1F, // (.577215665) https://oeis.org/A001620
-                ln2f        = 0x1.62e430P-1F, // (.693147181) https://oeis.org/A002162
-                ln10f       = 0x1.26bb1cP+1F, // (2.30258509) https://oeis.org/A002392
-                log2ef      = 0x1.715476P+0F, // (1.44269504)
-                log10ef     = 0x1.bcb7b2P-2F, // (.434294482)
-                pif         = 0x1.921fb6P+1F, // (3.14159265) https://oeis.org/A000796
-                inv_pif     = 0x1.45f306P-2F, // (.318309886) https://oeis.org/A049541
-                sqrtpif     = 0x1.c5bf8aP+0F, // (1.77245385) https://oeis.org/A002161
-                inv_sqrtpif = 0x1.20dd76P-1F, // (.564189584) https://oeis.org/A087197
-                sqrt2f      = 0x1.6a09e6P+0F, // (1.41421356) https://oeis.org/A002193
-                inv_sqrt2f  = 0x1.6a09e6P-1F, // (.707106781)
-                sqrt3f      = 0x1.bb67aeP+0F, // (1.73205081) https://oeis.org/A002194
-                inv_sqrt3f  = 0x1.279a74P-1F, // (.577350269)
-                phif        = 0x1.9e377aP+0F; // (1.61803399) https://oeis.org/A001622
+inline constexpr float ef          = e_v<float>;
+inline constexpr float egammaf     = egamma_v<float>;
+inline constexpr float ln2f        = ln2_v<float>;
+inline constexpr float ln10f       = ln10_v<float>;
+inline constexpr float log2ef      = log2e_v<float>;
+inline constexpr float log10ef     = log10e_v<float>;
+inline constexpr float pif         = pi_v<float>;
+inline constexpr float inv_pif     = inv_pi_v<float>;
+inline constexpr float inv_sqrtpif = inv_sqrtpi_v<float>;
+inline constexpr float sqrt2f      = sqrt2_v<float>;
+inline constexpr float inv_sqrt2f  = inv_sqrt2_v<float>;
+inline constexpr float sqrt3f      = sqrt3_v<float>;
+inline constexpr float inv_sqrt3f  = inv_sqrt3_v<float>;
+inline constexpr float phif        = phi_v<float>;
+
+// sqrtpi is not in C++20 std::numbers.
+template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>>
+inline constexpr T sqrtpi_v = T(0x1.c5bf891b4ef6bP+0); // (1.7724538509055160273) https://oeis.org/A002161
+inline constexpr double sqrtpi  = sqrtpi_v<double>;
+inline constexpr float  sqrtpif = sqrtpi_v<float>;
+
 // These string literals are taken from below:
 // https://github.com/bminor/glibc/blob/8543577b04ded6d979ffcc5a818930e4d74d0645/math/math.h#L1215-L1229
 constexpr const char *pis     = "3.141592653589793238462643383279502884",
diff --git a/llvm/include/llvm/Support/Timer.h b/llvm/include/llvm/Support/Timer.h
index 40709d4..a4ed712 100644
--- a/llvm/include/llvm/Support/Timer.h
+++ b/llvm/include/llvm/Support/Timer.h
@@ -167,7 +167,7 @@ public:
 /// you to declare a new timer, AND specify the region to time, all in one
 /// statement.  All timers with the same name are merged.  This is primarily
 /// used for debugging and for hunting performance problems.
-struct NamedRegionTimer : public TimeRegion {
+struct NamedRegionTimer : TimeRegion {
   LLVM_ABI explicit NamedRegionTimer(StringRef Name, StringRef Description,
                                      StringRef GroupName,
                                      StringRef GroupDescription,
diff --git a/llvm/lib/Analysis/LazyValueInfo.cpp b/llvm/lib/Analysis/LazyValueInfo.cpp
index 0e5bc48..df75999 100644
--- a/llvm/lib/Analysis/LazyValueInfo.cpp
+++ b/llvm/lib/Analysis/LazyValueInfo.cpp
@@ -947,9 +947,8 @@ LazyValueInfoImpl::solveBlockValueSelect(SelectInst *SI, BasicBlock *BB) {
                                                   /*UseBlockValue*/ false));
   }
 
-  ValueLatticeElement Result = TrueVal;
-  Result.mergeIn(FalseVal);
-  return Result;
+  TrueVal.mergeIn(FalseVal);
+  return TrueVal;
 }
 
 std::optional<ConstantRange>
@@ -1778,9 +1777,8 @@ ValueLatticeElement LazyValueInfoImpl::getValueInBlock(Value *V, BasicBlock *BB,
     assert(OptResult && "Value not available after solving");
   }
 
-  ValueLatticeElement Result = *OptResult;
-  LLVM_DEBUG(dbgs() << "  Result = " << Result << "\n");
-  return Result;
+  LLVM_DEBUG(dbgs() << "  Result = " << *OptResult << "\n");
+  return *OptResult;
 }
 
 ValueLatticeElement LazyValueInfoImpl::getValueAt(Value *V, Instruction *CxtI) {
diff --git a/llvm/lib/Support/APFloat.cpp b/llvm/lib/Support/APFloat.cpp
index 79b5c15..b4de79a 100644
--- a/llvm/lib/Support/APFloat.cpp
+++ b/llvm/lib/Support/APFloat.cpp
@@ -6165,6 +6165,70 @@ float APFloat::convertToFloat() const {
   return Temp.getIEEE().convertToFloat();
 }
 
+APFloat::Storage::~Storage() {
+  if (usesLayout<IEEEFloat>(*semantics)) {
+    IEEE.~IEEEFloat();
+    return;
+  }
+  if (usesLayout<DoubleAPFloat>(*semantics)) {
+    Double.~DoubleAPFloat();
+    return;
+  }
+  llvm_unreachable("Unexpected semantics");
+}
+
+APFloat::Storage::Storage(const APFloat::Storage &RHS) {
+  if (usesLayout<IEEEFloat>(*RHS.semantics)) {
+    new (this) IEEEFloat(RHS.IEEE);
+    return;
+  }
+  if (usesLayout<DoubleAPFloat>(*RHS.semantics)) {
+    new (this) DoubleAPFloat(RHS.Double);
+    return;
+  }
+  llvm_unreachable("Unexpected semantics");
+}
+
+APFloat::Storage::Storage(APFloat::Storage &&RHS) {
+  if (usesLayout<IEEEFloat>(*RHS.semantics)) {
+    new (this) IEEEFloat(std::move(RHS.IEEE));
+    return;
+  }
+  if (usesLayout<DoubleAPFloat>(*RHS.semantics)) {
+    new (this) DoubleAPFloat(std::move(RHS.Double));
+    return;
+  }
+  llvm_unreachable("Unexpected semantics");
+}
+
+APFloat::Storage &APFloat::Storage::operator=(const APFloat::Storage &RHS) {
+  if (usesLayout<IEEEFloat>(*semantics) &&
+      usesLayout<IEEEFloat>(*RHS.semantics)) {
+    IEEE = RHS.IEEE;
+  } else if (usesLayout<DoubleAPFloat>(*semantics) &&
+             usesLayout<DoubleAPFloat>(*RHS.semantics)) {
+    Double = RHS.Double;
+  } else if (this != &RHS) {
+    this->~Storage();
+    new (this) Storage(RHS);
+  }
+  return *this;
+}
+
+APFloat::Storage &APFloat::Storage::operator=(APFloat::Storage &&RHS) {
+  if (usesLayout<IEEEFloat>(*semantics) &&
+      usesLayout<IEEEFloat>(*RHS.semantics)) {
+    IEEE = std::move(RHS.IEEE);
+  } else if (usesLayout<DoubleAPFloat>(*semantics) &&
+             usesLayout<DoubleAPFloat>(*RHS.semantics)) {
+    Double = std::move(RHS.Double);
+  } else if (this != &RHS) {
+    this->~Storage();
+    new (this) Storage(std::move(RHS));
+  }
+  return *this;
+}
+
 } // namespace llvm
 
 #undef APFLOAT_DISPATCH_ON_SEMANTICS
diff --git a/llvm/lib/Target/AMDGPU/GCNRegPressure.h b/llvm/lib/Target/AMDGPU/GCNRegPressure.h
index 979a8b0..4b22c68 100644
--- a/llvm/lib/Target/AMDGPU/GCNRegPressure.h
+++ b/llvm/lib/Target/AMDGPU/GCNRegPressure.h
@@ -21,6 +21,7 @@
 #include "llvm/CodeGen/LiveIntervals.h"
 #include "llvm/CodeGen/RegisterPressure.h"
 #include <algorithm>
+#include <array>
 
 namespace llvm {
 
@@ -45,7 +46,7 @@ struct GCNRegPressure {
     return !Value[SGPR] && !Value[VGPR] && !Value[AGPR] && !Value[AVGPR];
   }
 
-  void clear() { std::fill(&Value[0], &Value[ValueArraySize], 0); }
+  void clear() { Value.fill(0); }
 
   unsigned getNumRegs(RegKind Kind) const {
     assert(Kind < TOTAL_KINDS);
@@ -127,9 +128,7 @@ struct GCNRegPressure {
   bool less(const MachineFunction &MF, const GCNRegPressure &O,
             unsigned MaxOccupancy = std::numeric_limits<unsigned>::max()) const;
 
-  bool operator==(const GCNRegPressure &O) const {
-    return std::equal(&Value[0], &Value[ValueArraySize], O.Value);
-  }
+  bool operator==(const GCNRegPressure &O) const { return Value == O.Value; }
 
   bool operator!=(const GCNRegPressure &O) const {
     return !(*this == O);
@@ -160,7 +159,7 @@ private:
 
   /// Pressure for all register kinds (first all regular registers kinds, then
   /// all tuple register kinds).
-  unsigned Value[ValueArraySize];
+  std::array<unsigned, ValueArraySize> Value;
 
   static unsigned getRegKind(const TargetRegisterClass *RC,
                              const SIRegisterInfo *STI);
diff --git a/llvm/lib/Target/AMDGPU/R600ISelLowering.cpp b/llvm/lib/Target/AMDGPU/R600ISelLowering.cpp
index 2aa54c9..09ef6ac 100644
--- a/llvm/lib/Target/AMDGPU/R600ISelLowering.cpp
+++ b/llvm/lib/Target/AMDGPU/R600ISelLowering.cpp
@@ -45,6 +45,9 @@ R600TargetLowering::R600TargetLowering(const TargetMachine &TM,
   // Legalize loads and stores to the private address space.
   setOperationAction(ISD::LOAD, {MVT::i32, MVT::v2i32, MVT::v4i32}, Custom);
 
+  // 32-bit ABS is legal for AMDGPU except for R600
+  setOperationAction(ISD::ABS, MVT::i32, Expand);
+
   // EXTLOAD should be the same as ZEXTLOAD. It is legal for some address
   // spaces, so it is custom lowered to handle those where it isn't.
   for (auto Op : {ISD::SEXTLOAD, ISD::ZEXTLOAD, ISD::EXTLOAD})
diff --git a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
index 942e784..50447f4 100644
--- a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
@@ -10626,59 +10626,6 @@ bool SIInstrInfo::optimizeCompareInstr(MachineInstr &CmpInstr, Register SrcReg,
   if (SrcReg2 && !getFoldableImm(SrcReg2, *MRI, CmpValue))
     return false;
 
-  const auto optimizeCmpSelect = [&CmpInstr, SrcReg, CmpValue, MRI,
-                                  this]() -> bool {
-    if (CmpValue != 0)
-      return false;
-
-    MachineInstr *Def = MRI->getUniqueVRegDef(SrcReg);
-    if (!Def || Def->getParent() != CmpInstr.getParent())
-      return false;
-
-    bool CanOptimize = false;
-
-    // For S_OP that set SCC = DST!=0, do the transformation
-    //
-    //   s_cmp_lg_* (S_OP ...), 0 => (S_OP ...)
-    if (setsSCCifResultIsNonZero(*Def))
-      CanOptimize = true;
-
-    // s_cmp_lg_* is redundant because the SCC input value for S_CSELECT* has
-    // the same value that will be calculated by s_cmp_lg_*
-    //
-    //   s_cmp_lg_* (S_CSELECT* (non-zero imm), 0), 0 => (S_CSELECT* (non-zero
-    //   imm), 0)
-    if (Def->getOpcode() == AMDGPU::S_CSELECT_B32 ||
-        Def->getOpcode() == AMDGPU::S_CSELECT_B64) {
-      bool Op1IsNonZeroImm =
-          Def->getOperand(1).isImm() && Def->getOperand(1).getImm() != 0;
-      bool Op2IsZeroImm =
-          Def->getOperand(2).isImm() && Def->getOperand(2).getImm() == 0;
-      if (Op1IsNonZeroImm && Op2IsZeroImm)
-        CanOptimize = true;
-    }
-
-    if (!CanOptimize)
-      return false;
-
-    MachineInstr *KillsSCC = nullptr;
-    for (MachineInstr &MI :
-         make_range(std::next(Def->getIterator()), CmpInstr.getIterator())) {
-      if (MI.modifiesRegister(AMDGPU::SCC, &RI))
-        return false;
-      if (MI.killsRegister(AMDGPU::SCC, &RI))
-        KillsSCC = &MI;
-    }
-
-    if (MachineOperand *SccDef =
-            Def->findRegisterDefOperand(AMDGPU::SCC, /*TRI=*/nullptr))
-      SccDef->setIsDead(false);
-    if (KillsSCC)
-      KillsSCC->clearRegisterKills(AMDGPU::SCC, /*TRI=*/nullptr);
-    CmpInstr.eraseFromParent();
-    return true;
-  };
-
   const auto optimizeCmpAnd = [&CmpInstr, SrcReg, CmpValue, MRI,
                                this](int64_t ExpectedValue, unsigned SrcSize,
                                      bool IsReversible, bool IsSigned) -> bool {
@@ -10753,20 +10700,16 @@ bool SIInstrInfo::optimizeCompareInstr(MachineInstr &CmpInstr, Register SrcReg,
     if (IsReversedCC && !MRI->hasOneNonDBGUse(DefReg))
       return false;
 
-    MachineInstr *KillsSCC = nullptr;
-    for (MachineInstr &MI :
-         make_range(std::next(Def->getIterator()), CmpInstr.getIterator())) {
-      if (MI.modifiesRegister(AMDGPU::SCC, &RI))
+    for (auto I = std::next(Def->getIterator()), E = CmpInstr.getIterator();
+         I != E; ++I) {
+      if (I->modifiesRegister(AMDGPU::SCC, &RI) ||
+          I->killsRegister(AMDGPU::SCC, &RI))
         return false;
-      if (MI.killsRegister(AMDGPU::SCC, &RI))
-        KillsSCC = &MI;
     }
 
     MachineOperand *SccDef =
         Def->findRegisterDefOperand(AMDGPU::SCC, /*TRI=*/nullptr);
     SccDef->setIsDead(false);
-    if (KillsSCC)
-      KillsSCC->clearRegisterKills(AMDGPU::SCC, /*TRI=*/nullptr);
     CmpInstr.eraseFromParent();
 
     if (!MRI->use_nodbg_empty(DefReg)) {
@@ -10810,7 +10753,7 @@ bool SIInstrInfo::optimizeCompareInstr(MachineInstr &CmpInstr, Register SrcReg,
   case AMDGPU::S_CMP_LG_I32:
   case AMDGPU::S_CMPK_LG_U32:
   case AMDGPU::S_CMPK_LG_I32:
-    return optimizeCmpAnd(0, 32, true, false) || optimizeCmpSelect();
+    return optimizeCmpAnd(0, 32, true, false);
   case AMDGPU::S_CMP_GT_U32:
   case AMDGPU::S_CMPK_GT_U32:
     return optimizeCmpAnd(0, 32, false, false);
@@ -10818,7 +10761,7 @@ bool SIInstrInfo::optimizeCompareInstr(MachineInstr &CmpInstr, Register SrcReg,
   case AMDGPU::S_CMPK_GT_I32:
     return optimizeCmpAnd(0, 32, false, true);
   case AMDGPU::S_CMP_LG_U64:
-    return optimizeCmpAnd(0, 64, true, false) || optimizeCmpSelect();
+    return optimizeCmpAnd(0, 64, true, false);
   }
 
   return false;
diff --git a/llvm/lib/Target/AMDGPU/SIInstrInfo.h b/llvm/lib/Target/AMDGPU/SIInstrInfo.h
index ee99a74..df27ec1 100644
--- a/llvm/lib/Target/AMDGPU/SIInstrInfo.h
+++ b/llvm/lib/Target/AMDGPU/SIInstrInfo.h
@@ -709,30 +709,6 @@ public:
     }
   }
 
-  static bool setsSCCifResultIsNonZero(const MachineInstr &MI) {
-    if (!MI.findRegisterDefOperand(AMDGPU::SCC, /*TRI=*/nullptr))
-      return false;
-    // Compares have no result
-    if (MI.isCompare())
-      return false;
-    switch (MI.getOpcode()) {
-    default:
-      return true;
-    case AMDGPU::S_ADD_I32:
-    case AMDGPU::S_ADD_U32:
-    case AMDGPU::S_ADDC_U32:
-    case AMDGPU::S_SUB_I32:
-    case AMDGPU::S_SUB_U32:
-    case AMDGPU::S_SUBB_U32:
-    case AMDGPU::S_MIN_I32:
-    case AMDGPU::S_MIN_U32:
-    case AMDGPU::S_MAX_I32:
-    case AMDGPU::S_MAX_U32:
-    case AMDGPU::S_ADDK_I32:
-      return false;
-    }
-  }
-
   static bool isEXP(const MachineInstr &MI) {
     return MI.getDesc().TSFlags & SIInstrFlags::EXP;
   }
diff --git a/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp b/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
index 1b7cb9b..636e31c 100644
--- a/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
+++ b/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
@@ -699,7 +699,8 @@ public:
            "Can't encode VTYPE for uninitialized or unknown");
     if (TWiden != 0)
       return RISCVVType::encodeXSfmmVType(SEW, TWiden, AltFmt);
-    return RISCVVType::encodeVTYPE(VLMul, SEW, TailAgnostic, MaskAgnostic);
+    return RISCVVType::encodeVTYPE(VLMul, SEW, TailAgnostic, MaskAgnostic,
+                                   AltFmt);
   }
 
   bool hasSEWLMULRatioOnly() const { return SEWLMULRatioOnly; }
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfo.cpp b/llvm/lib/Target/RISCV/RISCVInstrInfo.cpp
index ddb53a2..12f776b 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfo.cpp
@@ -3775,11 +3775,13 @@ std::string RISCVInstrInfo::createMIROperandComment(
 
 #define CASE_VFMA_OPCODE_VV(OP)                                                \
   CASE_VFMA_OPCODE_LMULS_MF4(OP, VV, E16):                                     \
+  case CASE_VFMA_OPCODE_LMULS_MF4(OP##_ALT, VV, E16):                          \
   case CASE_VFMA_OPCODE_LMULS_MF2(OP, VV, E32):                                \
   case CASE_VFMA_OPCODE_LMULS_M1(OP, VV, E64)
 
 #define CASE_VFMA_SPLATS(OP)                                                   \
   CASE_VFMA_OPCODE_LMULS_MF4(OP, VFPR16, E16):                                 \
+  case CASE_VFMA_OPCODE_LMULS_MF4(OP##_ALT, VFPR16, E16):                      \
   case CASE_VFMA_OPCODE_LMULS_MF2(OP, VFPR32, E32):                            \
   case CASE_VFMA_OPCODE_LMULS_M1(OP, VFPR64, E64)
 // clang-format on
@@ -4003,11 +4005,13 @@ bool RISCVInstrInfo::findCommutedOpIndices(const MachineInstr &MI,
 
 #define CASE_VFMA_CHANGE_OPCODE_VV(OLDOP, NEWOP)                               \
   CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(OLDOP, NEWOP, VV, E16)                     \
+  CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(OLDOP##_ALT, NEWOP##_ALT, VV, E16)         \
   CASE_VFMA_CHANGE_OPCODE_LMULS_MF2(OLDOP, NEWOP, VV, E32)                     \
   CASE_VFMA_CHANGE_OPCODE_LMULS_M1(OLDOP, NEWOP, VV, E64)
 
 #define CASE_VFMA_CHANGE_OPCODE_SPLATS(OLDOP, NEWOP)                           \
   CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(OLDOP, NEWOP, VFPR16, E16)                 \
+  CASE_VFMA_CHANGE_OPCODE_LMULS_MF4(OLDOP##_ALT, NEWOP##_ALT, VFPR16, E16)     \
   CASE_VFMA_CHANGE_OPCODE_LMULS_MF2(OLDOP, NEWOP, VFPR32, E32)                 \
   CASE_VFMA_CHANGE_OPCODE_LMULS_M1(OLDOP, NEWOP, VFPR64, E64)
 // clang-format on
@@ -4469,6 +4473,20 @@ bool RISCVInstrInfo::simplifyInstruction(MachineInstr &MI) const {
   CASE_FP_WIDEOP_CHANGE_OPCODE_COMMON(OP, M2, E32)                             \
   CASE_FP_WIDEOP_CHANGE_OPCODE_COMMON(OP, M4, E16)                             \
   CASE_FP_WIDEOP_CHANGE_OPCODE_COMMON(OP, M4, E32)                             \
+
+#define CASE_FP_WIDEOP_OPCODE_LMULS_ALT(OP)                                    \
+  CASE_FP_WIDEOP_OPCODE_COMMON(OP, MF4, E16):                                  \
+  case CASE_FP_WIDEOP_OPCODE_COMMON(OP, MF2, E16):                             \
+  case CASE_FP_WIDEOP_OPCODE_COMMON(OP, M1, E16):                              \
+  case CASE_FP_WIDEOP_OPCODE_COMMON(OP, M2, E16):                              \
+  case CASE_FP_WIDEOP_OPCODE_COMMON(OP, M4, E16)
+
+#define CASE_FP_WIDEOP_CHANGE_OPCODE_LMULS_ALT(OP)                             \
+  CASE_FP_WIDEOP_CHANGE_OPCODE_COMMON(OP, MF4, E16)                            \
+  CASE_FP_WIDEOP_CHANGE_OPCODE_COMMON(OP, MF2, E16)                            \
+  CASE_FP_WIDEOP_CHANGE_OPCODE_COMMON(OP, M1, E16)                             \
+  CASE_FP_WIDEOP_CHANGE_OPCODE_COMMON(OP, M2, E16)                             \
+  CASE_FP_WIDEOP_CHANGE_OPCODE_COMMON(OP, M4, E16)
 // clang-format on
 
 MachineInstr *RISCVInstrInfo::convertToThreeAddress(MachineInstr &MI,
@@ -4478,6 +4496,8 @@ MachineInstr *RISCVInstrInfo::convertToThreeAddress(MachineInstr &MI,
   switch (MI.getOpcode()) {
   default:
     return nullptr;
+  case CASE_FP_WIDEOP_OPCODE_LMULS_ALT(FWADD_ALT_WV):
+  case CASE_FP_WIDEOP_OPCODE_LMULS_ALT(FWSUB_ALT_WV):
   case CASE_FP_WIDEOP_OPCODE_LMULS(FWADD_WV):
   case CASE_FP_WIDEOP_OPCODE_LMULS(FWSUB_WV): {
     assert(RISCVII::hasVecPolicyOp(MI.getDesc().TSFlags) &&
@@ -4494,6 +4514,8 @@ MachineInstr *RISCVInstrInfo::convertToThreeAddress(MachineInstr &MI,
       llvm_unreachable("Unexpected opcode");
     CASE_FP_WIDEOP_CHANGE_OPCODE_LMULS(FWADD_WV)
     CASE_FP_WIDEOP_CHANGE_OPCODE_LMULS(FWSUB_WV)
+    CASE_FP_WIDEOP_CHANGE_OPCODE_LMULS_ALT(FWADD_ALT_WV)
+    CASE_FP_WIDEOP_CHANGE_OPCODE_LMULS_ALT(FWSUB_ALT_WV)
     }
     // clang-format on
 
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoZvfbf.td b/llvm/lib/Target/RISCV/RISCVInstrInfoZvfbf.td
index c9c1246..9358486 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoZvfbf.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoZvfbf.td
@@ -44,6 +44,336 @@ let Predicates = [HasStdExtZvfbfmin] in {
 let mayRaiseFPException = true, Predicates = [HasStdExtZvfbfwma] in
   defm PseudoVFWMACCBF16 : VPseudoVWMAC_VV_VF_BF_RM;
 
+defset list<VTypeInfoToWide> AllWidenableIntToBF16Vectors = {
+  def : VTypeInfoToWide<VI8MF8, VBF16MF4>;
+  def : VTypeInfoToWide<VI8MF4, VBF16MF2>;
+  def : VTypeInfoToWide<VI8MF2, VBF16M1>;
+  def : VTypeInfoToWide<VI8M1, VBF16M2>;
+  def : VTypeInfoToWide<VI8M2, VBF16M4>;
+  def : VTypeInfoToWide<VI8M4, VBF16M8>;
+}
+
+multiclass VPseudoVALU_VV_VF_RM_BF16 {
+  foreach m = MxListF in {
+    defm "" : VPseudoBinaryFV_VV_RM<m, 16/*sew*/>,
+              SchedBinary<"WriteVFALUV", "ReadVFALUV", "ReadVFALUV", m.MX,
+                          16/*sew*/, forcePassthruRead=true>;
+  }
+
+  defvar f = SCALAR_F16;
+  foreach m = f.MxList in {
+    defm "" : VPseudoBinaryV_VF_RM<m, f, f.SEW>,
+              SchedBinary<"WriteVFALUF", "ReadVFALUV", "ReadVFALUF", m.MX,
+                          f.SEW, forcePassthruRead=true>;
+  }
+}
+
+multiclass VPseudoVALU_VF_RM_BF16 {
+  defvar f = SCALAR_F16;
+  foreach m = f.MxList in {
+    defm "" : VPseudoBinaryV_VF_RM<m, f, f.SEW>,
+              SchedBinary<"WriteVFALUF", "ReadVFALUV", "ReadVFALUF", m.MX,
+                          f.SEW, forcePassthruRead=true>;
+  }
+}
+
+multiclass VPseudoVFWALU_VV_VF_RM_BF16 {
+  foreach m = MxListFW in {
+    defm "" : VPseudoBinaryW_VV_RM<m, sew=16>,
+              SchedBinary<"WriteVFWALUV", "ReadVFWALUV", "ReadVFWALUV", m.MX,
+                          16/*sew*/, forcePassthruRead=true>;
+  }
+
+  defvar f = SCALAR_F16;
+  foreach m = f.MxListFW in {
+    defm "" : VPseudoBinaryW_VF_RM<m, f, sew=f.SEW>,
+              SchedBinary<"WriteVFWALUF", "ReadVFWALUV", "ReadVFWALUF", m.MX,
+                        f.SEW, forcePassthruRead=true>;
+  }
+}
+
+multiclass VPseudoVFWALU_WV_WF_RM_BF16 {
+  foreach m = MxListFW in {
+    defm "" : VPseudoBinaryW_WV_RM<m, sew=16>,
+              SchedBinary<"WriteVFWALUV", "ReadVFWALUV", "ReadVFWALUV", m.MX,
+                          16/*sew*/, forcePassthruRead=true>;
+  }
+  defvar f = SCALAR_F16;
+  foreach m = f.MxListFW in {
+    defm "" : VPseudoBinaryW_WF_RM<m, f, sew=f.SEW>,
+              SchedBinary<"WriteVFWALUF", "ReadVFWALUV", "ReadVFWALUF", m.MX,
+                          f.SEW, forcePassthruRead=true>;
+  }
+}
+
+multiclass VPseudoVFMUL_VV_VF_RM_BF16 {
+  foreach m = MxListF in {
+    defm "" : VPseudoBinaryFV_VV_RM<m, 16/*sew*/>,
+              SchedBinary<"WriteVFMulV", "ReadVFMulV", "ReadVFMulV", m.MX,
+                          16/*sew*/, forcePassthruRead=true>;
+  }
+
+  defvar f = SCALAR_F16;
+  foreach m = f.MxList in {
+    defm "" : VPseudoBinaryV_VF_RM<m, f, f.SEW>,
+              SchedBinary<"WriteVFMulF", "ReadVFMulV", "ReadVFMulF", m.MX,
+                          f.SEW, forcePassthruRead=true>;
+  }
+}
+
+multiclass VPseudoVWMUL_VV_VF_RM_BF16 {
+  foreach m = MxListFW in {
+    defm "" : VPseudoBinaryW_VV_RM<m, sew=16>,
+              SchedBinary<"WriteVFWMulV", "ReadVFWMulV", "ReadVFWMulV", m.MX,
+                          16/*sew*/, forcePassthruRead=true>;
+  }
+
+  defvar f = SCALAR_F16;
+  foreach m = f.MxListFW in {
+    defm "" : VPseudoBinaryW_VF_RM<m, f, sew=f.SEW>,
+              SchedBinary<"WriteVFWMulF", "ReadVFWMulV", "ReadVFWMulF", m.MX,
+                          f.SEW, forcePassthruRead=true>;
+  }
+}
+
+multiclass VPseudoVMAC_VV_VF_AAXA_RM_BF16 {
+  foreach m = MxListF in {
+    defm "" : VPseudoTernaryV_VV_AAXA_RM<m, 16/*sew*/>,
+              SchedTernary<"WriteVFMulAddV", "ReadVFMulAddV", "ReadVFMulAddV", 
+                           "ReadVFMulAddV", m.MX, 16/*sew*/>;
+  }
+
+  defvar f = SCALAR_F16;
+  foreach m = f.MxList in {
+    defm "" : VPseudoTernaryV_VF_AAXA_RM<m, f, f.SEW>,
+              SchedTernary<"WriteVFMulAddF", "ReadVFMulAddV", "ReadVFMulAddF", 
+                           "ReadVFMulAddV", m.MX, f.SEW>;
+  }
+}
+
+multiclass VPseudoVWMAC_VV_VF_RM_BF16 {
+  foreach m = MxListFW in {
+    defm "" : VPseudoTernaryW_VV_RM<m, sew=16>,
+              SchedTernary<"WriteVFWMulAddV", "ReadVFWMulAddV",
+                           "ReadVFWMulAddV", "ReadVFWMulAddV", m.MX, 16/*sew*/>;
+  }
+
+  defvar f = SCALAR_F16;
+  foreach m = f.MxListFW in {
+    defm "" : VPseudoTernaryW_VF_RM<m, f, sew=f.SEW>,
+              SchedTernary<"WriteVFWMulAddF", "ReadVFWMulAddV",
+                           "ReadVFWMulAddF", "ReadVFWMulAddV", m.MX, f.SEW>;
+  }
+}
+
+multiclass VPseudoVRCP_V_BF16 {
+  foreach m = MxListF in {
+    defvar mx = m.MX;
+    let VLMul = m.value in {
+      def "_V_" # mx # "_E16"
+          : VPseudoUnaryNoMask<m.vrclass, m.vrclass>,
+            SchedUnary<"WriteVFRecpV", "ReadVFRecpV", mx, 16/*sew*/,
+                       forcePassthruRead=true>;
+      def "_V_" # mx # "_E16_MASK"
+          : VPseudoUnaryMask<m.vrclass, m.vrclass>,
+            RISCVMaskedPseudo<MaskIdx = 2>,
+            SchedUnary<"WriteVFRecpV", "ReadVFRecpV", mx, 16/*sew*/,
+                       forcePassthruRead=true>;
+    }
+  }
+}
+
+multiclass VPseudoVRCP_V_RM_BF16 {
+  foreach m = MxListF in {
+    defvar mx = m.MX;
+    let VLMul = m.value in {
+      def "_V_" # mx # "_E16"
+          : VPseudoUnaryNoMaskRoundingMode<m.vrclass, m.vrclass>,
+            SchedUnary<"WriteVFRecpV", "ReadVFRecpV", mx, 16/*sew*/,
+                       forcePassthruRead=true>;
+      def "_V_" # mx # "_E16_MASK"
+          : VPseudoUnaryMaskRoundingMode<m.vrclass, m.vrclass>,
+            RISCVMaskedPseudo<MaskIdx = 2>,
+            SchedUnary<"WriteVFRecpV", "ReadVFRecpV", mx, 16/*sew*/,
+                       forcePassthruRead=true>;
+    }
+  }
+}
+
+multiclass VPseudoVMAX_VV_VF_BF16 {
+  foreach m = MxListF in {
+    defm "" : VPseudoBinaryV_VV<m, sew=16>,
+              SchedBinary<"WriteVFMinMaxV", "ReadVFMinMaxV", "ReadVFMinMaxV", 
+                          m.MX, 16/*sew*/, forcePassthruRead=true>;
+  }
+
+  defvar f = SCALAR_F16;
+  foreach m = f.MxList in {
+    defm "" : VPseudoBinaryV_VF<m, f, f.SEW>,
+              SchedBinary<"WriteVFMinMaxF", "ReadVFMinMaxV", "ReadVFMinMaxF", 
+                          m.MX, f.SEW, forcePassthruRead=true>;
+  }
+}
+
+multiclass VPseudoVSGNJ_VV_VF_BF16 {
+  foreach m = MxListF in {
+    defm "" : VPseudoBinaryV_VV<m, sew=16>,
+              SchedBinary<"WriteVFSgnjV", "ReadVFSgnjV", "ReadVFSgnjV", m.MX,
+                          16/*sew*/, forcePassthruRead=true>;
+  }
+
+  defvar f = SCALAR_F16;
+  foreach m = f.MxList in {
+    defm "" : VPseudoBinaryV_VF<m, f, f.SEW>,
+              SchedBinary<"WriteVFSgnjF", "ReadVFSgnjV", "ReadVFSgnjF", m.MX,
+                          f.SEW, forcePassthruRead=true>;
+  }
+}
+
+multiclass VPseudoVWCVTF_V_BF16 {
+  defvar constraint = "@earlyclobber $rd";
+  foreach m = MxListW in
+    defm _V : VPseudoConversion<m.wvrclass, m.vrclass, m, constraint, sew=8,
+                                TargetConstraintType=3>,
+              SchedUnary<"WriteVFWCvtIToFV", "ReadVFWCvtIToFV", m.MX, 8/*sew*/,
+                         forcePassthruRead=true>;
+}
+
+multiclass VPseudoVWCVTD_V_BF16 {
+  defvar constraint = "@earlyclobber $rd";
+  foreach m = MxListFW in
+    defm _V : VPseudoConversion<m.wvrclass, m.vrclass, m, constraint, sew=16,
+                                TargetConstraintType=3>,
+              SchedUnary<"WriteVFWCvtFToFV", "ReadVFWCvtFToFV", m.MX, 16/*sew*/,
+                         forcePassthruRead=true>;
+}
+
+multiclass VPseudoVNCVTD_W_BF16 {
+  defvar constraint = "@earlyclobber $rd";
+  foreach m = MxListFW in
+    defm _W : VPseudoConversion<m.vrclass, m.wvrclass, m, constraint, sew=16,
+                                TargetConstraintType=2>,
+              SchedUnary<"WriteVFNCvtFToFV", "ReadVFNCvtFToFV", m.MX, 16/*sew*/,
+                         forcePassthruRead=true>;
+}
+
+multiclass VPseudoVNCVTD_W_RM_BF16 {
+  defvar constraint = "@earlyclobber $rd";
+  foreach m = MxListFW in
+    defm _W : VPseudoConversionRoundingMode<m.vrclass, m.wvrclass, m,
+                                            constraint, sew=16,
+                                            TargetConstraintType=2>,
+              SchedUnary<"WriteVFNCvtFToFV", "ReadVFNCvtFToFV", m.MX, 16/*sew*/,
+                         forcePassthruRead=true>;
+}
+
+let Predicates = [HasStdExtZvfbfa], AltFmtType = IS_ALTFMT in {
+let mayRaiseFPException = true in {
+defm PseudoVFADD_ALT : VPseudoVALU_VV_VF_RM_BF16;
+defm PseudoVFSUB_ALT  : VPseudoVALU_VV_VF_RM_BF16;
+defm PseudoVFRSUB_ALT : VPseudoVALU_VF_RM_BF16;
+}
+
+let mayRaiseFPException = true in {
+defm PseudoVFWADD_ALT : VPseudoVFWALU_VV_VF_RM_BF16;
+defm PseudoVFWSUB_ALT : VPseudoVFWALU_VV_VF_RM_BF16;
+defm PseudoVFWADD_ALT : VPseudoVFWALU_WV_WF_RM_BF16;
+defm PseudoVFWSUB_ALT : VPseudoVFWALU_WV_WF_RM_BF16;
+}
+
+let mayRaiseFPException = true in
+defm PseudoVFMUL_ALT : VPseudoVFMUL_VV_VF_RM_BF16;
+
+let mayRaiseFPException = true in
+defm PseudoVFWMUL_ALT : VPseudoVWMUL_VV_VF_RM_BF16;
+
+let mayRaiseFPException = true in {
+defm PseudoVFMACC_ALT  : VPseudoVMAC_VV_VF_AAXA_RM_BF16;
+defm PseudoVFNMACC_ALT : VPseudoVMAC_VV_VF_AAXA_RM_BF16;
+defm PseudoVFMSAC_ALT  : VPseudoVMAC_VV_VF_AAXA_RM_BF16;
+defm PseudoVFNMSAC_ALT : VPseudoVMAC_VV_VF_AAXA_RM_BF16;
+defm PseudoVFMADD_ALT  : VPseudoVMAC_VV_VF_AAXA_RM_BF16;
+defm PseudoVFNMADD_ALT : VPseudoVMAC_VV_VF_AAXA_RM_BF16;
+defm PseudoVFMSUB_ALT  : VPseudoVMAC_VV_VF_AAXA_RM_BF16;
+defm PseudoVFNMSUB_ALT : VPseudoVMAC_VV_VF_AAXA_RM_BF16;
+}
+
+let mayRaiseFPException = true in {
+defm PseudoVFWMACC_ALT  : VPseudoVWMAC_VV_VF_RM_BF16;
+defm PseudoVFWNMACC_ALT : VPseudoVWMAC_VV_VF_RM_BF16;
+defm PseudoVFWMSAC_ALT  : VPseudoVWMAC_VV_VF_RM_BF16;
+defm PseudoVFWNMSAC_ALT : VPseudoVWMAC_VV_VF_RM_BF16;
+}
+
+let mayRaiseFPException = true in
+defm PseudoVFRSQRT7_ALT : VPseudoVRCP_V_BF16;
+
+let mayRaiseFPException = true in
+defm PseudoVFREC7_ALT : VPseudoVRCP_V_RM_BF16;
+
+let mayRaiseFPException = true in {
+defm PseudoVFMIN_ALT : VPseudoVMAX_VV_VF_BF16;
+defm PseudoVFMAX_ALT : VPseudoVMAX_VV_VF_BF16;
+}
+
+defm PseudoVFSGNJ_ALT  : VPseudoVSGNJ_VV_VF_BF16;
+defm PseudoVFSGNJN_ALT : VPseudoVSGNJ_VV_VF_BF16;
+defm PseudoVFSGNJX_ALT : VPseudoVSGNJ_VV_VF_BF16;
+
+let mayRaiseFPException = true in {
+defm PseudoVMFEQ_ALT : VPseudoVCMPM_VV_VF;
+defm PseudoVMFNE_ALT : VPseudoVCMPM_VV_VF;
+defm PseudoVMFLT_ALT : VPseudoVCMPM_VV_VF;
+defm PseudoVMFLE_ALT : VPseudoVCMPM_VV_VF;
+defm PseudoVMFGT_ALT : VPseudoVCMPM_VF;
+defm PseudoVMFGE_ALT : VPseudoVCMPM_VF;
+}
+
+defm PseudoVFCLASS_ALT : VPseudoVCLS_V;
+
+defm PseudoVFMERGE_ALT : VPseudoVMRG_FM;
+
+defm PseudoVFMV_V_ALT : VPseudoVMV_F;
+
+let mayRaiseFPException = true in {
+defm PseudoVFWCVT_F_XU_ALT : VPseudoVWCVTF_V_BF16;
+defm PseudoVFWCVT_F_X_ALT  : VPseudoVWCVTF_V_BF16;
+
+defm PseudoVFWCVT_F_F_ALT  : VPseudoVWCVTD_V_BF16;
+} // mayRaiseFPException = true
+
+let mayRaiseFPException = true in {
+let hasSideEffects = 0, hasPostISelHook = 1 in {
+defm PseudoVFNCVT_XU_F_ALT : VPseudoVNCVTI_W_RM;
+defm PseudoVFNCVT_X_F_ALT  : VPseudoVNCVTI_W_RM;
+}
+
+defm PseudoVFNCVT_RTZ_XU_F_ALT : VPseudoVNCVTI_W;
+defm PseudoVFNCVT_RTZ_X_F_ALT  : VPseudoVNCVTI_W;
+
+defm PseudoVFNCVT_F_F_ALT  : VPseudoVNCVTD_W_RM_BF16;
+
+defm PseudoVFNCVT_ROD_F_F_ALT : VPseudoVNCVTD_W_BF16;
+} // mayRaiseFPException = true
+
+let mayLoad = 0, mayStore = 0, hasSideEffects = 0 in {
+  defvar f = SCALAR_F16;
+  let HasSEWOp = 1, BaseInstr = VFMV_F_S in
+  def "PseudoVFMV_" # f.FX # "_S_ALT" :
+    RISCVVPseudo<(outs f.fprclass:$rd), (ins VR:$rs2, sew:$sew)>,
+    Sched<[WriteVMovFS, ReadVMovFS]>;
+  let HasVLOp = 1, HasSEWOp = 1, BaseInstr = VFMV_S_F, isReMaterializable = 1,
+      Constraints = "$rd = $passthru" in
+  def "PseudoVFMV_S_" # f.FX # "_ALT" :
+    RISCVVPseudo<(outs VR:$rd),
+                 (ins VR:$passthru, f.fprclass:$rs1, AVL:$vl, sew:$sew)>,
+    Sched<[WriteVMovSF, ReadVMovSF_V, ReadVMovSF_F]>;
+}
+
+defm PseudoVFSLIDE1UP_ALT   : VPseudoVSLD1_VF<"@earlyclobber $rd">;
+defm PseudoVFSLIDE1DOWN_ALT : VPseudoVSLD1_VF;
+} // Predicates = [HasStdExtZvfbfa], AltFmtType = IS_ALTFMT
+
 //===----------------------------------------------------------------------===//
 // Patterns
 //===----------------------------------------------------------------------===//
@@ -108,6 +438,130 @@ let Predicates = [HasStdExtZvfbfmin] in {
                   FRM_DYN,
                   fvti.AVL, fvti.Log2SEW, TA_MA)>;
   }
+
+  defm : VPatUnaryV_V_AnyMask<"int_riscv_vcompress", "PseudoVCOMPRESS", AllBF16Vectors>;
+  defm : VPatBinaryV_VV_VX_VI_INT<"int_riscv_vrgather", "PseudoVRGATHER",
+                                  AllBF16Vectors, uimm5>;
+  defm : VPatBinaryV_VV_INT_EEW<"int_riscv_vrgatherei16_vv", "PseudoVRGATHEREI16",
+                                eew=16, vtilist=AllBF16Vectors>;
+  defm : VPatTernaryV_VX_VI<"int_riscv_vslideup", "PseudoVSLIDEUP", AllBF16Vectors, uimm5>;
+  defm : VPatTernaryV_VX_VI<"int_riscv_vslidedown", "PseudoVSLIDEDOWN", AllBF16Vectors, uimm5>;
+
+  foreach fvti = AllBF16Vectors in {
+    defm : VPatBinaryCarryInTAIL<"int_riscv_vmerge", "PseudoVMERGE", "VVM",
+                                 fvti.Vector,
+                                 fvti.Vector, fvti.Vector, fvti.Mask,
+                                 fvti.Log2SEW, fvti.LMul, fvti.RegClass,
+                                 fvti.RegClass, fvti.RegClass>;
+    defm : VPatBinaryCarryInTAIL<"int_riscv_vfmerge", "PseudoVFMERGE",
+                                 "V"#fvti.ScalarSuffix#"M",
+                                 fvti.Vector,
+                                 fvti.Vector, fvti.Scalar, fvti.Mask,
+                                 fvti.Log2SEW, fvti.LMul, fvti.RegClass,
+                                 fvti.RegClass, fvti.ScalarRegClass>;
+    defvar instr = !cast<Instruction>("PseudoVMERGE_VIM_"#fvti.LMul.MX);
+    def : Pat<(fvti.Vector (int_riscv_vfmerge (fvti.Vector fvti.RegClass:$passthru),
+                                              (fvti.Vector fvti.RegClass:$rs2),
+                                              (fvti.Scalar (fpimm0)),
+                                              (fvti.Mask VMV0:$vm), VLOpFrag)),
+              (instr fvti.RegClass:$passthru, fvti.RegClass:$rs2, 0,
+                     (fvti.Mask VMV0:$vm), GPR:$vl, fvti.Log2SEW)>;
+
+    defvar ivti = GetIntVTypeInfo<fvti>.Vti;
+    def : Pat<(fvti.Vector (vselect (fvti.Mask VMV0:$vm), fvti.RegClass:$rs1,
+                                                          fvti.RegClass:$rs2)),
+              (!cast<Instruction>("PseudoVMERGE_VVM_"#fvti.LMul.MX)
+                   (fvti.Vector (IMPLICIT_DEF)),
+                   fvti.RegClass:$rs2, fvti.RegClass:$rs1, (fvti.Mask VMV0:$vm),
+                   fvti.AVL, fvti.Log2SEW)>;
+
+    def : Pat<(fvti.Vector (vselect (fvti.Mask VMV0:$vm),
+                                    (SplatFPOp (SelectScalarFPAsInt (XLenVT GPR:$imm))),
+                                    fvti.RegClass:$rs2)),
+              (!cast<Instruction>("PseudoVMERGE_VXM_"#fvti.LMul.MX)
+                   (fvti.Vector (IMPLICIT_DEF)),
+                   fvti.RegClass:$rs2, GPR:$imm, (fvti.Mask VMV0:$vm), fvti.AVL, fvti.Log2SEW)>;
+
+    def : Pat<(fvti.Vector (vselect (fvti.Mask VMV0:$vm),
+                                    (SplatFPOp (fvti.Scalar fpimm0)),
+                                    fvti.RegClass:$rs2)),
+              (!cast<Instruction>("PseudoVMERGE_VIM_"#fvti.LMul.MX)
+                   (fvti.Vector (IMPLICIT_DEF)),
+                   fvti.RegClass:$rs2, 0, (fvti.Mask VMV0:$vm), fvti.AVL, fvti.Log2SEW)>;
+
+    def : Pat<(fvti.Vector (vselect (fvti.Mask VMV0:$vm),
+                                    (SplatFPOp fvti.ScalarRegClass:$rs1),
+                                    fvti.RegClass:$rs2)),
+              (!cast<Instruction>("PseudoVFMERGE_V"#fvti.ScalarSuffix#"M_"#fvti.LMul.MX)
+                   (fvti.Vector (IMPLICIT_DEF)),
+                   fvti.RegClass:$rs2,
+                   (fvti.Scalar fvti.ScalarRegClass:$rs1),
+                   (fvti.Mask VMV0:$vm), fvti.AVL, fvti.Log2SEW)>;
+
+    def : Pat<(fvti.Vector (riscv_vmerge_vl (fvti.Mask VMV0:$vm),
+                                            fvti.RegClass:$rs1,
+                                            fvti.RegClass:$rs2,
+                                            fvti.RegClass:$passthru,
+                                            VLOpFrag)),
+              (!cast<Instruction>("PseudoVMERGE_VVM_"#fvti.LMul.MX)
+                   fvti.RegClass:$passthru, fvti.RegClass:$rs2, fvti.RegClass:$rs1, (fvti.Mask VMV0:$vm),
+                   GPR:$vl, fvti.Log2SEW)>;
+
+    def : Pat<(fvti.Vector (riscv_vmerge_vl (fvti.Mask VMV0:$vm),
+                                            (SplatFPOp (SelectScalarFPAsInt (XLenVT GPR:$imm))),
+                                            fvti.RegClass:$rs2,
+                                            fvti.RegClass:$passthru,
+                                            VLOpFrag)),
+              (!cast<Instruction>("PseudoVMERGE_VXM_"#fvti.LMul.MX)
+                   fvti.RegClass:$passthru, fvti.RegClass:$rs2, GPR:$imm, (fvti.Mask VMV0:$vm),
+                   GPR:$vl, fvti.Log2SEW)>;
+
+
+    def : Pat<(fvti.Vector (riscv_vmerge_vl (fvti.Mask VMV0:$vm),
+                                            (SplatFPOp (fvti.Scalar fpimm0)),
+                                            fvti.RegClass:$rs2,
+                                            fvti.RegClass:$passthru,
+                                            VLOpFrag)),
+              (!cast<Instruction>("PseudoVMERGE_VIM_"#fvti.LMul.MX)
+                   fvti.RegClass:$passthru, fvti.RegClass:$rs2, 0, (fvti.Mask VMV0:$vm),
+                   GPR:$vl, fvti.Log2SEW)>;
+
+    def : Pat<(fvti.Vector (riscv_vmerge_vl (fvti.Mask VMV0:$vm),
+                                            (SplatFPOp fvti.ScalarRegClass:$rs1),
+                                            fvti.RegClass:$rs2,
+                                            fvti.RegClass:$passthru,
+                                            VLOpFrag)),
+              (!cast<Instruction>("PseudoVFMERGE_V"#fvti.ScalarSuffix#"M_"#fvti.LMul.MX)
+                   fvti.RegClass:$passthru, fvti.RegClass:$rs2,
+                   (fvti.Scalar fvti.ScalarRegClass:$rs1),
+                   (fvti.Mask VMV0:$vm), GPR:$vl, fvti.Log2SEW)>;
+
+    def : Pat<(fvti.Vector
+               (riscv_vrgather_vv_vl fvti.RegClass:$rs2,
+                                     (ivti.Vector fvti.RegClass:$rs1),
+                                     fvti.RegClass:$passthru,
+                                     (fvti.Mask VMV0:$vm),
+                                     VLOpFrag)),
+              (!cast<Instruction>("PseudoVRGATHER_VV_"# fvti.LMul.MX#"_E"# fvti.SEW#"_MASK")
+                   fvti.RegClass:$passthru, fvti.RegClass:$rs2, fvti.RegClass:$rs1,
+                   (fvti.Mask VMV0:$vm), GPR:$vl, fvti.Log2SEW, TAIL_AGNOSTIC)>;
+    def : Pat<(fvti.Vector (riscv_vrgather_vx_vl fvti.RegClass:$rs2, GPR:$rs1,
+                                                fvti.RegClass:$passthru,
+                                                (fvti.Mask VMV0:$vm),
+                                                VLOpFrag)),
+              (!cast<Instruction>("PseudoVRGATHER_VX_"# fvti.LMul.MX#"_MASK")
+                   fvti.RegClass:$passthru, fvti.RegClass:$rs2, GPR:$rs1,
+                   (fvti.Mask VMV0:$vm), GPR:$vl, fvti.Log2SEW, TAIL_AGNOSTIC)>;
+    def : Pat<(fvti.Vector
+               (riscv_vrgather_vx_vl fvti.RegClass:$rs2,
+                                     uimm5:$imm,
+                                     fvti.RegClass:$passthru,
+                                     (fvti.Mask VMV0:$vm),
+                                     VLOpFrag)),
+              (!cast<Instruction>("PseudoVRGATHER_VI_"# fvti.LMul.MX#"_MASK")
+                   fvti.RegClass:$passthru, fvti.RegClass:$rs2, uimm5:$imm,
+                   (fvti.Mask VMV0:$vm), GPR:$vl, fvti.Log2SEW, TAIL_AGNOSTIC)>;
+  }
 }
 
 let Predicates = [HasStdExtZvfbfwma] in {
@@ -118,3 +572,224 @@ let Predicates = [HasStdExtZvfbfwma] in {
   defm : VPatWidenFPMulAccSDNode_VV_VF_RM<"PseudoVFWMACCBF16",
                                           AllWidenableBF16ToFloatVectors>;
 }
+
+multiclass VPatConversionVI_VF_BF16<string intrinsic, string instruction> {
+  foreach fvti = AllBF16Vectors in {
+    defvar ivti = GetIntVTypeInfo<fvti>.Vti;
+    let Predicates = !listconcat(GetVTypePredicates<fvti>.Predicates,
+                                 GetVTypePredicates<ivti>.Predicates) in
+    defm : VPatConversion<intrinsic, instruction, "V",
+                          ivti.Vector, fvti.Vector, ivti.Mask, fvti.Log2SEW,
+                          fvti.LMul, ivti.RegClass, fvti.RegClass>;
+  }
+}
+
+multiclass VPatConversionWF_VI_BF16<string intrinsic, string instruction,
+                                    bit isSEWAware = 0> {
+  foreach vtiToWti = AllWidenableIntToBF16Vectors in {
+    defvar vti = vtiToWti.Vti;
+    defvar fwti = vtiToWti.Wti;
+    let Predicates = !listconcat(GetVTypePredicates<vti>.Predicates,
+                                 GetVTypePredicates<fwti>.Predicates) in
+    defm : VPatConversion<intrinsic, instruction, "V",
+                          fwti.Vector, vti.Vector, fwti.Mask, vti.Log2SEW,
+                          vti.LMul, fwti.RegClass, vti.RegClass, isSEWAware>;
+  }
+}
+
+multiclass VPatConversionWF_VF_BF16<string intrinsic, string instruction,
+                                    bit isSEWAware = 0> {
+  foreach fvtiToFWti = AllWidenableBF16ToFloatVectors in {
+    defvar fvti = fvtiToFWti.Vti;
+    defvar fwti = fvtiToFWti.Wti;
+    let Predicates = !listconcat(GetVTypeMinimalPredicates<fvti>.Predicates,
+                                 GetVTypeMinimalPredicates<fwti>.Predicates) in
+    defm : VPatConversion<intrinsic, instruction, "V",
+                          fwti.Vector, fvti.Vector, fwti.Mask, fvti.Log2SEW,
+                          fvti.LMul, fwti.RegClass, fvti.RegClass, isSEWAware>;
+  }
+}
+
+multiclass VPatConversionVI_WF_BF16<string intrinsic, string instruction> {
+  foreach vtiToWti = AllWidenableIntToBF16Vectors in {
+    defvar vti = vtiToWti.Vti;
+    defvar fwti = vtiToWti.Wti;
+    let Predicates = !listconcat(GetVTypePredicates<vti>.Predicates,
+                                 GetVTypePredicates<fwti>.Predicates) in
+    defm : VPatConversion<intrinsic, instruction, "W",
+                          vti.Vector, fwti.Vector, vti.Mask, vti.Log2SEW,
+                          vti.LMul, vti.RegClass, fwti.RegClass>;
+  }
+}
+
+multiclass VPatConversionVI_WF_RM_BF16<string intrinsic, string instruction> {
+  foreach vtiToWti = AllWidenableIntToBF16Vectors in {
+    defvar vti = vtiToWti.Vti;
+    defvar fwti = vtiToWti.Wti;
+    let Predicates = !listconcat(GetVTypePredicates<vti>.Predicates,
+                                 GetVTypePredicates<fwti>.Predicates) in
+    defm : VPatConversionRoundingMode<intrinsic, instruction, "W",
+                                      vti.Vector, fwti.Vector, vti.Mask, vti.Log2SEW,
+                                      vti.LMul, vti.RegClass, fwti.RegClass>;
+  }
+}
+
+multiclass VPatConversionVF_WF_BF16<string intrinsic, string instruction,
+                                    bit isSEWAware = 0> {
+  foreach fvtiToFWti = AllWidenableBF16ToFloatVectors in {
+    defvar fvti = fvtiToFWti.Vti;
+    defvar fwti = fvtiToFWti.Wti;
+    let Predicates = !listconcat(GetVTypePredicates<fvti>.Predicates,
+                                 GetVTypePredicates<fwti>.Predicates) in
+    defm : VPatConversion<intrinsic, instruction, "W",
+                          fvti.Vector, fwti.Vector, fvti.Mask, fvti.Log2SEW,
+                          fvti.LMul, fvti.RegClass, fwti.RegClass, isSEWAware>;
+  }
+}
+
+let Predicates = [HasStdExtZvfbfa] in {
+defm : VPatBinaryV_VV_VX_RM<"int_riscv_vfadd", "PseudoVFADD_ALT",
+                            AllBF16Vectors, isSEWAware = 1>;
+defm : VPatBinaryV_VV_VX_RM<"int_riscv_vfsub", "PseudoVFSUB_ALT",
+                            AllBF16Vectors, isSEWAware = 1>;
+defm : VPatBinaryV_VX_RM<"int_riscv_vfrsub", "PseudoVFRSUB_ALT",
+                         AllBF16Vectors, isSEWAware = 1>;
+defm : VPatBinaryW_VV_VX_RM<"int_riscv_vfwadd", "PseudoVFWADD_ALT",
+                            AllWidenableBF16ToFloatVectors, isSEWAware=1>;
+defm : VPatBinaryW_VV_VX_RM<"int_riscv_vfwsub", "PseudoVFWSUB_ALT",
+                            AllWidenableBF16ToFloatVectors, isSEWAware=1>;
+defm : VPatBinaryW_WV_WX_RM<"int_riscv_vfwadd_w", "PseudoVFWADD_ALT",
+                            AllWidenableBF16ToFloatVectors, isSEWAware=1>;
+defm : VPatBinaryW_WV_WX_RM<"int_riscv_vfwsub_w", "PseudoVFWSUB_ALT",
+                            AllWidenableBF16ToFloatVectors, isSEWAware=1>;
+defm : VPatBinaryV_VV_VX_RM<"int_riscv_vfmul", "PseudoVFMUL_ALT",
+                            AllBF16Vectors, isSEWAware=1>;
+defm : VPatBinaryW_VV_VX_RM<"int_riscv_vfwmul", "PseudoVFWMUL_ALT",
+                            AllWidenableBF16ToFloatVectors, isSEWAware=1>;
+defm : VPatTernaryV_VV_VX_AAXA_RM<"int_riscv_vfmacc", "PseudoVFMACC_ALT",
+                                  AllBF16Vectors, isSEWAware=1>;
+defm : VPatTernaryV_VV_VX_AAXA_RM<"int_riscv_vfnmacc", "PseudoVFNMACC_ALT",
+                                  AllBF16Vectors, isSEWAware=1>;
+defm : VPatTernaryV_VV_VX_AAXA_RM<"int_riscv_vfmsac", "PseudoVFMSAC_ALT",
+                                  AllBF16Vectors, isSEWAware=1>;
+defm : VPatTernaryV_VV_VX_AAXA_RM<"int_riscv_vfnmsac", "PseudoVFNMSAC_ALT",
+                                  AllBF16Vectors, isSEWAware=1>;
+defm : VPatTernaryV_VV_VX_AAXA_RM<"int_riscv_vfmadd", "PseudoVFMADD_ALT",
+                                  AllBF16Vectors, isSEWAware=1>;
+defm : VPatTernaryV_VV_VX_AAXA_RM<"int_riscv_vfnmadd", "PseudoVFNMADD_ALT",
+                                  AllBF16Vectors, isSEWAware=1>;
+defm : VPatTernaryV_VV_VX_AAXA_RM<"int_riscv_vfmsub", "PseudoVFMSUB_ALT",
+                                  AllBF16Vectors, isSEWAware=1>;
+defm : VPatTernaryV_VV_VX_AAXA_RM<"int_riscv_vfnmsub", "PseudoVFNMSUB_ALT",
+                                  AllBF16Vectors, isSEWAware=1>;
+defm : VPatTernaryW_VV_VX_RM<"int_riscv_vfwmacc", "PseudoVFWMACC_ALT",
+                             AllWidenableBF16ToFloatVectors, isSEWAware=1>;
+defm : VPatTernaryW_VV_VX_RM<"int_riscv_vfwnmacc", "PseudoVFWNMACC_ALT",
+                             AllWidenableBF16ToFloatVectors, isSEWAware=1>;
+defm : VPatTernaryW_VV_VX_RM<"int_riscv_vfwmsac", "PseudoVFWMSAC_ALT",
+                             AllWidenableBF16ToFloatVectors, isSEWAware=1>;
+defm : VPatTernaryW_VV_VX_RM<"int_riscv_vfwnmsac", "PseudoVFWNMSAC_ALT",
+                             AllWidenableBF16ToFloatVectors, isSEWAware=1>;
+defm : VPatUnaryV_V<"int_riscv_vfrsqrt7", "PseudoVFRSQRT7_ALT",
+                    AllBF16Vectors, isSEWAware=1>;
+defm : VPatUnaryV_V_RM<"int_riscv_vfrec7", "PseudoVFREC7_ALT",
+                       AllBF16Vectors, isSEWAware=1>;
+defm : VPatBinaryV_VV_VX<"int_riscv_vfmin", "PseudoVFMIN_ALT",
+                         AllBF16Vectors, isSEWAware=1>;
+defm : VPatBinaryV_VV_VX<"int_riscv_vfmax", "PseudoVFMAX_ALT",
+                         AllBF16Vectors, isSEWAware=1>;
+defm : VPatBinaryV_VV_VX<"int_riscv_vfsgnj", "PseudoVFSGNJ_ALT",
+                         AllBF16Vectors, isSEWAware=1>;
+defm : VPatBinaryV_VV_VX<"int_riscv_vfsgnjn", "PseudoVFSGNJN_ALT",
+                         AllBF16Vectors, isSEWAware=1>;
+defm : VPatBinaryV_VV_VX<"int_riscv_vfsgnjx", "PseudoVFSGNJX_ALT",
+                         AllBF16Vectors, isSEWAware=1>;
+defm : VPatBinaryM_VV_VX<"int_riscv_vmfeq", "PseudoVMFEQ_ALT", AllBF16Vectors>;
+defm : VPatBinaryM_VV_VX<"int_riscv_vmfle", "PseudoVMFLE_ALT", AllBF16Vectors>;
+defm : VPatBinaryM_VV_VX<"int_riscv_vmflt", "PseudoVMFLT_ALT", AllBF16Vectors>;
+defm : VPatBinaryM_VV_VX<"int_riscv_vmfne", "PseudoVMFNE_ALT", AllBF16Vectors>;
+defm : VPatBinaryM_VX<"int_riscv_vmfgt", "PseudoVMFGT_ALT", AllBF16Vectors>;
+defm : VPatBinaryM_VX<"int_riscv_vmfge", "PseudoVMFGE_ALT", AllBF16Vectors>;
+defm : VPatBinarySwappedM_VV<"int_riscv_vmfgt", "PseudoVMFLT_ALT", AllBF16Vectors>;
+defm : VPatBinarySwappedM_VV<"int_riscv_vmfge", "PseudoVMFLE_ALT", AllBF16Vectors>;
+defm : VPatConversionVI_VF_BF16<"int_riscv_vfclass", "PseudoVFCLASS_ALT">;
+foreach vti = AllBF16Vectors in {
+  let Predicates = GetVTypePredicates<vti>.Predicates in
+    defm : VPatBinaryCarryInTAIL<"int_riscv_vfmerge", "PseudoVFMERGE_ALT",
+                                 "V"#vti.ScalarSuffix#"M",
+                                 vti.Vector,
+                                 vti.Vector, vti.Scalar, vti.Mask,
+                                 vti.Log2SEW, vti.LMul, vti.RegClass,
+                                 vti.RegClass, vti.ScalarRegClass>;
+}
+defm : VPatConversionWF_VI_BF16<"int_riscv_vfwcvt_f_xu_v", "PseudoVFWCVT_F_XU_ALT",
+                                isSEWAware=1>;
+defm : VPatConversionWF_VI_BF16<"int_riscv_vfwcvt_f_x_v", "PseudoVFWCVT_F_X_ALT",
+                                isSEWAware=1>;
+defm : VPatConversionWF_VF_BF16<"int_riscv_vfwcvt_f_f_v", "PseudoVFWCVT_F_F_ALT",
+                                isSEWAware=1>;
+defm : VPatConversionVI_WF_RM_BF16<"int_riscv_vfncvt_xu_f_w", "PseudoVFNCVT_XU_F_ALT">;
+defm : VPatConversionVI_WF_RM_BF16<"int_riscv_vfncvt_x_f_w", "PseudoVFNCVT_X_F_ALT">;
+defm : VPatConversionVI_WF_BF16<"int_riscv_vfncvt_rtz_xu_f_w", "PseudoVFNCVT_RTZ_XU_F_ALT">;
+defm : VPatConversionVI_WF_BF16<"int_riscv_vfncvt_rtz_x_f_w", "PseudoVFNCVT_RTZ_X_F_ALT">;
+defm : VPatConversionVF_WF_RM<"int_riscv_vfncvt_f_f_w", "PseudoVFNCVT_F_F_ALT",
+                              AllWidenableBF16ToFloatVectors, isSEWAware=1>;
+defm : VPatConversionVF_WF_BF16<"int_riscv_vfncvt_rod_f_f_w", "PseudoVFNCVT_ROD_F_F_ALT",
+                                isSEWAware=1>;
+defm : VPatBinaryV_VX<"int_riscv_vfslide1up", "PseudoVFSLIDE1UP_ALT", AllBF16Vectors>;
+defm : VPatBinaryV_VX<"int_riscv_vfslide1down", "PseudoVFSLIDE1DOWN_ALT", AllBF16Vectors>;
+
+foreach fvti = AllBF16Vectors in {
+  defvar ivti = GetIntVTypeInfo<fvti>.Vti;
+  let Predicates = GetVTypePredicates<ivti>.Predicates in {
+    // 13.16. Vector Floating-Point Move Instruction
+    // If we're splatting fpimm0, use vmv.v.x vd, x0.
+    def : Pat<(fvti.Vector (riscv_vfmv_v_f_vl
+                           fvti.Vector:$passthru, (fvti.Scalar (fpimm0)), VLOpFrag)),
+              (!cast<Instruction>("PseudoVMV_V_I_"#fvti.LMul.MX)
+               $passthru, 0, GPR:$vl, fvti.Log2SEW, TU_MU)>;
+    def : Pat<(fvti.Vector (riscv_vfmv_v_f_vl
+                           fvti.Vector:$passthru, (fvti.Scalar (SelectScalarFPAsInt (XLenVT GPR:$imm))), VLOpFrag)),
+              (!cast<Instruction>("PseudoVMV_V_X_"#fvti.LMul.MX)
+               $passthru, GPR:$imm, GPR:$vl, fvti.Log2SEW, TU_MU)>;
+  }
+
+  let Predicates = GetVTypePredicates<fvti>.Predicates in {
+    def : Pat<(fvti.Vector (riscv_vfmv_v_f_vl
+                           fvti.Vector:$passthru, (fvti.Scalar fvti.ScalarRegClass:$rs2), VLOpFrag)),
+              (!cast<Instruction>("PseudoVFMV_V_ALT_" # fvti.ScalarSuffix # "_" #
+                                  fvti.LMul.MX)
+               $passthru, (fvti.Scalar fvti.ScalarRegClass:$rs2),
+               GPR:$vl, fvti.Log2SEW, TU_MU)>;
+  }
+}
+
+foreach vti = NoGroupBF16Vectors in {
+  let Predicates = GetVTypePredicates<vti>.Predicates in {
+    def : Pat<(vti.Vector (riscv_vfmv_s_f_vl (vti.Vector vti.RegClass:$passthru),
+                                             (vti.Scalar (fpimm0)),
+                                             VLOpFrag)),
+              (PseudoVMV_S_X $passthru, (XLenVT X0), GPR:$vl, vti.Log2SEW)>;
+    def : Pat<(vti.Vector (riscv_vfmv_s_f_vl (vti.Vector vti.RegClass:$passthru),
+                                             (vti.Scalar (SelectScalarFPAsInt (XLenVT GPR:$imm))),
+                                             VLOpFrag)),
+              (PseudoVMV_S_X $passthru, GPR:$imm, GPR:$vl, vti.Log2SEW)>;
+    def : Pat<(vti.Vector (riscv_vfmv_s_f_vl (vti.Vector vti.RegClass:$passthru),
+                                             vti.ScalarRegClass:$rs1,
+                                             VLOpFrag)),
+              (!cast<Instruction>("PseudoVFMV_S_"#vti.ScalarSuffix#"_ALT")
+                  vti.RegClass:$passthru,
+                  (vti.Scalar vti.ScalarRegClass:$rs1), GPR:$vl, vti.Log2SEW)>;
+  }
+
+  defvar vfmv_f_s_inst = !cast<Instruction>(!strconcat("PseudoVFMV_",
+                                                       vti.ScalarSuffix,
+                                                       "_S_ALT"));
+  // Only pattern-match extract-element operations where the index is 0. Any
+  // other index will have been custom-lowered to slide the vector correctly
+  // into place.
+  let Predicates = GetVTypePredicates<vti>.Predicates in
+  def : Pat<(vti.Scalar (extractelt (vti.Vector vti.RegClass:$rs2), 0)),
+            (vfmv_f_s_inst vti.RegClass:$rs2, vti.Log2SEW)>;
+}
+} // Predicates = [HasStdExtZvfbfa]
diff --git a/llvm/lib/Target/RISCV/RISCVSubtarget.h b/llvm/lib/Target/RISCV/RISCVSubtarget.h
index 6acf799..334db4b 100644
--- a/llvm/lib/Target/RISCV/RISCVSubtarget.h
+++ b/llvm/lib/Target/RISCV/RISCVSubtarget.h
@@ -288,9 +288,12 @@ public:
   bool hasVInstructionsI64() const { return HasStdExtZve64x; }
   bool hasVInstructionsF16Minimal() const { return HasStdExtZvfhmin; }
   bool hasVInstructionsF16() const { return HasStdExtZvfh; }
-  bool hasVInstructionsBF16Minimal() const { return HasStdExtZvfbfmin; }
+  bool hasVInstructionsBF16Minimal() const {
+    return HasStdExtZvfbfmin || HasStdExtZvfbfa;
+  }
   bool hasVInstructionsF32() const { return HasStdExtZve32f; }
   bool hasVInstructionsF64() const { return HasStdExtZve64d; }
+  bool hasVInstructionsBF16() const { return HasStdExtZvfbfa; }
   // F16 and F64 both require F32.
   bool hasVInstructionsAnyF() const { return hasVInstructionsF32(); }
   bool hasVInstructionsFullMultiply() const { return HasStdExtV; }
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
index 09cb225..a8eb9b9 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
@@ -3757,6 +3757,10 @@ static Instruction *foldBitCeil(SelectInst &SI, IRBuilderBase &Builder,
 //   (x < y) ? -1 : zext(x > y)
 //   (x > y) ? 1 : sext(x != y)
 //   (x > y) ? 1 : sext(x < y)
+//   (x == y) ? 0 : (x > y ? 1 : -1)
+//   (x == y) ? 0 : (x < y ? -1 : 1)
+//   Special case: x == C ? 0 : (x > C - 1 ? 1 : -1)
+//   Special case: x == C ? 0 : (x < C + 1 ? -1 : 1)
 // Into ucmp/scmp(x, y), where signedness is determined by the signedness
 // of the comparison in the original sequence.
 Instruction *InstCombinerImpl::foldSelectToCmp(SelectInst &SI) {
@@ -3849,6 +3853,44 @@ Instruction *InstCombinerImpl::foldSelectToCmp(SelectInst &SI) {
     }
   }
 
+  // Special cases with constants: x == C ? 0 : (x > C-1 ? 1 : -1)
+  if (Pred == ICmpInst::ICMP_EQ && match(TV, m_Zero())) {
+    const APInt *C;
+    if (match(RHS, m_APInt(C))) {
+      CmpPredicate InnerPred;
+      Value *InnerRHS;
+      const APInt *InnerTV, *InnerFV;
+      if (match(FV,
+                m_Select(m_ICmp(InnerPred, m_Specific(LHS), m_Value(InnerRHS)),
+                         m_APInt(InnerTV), m_APInt(InnerFV)))) {
+
+        // x == C ? 0 : (x > C-1 ? 1 : -1)
+        if (ICmpInst::isGT(InnerPred) && InnerTV->isOne() &&
+            InnerFV->isAllOnes()) {
+          IsSigned = ICmpInst::isSigned(InnerPred);
+          bool CanSubOne = IsSigned ? !C->isMinSignedValue() : !C->isMinValue();
+          if (CanSubOne) {
+            APInt Cminus1 = *C - 1;
+            if (match(InnerRHS, m_SpecificInt(Cminus1)))
+              Replace = true;
+          }
+        }
+
+        // x == C ? 0 : (x < C+1 ? -1 : 1)
+        if (ICmpInst::isLT(InnerPred) && InnerTV->isAllOnes() &&
+            InnerFV->isOne()) {
+          IsSigned = ICmpInst::isSigned(InnerPred);
+          bool CanAddOne = IsSigned ? !C->isMaxSignedValue() : !C->isMaxValue();
+          if (CanAddOne) {
+            APInt Cplus1 = *C + 1;
+            if (match(InnerRHS, m_SpecificInt(Cplus1)))
+              Replace = true;
+          }
+        }
+      }
+    }
+  }
+
   Intrinsic::ID IID = IsSigned ? Intrinsic::scmp : Intrinsic::ucmp;
   if (Replace)
     return replaceInstUsesWith(
diff --git a/llvm/lib/Transforms/Utils/SCCPSolver.cpp b/llvm/lib/Transforms/Utils/SCCPSolver.cpp
index 9693ae6..4947d03 100644
--- a/llvm/lib/Transforms/Utils/SCCPSolver.cpp
+++ b/llvm/lib/Transforms/Utils/SCCPSolver.cpp
@@ -20,6 +20,7 @@
 #include "llvm/Analysis/ValueLatticeUtils.h"
 #include "llvm/Analysis/ValueTracking.h"
 #include "llvm/IR/ConstantRange.h"
+#include "llvm/IR/DerivedTypes.h"
 #include "llvm/IR/IRBuilder.h"
 #include "llvm/IR/InstVisitor.h"
 #include "llvm/IR/Instructions.h"
@@ -634,18 +635,10 @@ private:
   /// Merge \p MergeWithV into \p IV and push \p V to the worklist, if \p IV
   /// changes.
   bool mergeInValue(ValueLatticeElement &IV, Value *V,
-                    ValueLatticeElement MergeWithV,
+                    const ValueLatticeElement &MergeWithV,
                     ValueLatticeElement::MergeOptions Opts = {
                         /*MayIncludeUndef=*/false, /*CheckWiden=*/false});
 
-  bool mergeInValue(Value *V, ValueLatticeElement MergeWithV,
-                    ValueLatticeElement::MergeOptions Opts = {
-                        /*MayIncludeUndef=*/false, /*CheckWiden=*/false}) {
-    assert(!V->getType()->isStructTy() &&
-           "non-structs should use markConstant");
-    return mergeInValue(ValueState[V], V, MergeWithV, Opts);
-  }
-
   /// getValueState - Return the ValueLatticeElement object that corresponds to
   /// the value.  This function handles the case when the value hasn't been seen
   /// yet by properly seeding constants etc.
@@ -768,6 +761,7 @@ private:
   void handleCallArguments(CallBase &CB);
   void handleExtractOfWithOverflow(ExtractValueInst &EVI,
                                    const WithOverflowInst *WO, unsigned Idx);
+  bool isInstFullyOverDefined(Instruction &Inst);
 
 private:
   friend class InstVisitor<SCCPInstVisitor>;
@@ -987,7 +981,7 @@ public:
   void trackValueOfArgument(Argument *A) {
     if (A->getType()->isStructTy())
       return (void)markOverdefined(A);
-    mergeInValue(A, getArgAttributeVL(A));
+    mergeInValue(ValueState[A], A, getArgAttributeVL(A));
   }
 
   bool isStructLatticeConstant(Function *F, StructType *STy);
@@ -1128,8 +1122,7 @@ bool SCCPInstVisitor::isStructLatticeConstant(Function *F, StructType *STy) {
   for (unsigned i = 0, e = STy->getNumElements(); i != e; ++i) {
     const auto &It = TrackedMultipleRetVals.find(std::make_pair(F, i));
     assert(It != TrackedMultipleRetVals.end());
-    ValueLatticeElement LV = It->second;
-    if (!SCCPSolver::isConstant(LV))
+    if (!SCCPSolver::isConstant(It->second))
       return false;
   }
   return true;
@@ -1160,7 +1153,7 @@ Constant *SCCPInstVisitor::getConstantOrNull(Value *V) const {
     std::vector<Constant *> ConstVals;
     auto *ST = cast<StructType>(V->getType());
     for (unsigned I = 0, E = ST->getNumElements(); I != E; ++I) {
-      ValueLatticeElement LV = LVs[I];
+      const ValueLatticeElement &LV = LVs[I];
       ConstVals.push_back(SCCPSolver::isConstant(LV)
                               ? getConstant(LV, ST->getElementType(I))
                               : UndefValue::get(ST->getElementType(I)));
@@ -1225,7 +1218,7 @@ void SCCPInstVisitor::visitInstruction(Instruction &I) {
 }
 
 bool SCCPInstVisitor::mergeInValue(ValueLatticeElement &IV, Value *V,
-                                   ValueLatticeElement MergeWithV,
+                                   const ValueLatticeElement &MergeWithV,
                                    ValueLatticeElement::MergeOptions Opts) {
   if (IV.mergeIn(MergeWithV, Opts)) {
     pushUsersToWorkList(V);
@@ -1264,7 +1257,7 @@ void SCCPInstVisitor::getFeasibleSuccessors(Instruction &TI,
       return;
     }
 
-    ValueLatticeElement BCValue = getValueState(BI->getCondition());
+    const ValueLatticeElement &BCValue = getValueState(BI->getCondition());
     ConstantInt *CI = getConstantInt(BCValue, BI->getCondition()->getType());
     if (!CI) {
       // Overdefined condition variables, and branches on unfoldable constant
@@ -1326,7 +1319,7 @@ void SCCPInstVisitor::getFeasibleSuccessors(Instruction &TI,
   // the target as executable.
   if (auto *IBR = dyn_cast<IndirectBrInst>(&TI)) {
     // Casts are folded by visitCastInst.
-    ValueLatticeElement IBRValue = getValueState(IBR->getAddress());
+    const ValueLatticeElement &IBRValue = getValueState(IBR->getAddress());
     BlockAddress *Addr = dyn_cast_or_null<BlockAddress>(
         getConstant(IBRValue, IBR->getAddress()->getType()));
     if (!Addr) { // Overdefined or unknown condition?
@@ -1383,49 +1376,66 @@ bool SCCPInstVisitor::isEdgeFeasible(BasicBlock *From, BasicBlock *To) const {
 // 7. If a conditional branch has a value that is overdefined, make all
 //    successors executable.
 void SCCPInstVisitor::visitPHINode(PHINode &PN) {
-  // If this PN returns a struct, just mark the result overdefined.
-  // TODO: We could do a lot better than this if code actually uses this.
-  if (PN.getType()->isStructTy())
-    return (void)markOverdefined(&PN);
-
-  if (getValueState(&PN).isOverdefined())
-    return; // Quick exit
-
   // Super-extra-high-degree PHI nodes are unlikely to ever be marked constant,
   // and slow us down a lot.  Just mark them overdefined.
   if (PN.getNumIncomingValues() > 64)
     return (void)markOverdefined(&PN);
 
-  unsigned NumActiveIncoming = 0;
+  if (isInstFullyOverDefined(PN))
+    return;
+  SmallVector<unsigned> FeasibleIncomingIndices;
+  for (unsigned i = 0, e = PN.getNumIncomingValues(); i != e; ++i) {
+    if (!isEdgeFeasible(PN.getIncomingBlock(i), PN.getParent()))
+      continue;
+    FeasibleIncomingIndices.push_back(i);
+  }
 
   // Look at all of the executable operands of the PHI node.  If any of them
   // are overdefined, the PHI becomes overdefined as well.  If they are all
   // constant, and they agree with each other, the PHI becomes the identical
   // constant.  If they are constant and don't agree, the PHI is a constant
   // range. If there are no executable operands, the PHI remains unknown.
-  ValueLatticeElement PhiState = getValueState(&PN);
-  for (unsigned i = 0, e = PN.getNumIncomingValues(); i != e; ++i) {
-    if (!isEdgeFeasible(PN.getIncomingBlock(i), PN.getParent()))
-      continue;
-
-    ValueLatticeElement IV = getValueState(PN.getIncomingValue(i));
-    PhiState.mergeIn(IV);
-    NumActiveIncoming++;
-    if (PhiState.isOverdefined())
-      break;
+  if (StructType *STy = dyn_cast<StructType>(PN.getType())) {
+    for (unsigned i = 0, e = STy->getNumElements(); i != e; ++i) {
+      ValueLatticeElement PhiState = getStructValueState(&PN, i);
+      if (PhiState.isOverdefined())
+        continue;
+      for (unsigned j : FeasibleIncomingIndices) {
+        const ValueLatticeElement &IV =
+            getStructValueState(PN.getIncomingValue(j), i);
+        PhiState.mergeIn(IV);
+        if (PhiState.isOverdefined())
+          break;
+      }
+      ValueLatticeElement &PhiStateRef = getStructValueState(&PN, i);
+      mergeInValue(PhiStateRef, &PN, PhiState,
+                   ValueLatticeElement::MergeOptions().setMaxWidenSteps(
+                       FeasibleIncomingIndices.size() + 1));
+      PhiStateRef.setNumRangeExtensions(
+          std::max((unsigned)FeasibleIncomingIndices.size(),
+                   PhiStateRef.getNumRangeExtensions()));
+    }
+  } else {
+    ValueLatticeElement PhiState = getValueState(&PN);
+    for (unsigned i : FeasibleIncomingIndices) {
+      const ValueLatticeElement &IV = getValueState(PN.getIncomingValue(i));
+      PhiState.mergeIn(IV);
+      if (PhiState.isOverdefined())
+        break;
+    }
+    // We allow up to 1 range extension per active incoming value and one
+    // additional extension. Note that we manually adjust the number of range
+    // extensions to match the number of active incoming values. This helps to
+    // limit multiple extensions caused by the same incoming value, if other
+    // incoming values are equal.
+    ValueLatticeElement &PhiStateRef = ValueState[&PN];
+    mergeInValue(PhiStateRef, &PN, PhiState,
+                 ValueLatticeElement::MergeOptions().setMaxWidenSteps(
+                     FeasibleIncomingIndices.size() + 1));
+    PhiStateRef.setNumRangeExtensions(
+        std::max((unsigned)FeasibleIncomingIndices.size(),
+                 PhiStateRef.getNumRangeExtensions()));
   }
-
-  // We allow up to 1 range extension per active incoming value and one
-  // additional extension. Note that we manually adjust the number of range
-  // extensions to match the number of active incoming values. This helps to
-  // limit multiple extensions caused by the same incoming value, if other
-  // incoming values are equal.
-  mergeInValue(&PN, PhiState,
-               ValueLatticeElement::MergeOptions().setMaxWidenSteps(
-                   NumActiveIncoming + 1));
-  ValueLatticeElement &PhiStateRef = getValueState(&PN);
-  PhiStateRef.setNumRangeExtensions(
-      std::max(NumActiveIncoming, PhiStateRef.getNumRangeExtensions()));
 }
 
 void SCCPInstVisitor::visitReturnInst(ReturnInst &I) {
@@ -1481,7 +1491,7 @@ void SCCPInstVisitor::visitCastInst(CastInst &I) {
     }
   }
 
-  ValueLatticeElement OpSt = getValueState(I.getOperand(0));
+  const ValueLatticeElement &OpSt = getValueState(I.getOperand(0));
   if (OpSt.isUnknownOrUndef())
     return;
 
@@ -1496,9 +1506,9 @@ void SCCPInstVisitor::visitCastInst(CastInst &I) {
   if (I.getDestTy()->isIntOrIntVectorTy() &&
       I.getSrcTy()->isIntOrIntVectorTy() &&
       I.getOpcode() != Instruction::BitCast) {
-    auto &LV = getValueState(&I);
     ConstantRange OpRange =
         OpSt.asConstantRange(I.getSrcTy(), /*UndefAllowed=*/false);
+    auto &LV = getValueState(&I);
 
     Type *DestTy = I.getDestTy();
     ConstantRange Res = ConstantRange::getEmpty(DestTy->getScalarSizeInBits());
@@ -1516,19 +1526,24 @@ void SCCPInstVisitor::handleExtractOfWithOverflow(ExtractValueInst &EVI,
                                                   const WithOverflowInst *WO,
                                                   unsigned Idx) {
   Value *LHS = WO->getLHS(), *RHS = WO->getRHS();
-  ValueLatticeElement L = getValueState(LHS);
-  ValueLatticeElement R = getValueState(RHS);
+  Type *Ty = LHS->getType();
+
   addAdditionalUser(LHS, &EVI);
   addAdditionalUser(RHS, &EVI);
-  if (L.isUnknownOrUndef() || R.isUnknownOrUndef())
-    return; // Wait to resolve.
 
-  Type *Ty = LHS->getType();
+  const ValueLatticeElement &L = getValueState(LHS);
+  if (L.isUnknownOrUndef())
+    return; // Wait to resolve.
   ConstantRange LR = L.asConstantRange(Ty, /*UndefAllowed=*/false);
+
+  const ValueLatticeElement &R = getValueState(RHS);
+  if (R.isUnknownOrUndef())
+    return; // Wait to resolve.
+
   ConstantRange RR = R.asConstantRange(Ty, /*UndefAllowed=*/false);
   if (Idx == 0) {
     ConstantRange Res = LR.binaryOp(WO->getBinaryOp(), RR);
-    mergeInValue(&EVI, ValueLatticeElement::getRange(Res));
+    mergeInValue(ValueState[&EVI], &EVI, ValueLatticeElement::getRange(Res));
   } else {
     assert(Idx == 1 && "Index can only be 0 or 1");
     ConstantRange NWRegion = ConstantRange::makeGuaranteedNoWrapRegion(
@@ -1560,7 +1575,7 @@ void SCCPInstVisitor::visitExtractValueInst(ExtractValueInst &EVI) {
     if (auto *WO = dyn_cast<WithOverflowInst>(AggVal))
       return handleExtractOfWithOverflow(EVI, WO, i);
     ValueLatticeElement EltVal = getStructValueState(AggVal, i);
-    mergeInValue(getValueState(&EVI), &EVI, EltVal);
+    mergeInValue(ValueState[&EVI], &EVI, EltVal);
   } else {
     // Otherwise, must be extracting from an array.
     return (void)markOverdefined(&EVI);
@@ -1616,14 +1631,18 @@ void SCCPInstVisitor::visitSelectInst(SelectInst &I) {
   if (ValueState[&I].isOverdefined())
     return (void)markOverdefined(&I);
 
-  ValueLatticeElement CondValue = getValueState(I.getCondition());
+  const ValueLatticeElement &CondValue = getValueState(I.getCondition());
   if (CondValue.isUnknownOrUndef())
     return;
 
   if (ConstantInt *CondCB =
           getConstantInt(CondValue, I.getCondition()->getType())) {
     Value *OpVal = CondCB->isZero() ? I.getFalseValue() : I.getTrueValue();
-    mergeInValue(&I, getValueState(OpVal));
+    const ValueLatticeElement &OpValState = getValueState(OpVal);
+    // Safety: ValueState[&I] doesn't invalidate OpValState since it is already
+    // in the map.
+    assert(ValueState.contains(&I) && "&I is not in ValueState map.");
+    mergeInValue(ValueState[&I], &I, OpValState);
     return;
   }
 
@@ -1721,7 +1740,7 @@ void SCCPInstVisitor::visitBinaryOperator(Instruction &I) {
       // being a special floating value.
       ValueLatticeElement NewV;
       NewV.markConstant(C, /*MayIncludeUndef=*/true);
-      return (void)mergeInValue(&I, NewV);
+      return (void)mergeInValue(ValueState[&I], &I, NewV);
     }
   }
 
@@ -1741,7 +1760,7 @@ void SCCPInstVisitor::visitBinaryOperator(Instruction &I) {
     R = A.overflowingBinaryOp(BO->getOpcode(), B, OBO->getNoWrapKind());
   else
     R = A.binaryOp(BO->getOpcode(), B);
-  mergeInValue(&I, ValueLatticeElement::getRange(R));
+  mergeInValue(ValueState[&I], &I, ValueLatticeElement::getRange(R));
 
   // TODO: Currently we do not exploit special values that produce something
   // better than overdefined with an overdefined operand for vector or floating
@@ -1767,7 +1786,7 @@ void SCCPInstVisitor::visitCmpInst(CmpInst &I) {
   if (C) {
     ValueLatticeElement CV;
     CV.markConstant(C);
-    mergeInValue(&I, CV);
+    mergeInValue(ValueState[&I], &I, CV);
     return;
   }
 
@@ -1802,7 +1821,7 @@ void SCCPInstVisitor::visitGetElementPtrInst(GetElementPtrInst &I) {
   Operands.reserve(I.getNumOperands());
 
   for (unsigned i = 0, e = I.getNumOperands(); i != e; ++i) {
-    ValueLatticeElement State = getValueState(I.getOperand(i));
+    const ValueLatticeElement &State = getValueState(I.getOperand(i));
     if (State.isUnknownOrUndef())
       return; // Operands are not resolved yet.
 
@@ -1881,14 +1900,13 @@ void SCCPInstVisitor::visitLoadInst(LoadInst &I) {
   if (ValueState[&I].isOverdefined())
     return (void)markOverdefined(&I);
 
-  ValueLatticeElement PtrVal = getValueState(I.getOperand(0));
+  const ValueLatticeElement &PtrVal = getValueState(I.getOperand(0));
   if (PtrVal.isUnknownOrUndef())
     return; // The pointer is not resolved yet!
 
-  ValueLatticeElement &IV = ValueState[&I];
-
   if (SCCPSolver::isConstant(PtrVal)) {
     Constant *Ptr = getConstant(PtrVal, I.getOperand(0)->getType());
+    ValueLatticeElement &IV = ValueState[&I];
 
     // load null is undefined.
     if (isa<ConstantPointerNull>(Ptr)) {
@@ -1916,7 +1934,7 @@ void SCCPInstVisitor::visitLoadInst(LoadInst &I) {
   }
 
   // Fall back to metadata.
-  mergeInValue(&I, getValueFromMetadata(&I));
+  mergeInValue(ValueState[&I], &I, getValueFromMetadata(&I));
 }
 
 void SCCPInstVisitor::visitCallBase(CallBase &CB) {
@@ -1944,7 +1962,7 @@ void SCCPInstVisitor::handleCallOverdefined(CallBase &CB) {
         return markOverdefined(&CB); // Can't handle struct args.
       if (A.get()->getType()->isMetadataTy())
         continue;                    // Carried in CB, not allowed in Operands.
-      ValueLatticeElement State = getValueState(A);
+      const ValueLatticeElement &State = getValueState(A);
 
       if (State.isUnknownOrUndef())
         return; // Operands are not resolved yet.
@@ -1964,7 +1982,7 @@ void SCCPInstVisitor::handleCallOverdefined(CallBase &CB) {
   }
 
   // Fall back to metadata.
-  mergeInValue(&CB, getValueFromMetadata(&CB));
+  mergeInValue(ValueState[&CB], &CB, getValueFromMetadata(&CB));
 }
 
 void SCCPInstVisitor::handleCallArguments(CallBase &CB) {
@@ -1992,10 +2010,11 @@ void SCCPInstVisitor::handleCallArguments(CallBase &CB) {
           mergeInValue(getStructValueState(&*AI, i), &*AI, CallArg,
                        getMaxWidenStepsOpts());
         }
-      } else
-        mergeInValue(&*AI,
-                     getValueState(*CAI).intersect(getArgAttributeVL(&*AI)),
-                     getMaxWidenStepsOpts());
+      } else {
+        ValueLatticeElement CallArg =
+            getValueState(*CAI).intersect(getArgAttributeVL(&*AI));
+        mergeInValue(ValueState[&*AI], &*AI, CallArg, getMaxWidenStepsOpts());
+      }
     }
   }
 }
@@ -2076,7 +2095,8 @@ void SCCPInstVisitor::handleCallResult(CallBase &CB) {
     if (II->getIntrinsicID() == Intrinsic::vscale) {
       unsigned BitWidth = CB.getType()->getScalarSizeInBits();
       const ConstantRange Result = getVScaleRange(II->getFunction(), BitWidth);
-      return (void)mergeInValue(II, ValueLatticeElement::getRange(Result));
+      return (void)mergeInValue(ValueState[II], II,
+                                ValueLatticeElement::getRange(Result));
     }
 
     if (ConstantRange::isIntrinsicSupported(II->getIntrinsicID())) {
@@ -2094,7 +2114,8 @@ void SCCPInstVisitor::handleCallResult(CallBase &CB) {
 
       ConstantRange Result =
           ConstantRange::intrinsic(II->getIntrinsicID(), OpRanges);
-      return (void)mergeInValue(II, ValueLatticeElement::getRange(Result));
+      return (void)mergeInValue(ValueState[II], II,
+                                ValueLatticeElement::getRange(Result));
     }
   }
 
@@ -2121,10 +2142,25 @@ void SCCPInstVisitor::handleCallResult(CallBase &CB) {
       return handleCallOverdefined(CB); // Not tracking this callee.
 
     // If so, propagate the return value of the callee into this call result.
-    mergeInValue(&CB, TFRVI->second, getMaxWidenStepsOpts());
+    mergeInValue(ValueState[&CB], &CB, TFRVI->second, getMaxWidenStepsOpts());
   }
 }
 
+bool SCCPInstVisitor::isInstFullyOverDefined(Instruction &Inst) {
+  // For structure Type, we handle each member separately.
+  // A structure object won't be considered as overdefined when
+  // there is at least one member that is not overdefined.
+  if (StructType *STy = dyn_cast<StructType>(Inst.getType())) {
+    for (unsigned i = 0, e = STy->getNumElements(); i < e; ++i) {
+      if (!getStructValueState(&Inst, i).isOverdefined())
+        return false;
+    }
+    return true;
+  }
+
+  return getValueState(&Inst).isOverdefined();
+}
+
 void SCCPInstVisitor::solve() {
   // Process the work lists until they are empty!
   while (!BBWorkList.empty() || !InstWorkList.empty()) {
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
index 0e0b042..84d2ea6 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.h
+++ b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -407,6 +407,10 @@ public:
   VPBasicBlock *getParent() { return Parent; }
   const VPBasicBlock *getParent() const { return Parent; }
 
+  /// \return the VPRegionBlock which the recipe belongs to.
+  VPRegionBlock *getRegion();
+  const VPRegionBlock *getRegion() const;
+
   /// The method which generates the output IR instructions that correspond to
   /// this VPRecipe, thereby "executing" the VPlan.
   virtual void execute(VPTransformState &State) = 0;
@@ -4075,6 +4079,14 @@ public:
   }
 };
 
+inline VPRegionBlock *VPRecipeBase::getRegion() {
+  return getParent()->getParent();
+}
+
+inline const VPRegionBlock *VPRecipeBase::getRegion() const {
+  return getParent()->getParent();
+}
+
 /// VPlan models a candidate for vectorization, encoding various decisions take
 /// to produce efficient output IR, including which branches, basic-blocks and
 /// output IR instructions to generate, and their cost. VPlan holds a
diff --git a/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp b/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
index f413c63..7e074c1 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanAnalysis.cpp
@@ -377,7 +377,7 @@ bool VPDominatorTree::properlyDominates(const VPRecipeBase *A,
 
 #ifndef NDEBUG
   auto GetReplicateRegion = [](VPRecipeBase *R) -> VPRegionBlock * {
-    auto *Region = dyn_cast_or_null<VPRegionBlock>(R->getParent()->getParent());
+    VPRegionBlock *Region = R->getRegion();
     if (Region && Region->isReplicator()) {
       assert(Region->getNumSuccessors() == 1 &&
              Region->getNumPredecessors() == 1 && "Expected SESE region!");
diff --git a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
index 7a98c75..d1e67e6b 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
@@ -2352,7 +2352,7 @@ bool VPWidenIntOrFpInductionRecipe::isCanonical() const {
     return false;
   auto *StepC = dyn_cast<ConstantInt>(getStepValue()->getLiveInIRValue());
   auto *StartC = dyn_cast<ConstantInt>(getStartValue()->getLiveInIRValue());
-  auto *CanIV = getParent()->getParent()->getCanonicalIV();
+  auto *CanIV = getRegion()->getCanonicalIV();
   return StartC && StartC->isZero() && StepC && StepC->isOne() &&
          getScalarType() == CanIV->getScalarType();
 }
@@ -3076,7 +3076,7 @@ static void scalarizeInstruction(const Instruction *Instr,
     State.AC->registerAssumption(II);
 
   assert(
-      (RepRecipe->getParent()->getParent() ||
+      (RepRecipe->getRegion() ||
        !RepRecipe->getParent()->getPlan()->getVectorLoopRegion() ||
        all_of(RepRecipe->operands(),
               [](VPValue *Op) { return Op->isDefinedOutsideLoopRegions(); })) &&
@@ -3268,7 +3268,7 @@ InstructionCost VPReplicateRecipe::computeCost(ElementCount VF,
                                               to_vector(operands()), VF);
     // If the recipe is not predicated (i.e. not in a replicate region), return
     // the scalar cost. Otherwise handle predicated cost.
-    if (!getParent()->getParent()->isReplicator())
+    if (!getRegion()->isReplicator())
       return ScalarCost;
 
     // Account for the phi nodes that we will create.
@@ -3284,7 +3284,7 @@ InstructionCost VPReplicateRecipe::computeCost(ElementCount VF,
   case Instruction::Store: {
     // TODO: See getMemInstScalarizationCost for how to handle replicating and
     // predicated cases.
-    const VPRegionBlock *ParentRegion = getParent()->getParent();
+    const VPRegionBlock *ParentRegion = getRegion();
     if (ParentRegion && ParentRegion->isReplicator())
       break;
 
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index cae9aee8..f5f616f 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -1858,8 +1858,8 @@ static bool hoistPreviousBeforeFORUsers(VPFirstOrderRecurrencePHIRecipe *FOR,
       return nullptr;
     VPRegionBlock *EnclosingLoopRegion =
         HoistCandidate->getParent()->getEnclosingLoopRegion();
-    assert((!HoistCandidate->getParent()->getParent() ||
-            HoistCandidate->getParent()->getParent() == EnclosingLoopRegion) &&
+    assert((!HoistCandidate->getRegion() ||
+            HoistCandidate->getRegion() == EnclosingLoopRegion) &&
            "CFG in VPlan should still be flat, without replicate regions");
     // Hoist candidate was already visited, no need to hoist.
     if (!Visited.insert(HoistCandidate).second)
@@ -2898,7 +2898,7 @@ void VPlanTransforms::replaceSymbolicStrides(
   // evolution.
   auto CanUseVersionedStride = [&Plan](VPUser &U, unsigned) {
     auto *R = cast<VPRecipeBase>(&U);
-    return R->getParent()->getParent() ||
+    return R->getRegion() ||
            R->getParent() == Plan.getVectorLoopRegion()->getSinglePredecessor();
   };
   ValueToSCEVMapTy RewriteMap;
@@ -3803,8 +3803,7 @@ void VPlanTransforms::materializeBuildVectors(VPlan &Plan) {
         continue;
       auto *DefR = cast<VPRecipeWithIRFlags>(&R);
       auto UsesVectorOrInsideReplicateRegion = [DefR, LoopRegion](VPUser *U) {
-        VPRegionBlock *ParentRegion =
-            cast<VPRecipeBase>(U)->getParent()->getParent();
+        VPRegionBlock *ParentRegion = cast<VPRecipeBase>(U)->getRegion();
         return !U->usesScalars(DefR) || ParentRegion != LoopRegion;
       };
       if ((isa<VPReplicateRecipe>(DefR) &&
diff --git a/llvm/lib/Transforms/Vectorize/VPlanUtils.h b/llvm/lib/Transforms/Vectorize/VPlanUtils.h
index cf95ac0..9a2497e 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanUtils.h
+++ b/llvm/lib/Transforms/Vectorize/VPlanUtils.h
@@ -64,7 +64,7 @@ inline bool isSingleScalar(const VPValue *VPV) {
     return true;
 
   if (auto *Rep = dyn_cast<VPReplicateRecipe>(VPV)) {
-    const VPRegionBlock *RegionOfR = Rep->getParent()->getParent();
+    const VPRegionBlock *RegionOfR = Rep->getRegion();
     // Don't consider recipes in replicate regions as uniform yet; their first
     // lane cannot be accessed when executing the replicate region for other
     // lanes.
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i32.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i32.ll
index 7714c03..5171403 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i32.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i32.ll
@@ -140,6 +140,7 @@ define amdgpu_cs i32 @branch_divergent_ballot_eq_zero_non_compare(i32 %v) {
 ; CHECK-NEXT:    v_and_b32_e32 v0, 1, v0
 ; CHECK-NEXT:    v_cmp_ne_u32_e32 vcc_lo, 0, v0
 ; CHECK-NEXT:    s_and_b32 s0, vcc_lo, exec_lo
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cbranch_scc0 .LBB9_2
 ; CHECK-NEXT:  ; %bb.1: ; %false
 ; CHECK-NEXT:    s_mov_b32 s0, 33
@@ -344,6 +345,7 @@ define amdgpu_cs i32 @branch_divergent_ballot_eq_zero_and(i32 %v1, i32 %v2) {
 ; CHECK-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 12, v0
 ; CHECK-NEXT:    v_cmp_lt_u32_e64 s0, 34, v1
 ; CHECK-NEXT:    s_and_b32 s0, vcc_lo, s0
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cbranch_scc0 .LBB17_2
 ; CHECK-NEXT:  ; %bb.1: ; %false
 ; CHECK-NEXT:    s_mov_b32 s0, 33
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i64.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i64.ll
index 7b81669..7b01f13 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i64.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i64.ll
@@ -143,6 +143,7 @@ define amdgpu_cs i32 @branch_divergent_ballot_eq_zero_non_compare(i32 %v) {
 ; CHECK-NEXT:    v_and_b32_e32 v0, 1, v0
 ; CHECK-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; CHECK-NEXT:    s_and_b64 s[0:1], vcc, exec
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_cbranch_scc0 .LBB9_2
 ; CHECK-NEXT:  ; %bb.1: ; %false
 ; CHECK-NEXT:    s_mov_b32 s0, 33
@@ -347,6 +348,7 @@ define amdgpu_cs i32 @branch_divergent_ballot_eq_zero_and(i32 %v1, i32 %v2) {
 ; CHECK-NEXT:    v_cmp_gt_u32_e32 vcc, 12, v0
 ; CHECK-NEXT:    v_cmp_lt_u32_e64 s[0:1], 34, v1
 ; CHECK-NEXT:    s_and_b64 s[0:1], vcc, s[0:1]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_cbranch_scc0 .LBB17_2
 ; CHECK-NEXT:  ; %bb.1: ; %false
 ; CHECK-NEXT:    s_mov_b32 s0, 33
diff --git a/llvm/test/CodeGen/AMDGPU/abs_i32.ll b/llvm/test/CodeGen/AMDGPU/abs_i32.ll
new file mode 100644
index 0000000..b53047f
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/abs_i32.ll
@@ -0,0 +1,92 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 < %s | FileCheck -check-prefixes=GFX9 %s
+; RUN: llc -mtriple=r600 -mcpu=cypress < %s | FileCheck -check-prefixes=R600 %s
+
+define amdgpu_kernel void @abs_v1(ptr addrspace(1) %out, i32 %arg) {
+; GFX9-LABEL: abs_v1:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x8
+; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    s_abs_i32 s2, s2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s2
+; GFX9-NEXT:    global_store_dword v0, v1, s[0:1]
+; GFX9-NEXT:    s_endpgm
+;
+; R600-LABEL: abs_v1:
+; R600:       ; %bb.0:
+; R600-NEXT:    ALU 4, @4, KC0[CB0:0-32], KC1[]
+; R600-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
+; R600-NEXT:    CF_END
+; R600-NEXT:    PAD
+; R600-NEXT:    ALU clause starting at 4:
+; R600-NEXT:     MOV * T0.W, KC0[2].Z,
+; R600-NEXT:     SUB_INT * T1.W, 0.0, PV.W,
+; R600-NEXT:     MAX_INT T0.X, T0.W, PV.W,
+; R600-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
+; R600-NEXT:    2(2.802597e-45), 0(0.000000e+00)
+  %res = call i32 @llvm.abs.i32(i32 %arg, i1 false)
+  store i32 %res, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+define amdgpu_kernel void @abs_v2(ptr addrspace(1) %out, i32 %arg) {
+; GFX9-LABEL: abs_v2:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x8
+; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    s_abs_i32 s2, s2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s2
+; GFX9-NEXT:    global_store_dword v0, v1, s[0:1]
+; GFX9-NEXT:    s_endpgm
+;
+; R600-LABEL: abs_v2:
+; R600:       ; %bb.0:
+; R600-NEXT:    ALU 3, @4, KC0[CB0:0-32], KC1[]
+; R600-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
+; R600-NEXT:    CF_END
+; R600-NEXT:    PAD
+; R600-NEXT:    ALU clause starting at 4:
+; R600-NEXT:     SUB_INT * T0.W, 0.0, KC0[2].Z,
+; R600-NEXT:     MAX_INT T0.X, KC0[2].Z, PV.W,
+; R600-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
+; R600-NEXT:    2(2.802597e-45), 0(0.000000e+00)
+  %neg = sub i32 0, %arg
+  %cond = icmp sgt i32 %arg, %neg
+  %res = select i1 %cond, i32 %arg, i32 %neg
+  store i32 %res, ptr addrspace(1) %out, align 4
+  ret void
+}
+
+define amdgpu_kernel void @abs_v3(ptr addrspace(1) %out, i32 %arg) {
+; GFX9-LABEL: abs_v3:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x8
+; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
+; GFX9-NEXT:    v_mov_b32_e32 v0, 0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    s_abs_i32 s2, s2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s2
+; GFX9-NEXT:    global_store_dword v0, v1, s[0:1]
+; GFX9-NEXT:    s_endpgm
+;
+; R600-LABEL: abs_v3:
+; R600:       ; %bb.0:
+; R600-NEXT:    ALU 3, @4, KC0[CB0:0-32], KC1[]
+; R600-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
+; R600-NEXT:    CF_END
+; R600-NEXT:    PAD
+; R600-NEXT:    ALU clause starting at 4:
+; R600-NEXT:     SUB_INT * T0.W, 0.0, KC0[2].Z,
+; R600-NEXT:     MAX_INT T0.X, PV.W, KC0[2].Z,
+; R600-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
+; R600-NEXT:    2(2.802597e-45), 0(0.000000e+00)
+  %neg = sub i32 0, %arg
+  %cond = icmp sgt i32 %neg, %arg
+  %res = select i1 %cond, i32 %neg, i32 %arg
+  store i32 %res, ptr addrspace(1) %out, align 4
+  ret void
+}
diff --git a/llvm/test/CodeGen/AMDGPU/addsub64_carry.ll b/llvm/test/CodeGen/AMDGPU/addsub64_carry.ll
index 8088c1b..b72eba8 100644
--- a/llvm/test/CodeGen/AMDGPU/addsub64_carry.ll
+++ b/llvm/test/CodeGen/AMDGPU/addsub64_carry.ll
@@ -180,7 +180,11 @@ define amdgpu_ps %struct.uint96 @s_add64_32(i64 inreg %val64A, i64 inreg %val64B
 ; CHECK-LABEL: s_add64_32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_add_u32 s0, s0, s2
+; CHECK-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[6:7], 0
 ; CHECK-NEXT:    s_addc_u32 s1, s1, s3
+; CHECK-NEXT:    s_cselect_b64 s[2:3], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; CHECK-NEXT:    s_addc_u32 s2, s4, 0
 ; CHECK-NEXT:    ; return to shader part epilog
   %sum64 = add i64 %val64A, %val64B
@@ -195,10 +199,14 @@ define amdgpu_ps %struct.uint96 @s_add64_32(i64 inreg %val64A, i64 inreg %val64B
 define amdgpu_ps <2 x i64> @s_uadd_v2i64(<2 x i64> inreg %val0, <2 x i64> inreg %val1, ptr %ptrval) {
 ; CHECK-LABEL: s_uadd_v2i64:
 ; CHECK:       ; %bb.0:
-; CHECK-NEXT:    s_add_u32 s6, s2, s6
-; CHECK-NEXT:    s_addc_u32 s7, s3, s7
+; CHECK-NEXT:    s_add_u32 s10, s2, s6
+; CHECK-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[8:9], 0
+; CHECK-NEXT:    s_addc_u32 s8, s3, s7
 ; CHECK-NEXT:    s_cselect_b64 s[2:3], -1, 0
 ; CHECK-NEXT:    s_add_u32 s0, s0, s4
+; CHECK-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[6:7], 0
 ; CHECK-NEXT:    s_addc_u32 s1, s1, s5
 ; CHECK-NEXT:    v_mov_b32_e32 v2, s0
 ; CHECK-NEXT:    v_mov_b32_e32 v3, s1
@@ -207,8 +215,8 @@ define amdgpu_ps <2 x i64> @s_uadd_v2i64(<2 x i64> inreg %val0, <2 x i64> inreg
 ; CHECK-NEXT:    v_cndmask_b32_e64 v7, 0, -1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v7
 ; CHECK-NEXT:    v_readfirstlane_b32 s2, v6
-; CHECK-NEXT:    v_mov_b32_e32 v4, s6
-; CHECK-NEXT:    v_mov_b32_e32 v5, s7
+; CHECK-NEXT:    v_mov_b32_e32 v4, s10
+; CHECK-NEXT:    v_mov_b32_e32 v5, s8
 ; CHECK-NEXT:    s_mov_b32 s1, s0
 ; CHECK-NEXT:    s_mov_b32 s3, s2
 ; CHECK-NEXT:    flat_store_dwordx4 v[0:1], v[2:5]
@@ -225,10 +233,14 @@ define amdgpu_ps <2 x i64> @s_uadd_v2i64(<2 x i64> inreg %val0, <2 x i64> inreg
 define amdgpu_ps <2 x i64> @s_usub_v2i64(<2 x i64> inreg %val0, <2 x i64> inreg %val1, ptr %ptrval) {
 ; CHECK-LABEL: s_usub_v2i64:
 ; CHECK:       ; %bb.0:
-; CHECK-NEXT:    s_sub_u32 s6, s2, s6
-; CHECK-NEXT:    s_subb_u32 s7, s3, s7
+; CHECK-NEXT:    s_sub_u32 s10, s2, s6
+; CHECK-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[8:9], 0
+; CHECK-NEXT:    s_subb_u32 s8, s3, s7
 ; CHECK-NEXT:    s_cselect_b64 s[2:3], -1, 0
 ; CHECK-NEXT:    s_sub_u32 s0, s0, s4
+; CHECK-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[6:7], 0
 ; CHECK-NEXT:    s_subb_u32 s1, s1, s5
 ; CHECK-NEXT:    v_mov_b32_e32 v2, s0
 ; CHECK-NEXT:    v_mov_b32_e32 v3, s1
@@ -237,8 +249,8 @@ define amdgpu_ps <2 x i64> @s_usub_v2i64(<2 x i64> inreg %val0, <2 x i64> inreg
 ; CHECK-NEXT:    v_cndmask_b32_e64 v7, 0, -1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v7
 ; CHECK-NEXT:    v_readfirstlane_b32 s2, v6
-; CHECK-NEXT:    v_mov_b32_e32 v4, s6
-; CHECK-NEXT:    v_mov_b32_e32 v5, s7
+; CHECK-NEXT:    v_mov_b32_e32 v4, s10
+; CHECK-NEXT:    v_mov_b32_e32 v5, s8
 ; CHECK-NEXT:    s_mov_b32 s1, s0
 ; CHECK-NEXT:    s_mov_b32 s3, s2
 ; CHECK-NEXT:    flat_store_dwordx4 v[0:1], v[2:5]
@@ -256,6 +268,8 @@ define amdgpu_ps i64 @s_uadd_i64(i64 inreg %val0, i64 inreg %val1, ptr %ptrval)
 ; CHECK-LABEL: s_uadd_i64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_add_u32 s0, s0, s2
+; CHECK-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[4:5], 0
 ; CHECK-NEXT:    s_addc_u32 s1, s1, s3
 ; CHECK-NEXT:    v_mov_b32_e32 v2, s0
 ; CHECK-NEXT:    v_mov_b32_e32 v3, s1
@@ -278,6 +292,8 @@ define amdgpu_ps i64 @s_uadd_p1(i64 inreg %val0, i64 inreg %val1, ptr %ptrval) {
 ; CHECK-LABEL: s_uadd_p1:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_add_u32 s0, s0, 1
+; CHECK-NEXT:    s_cselect_b64 s[2:3], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; CHECK-NEXT:    s_addc_u32 s1, s1, 0
 ; CHECK-NEXT:    v_mov_b32_e32 v2, s0
 ; CHECK-NEXT:    v_mov_b32_e32 v3, s1
@@ -323,6 +339,8 @@ define amdgpu_ps i64 @s_usub_p1(i64 inreg %val0, i64 inreg %val1, ptr %ptrval) {
 ; CHECK-LABEL: s_usub_p1:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_sub_u32 s0, s0, 1
+; CHECK-NEXT:    s_cselect_b64 s[2:3], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; CHECK-NEXT:    s_subb_u32 s1, s1, 0
 ; CHECK-NEXT:    v_mov_b32_e32 v2, s0
 ; CHECK-NEXT:    v_mov_b32_e32 v3, s1
@@ -345,6 +363,8 @@ define amdgpu_ps i64 @s_usub_n1(i64 inreg %val0, i64 inreg %val1, ptr %ptrval) {
 ; CHECK-LABEL: s_usub_n1:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_sub_u32 s0, s0, -1
+; CHECK-NEXT:    s_cselect_b64 s[2:3], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; CHECK-NEXT:    s_subb_u32 s1, s1, -1
 ; CHECK-NEXT:    v_mov_b32_e32 v2, s0
 ; CHECK-NEXT:    v_mov_b32_e32 v3, s1
diff --git a/llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll b/llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll
index 51df8c3..948811e 100644
--- a/llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll
+++ b/llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll
@@ -7821,9 +7821,10 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX6-NEXT:    s_addc_u32 s15, 0, s16
 ; GFX6-NEXT:    s_add_u32 s16, s0, s1
 ; GFX6-NEXT:    v_mov_b32_e32 v0, s16
-; GFX6-NEXT:    v_mul_hi_u32 v0, s12, v0
 ; GFX6-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GFX6-NEXT:    v_mul_hi_u32 v0, s12, v0
 ; GFX6-NEXT:    s_or_b32 s0, s0, s1
+; GFX6-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX6-NEXT:    s_addc_u32 s14, s14, s15
 ; GFX6-NEXT:    s_mul_i32 s0, s12, s14
 ; GFX6-NEXT:    v_readfirstlane_b32 s1, v0
@@ -7854,6 +7855,7 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX6-NEXT:    s_add_u32 s15, s16, s0
 ; GFX6-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GFX6-NEXT:    s_or_b32 s0, s0, s1
+; GFX6-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX6-NEXT:    s_addc_u32 s14, s14, s12
 ; GFX6-NEXT:    s_ashr_i32 s12, s7, 31
 ; GFX6-NEXT:    s_add_u32 s0, s6, s12
@@ -7879,50 +7881,52 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX6-NEXT:    v_readfirstlane_b32 s4, v0
 ; GFX6-NEXT:    s_addc_u32 s4, s4, 0
 ; GFX6-NEXT:    s_mul_i32 s14, s7, s14
-; GFX6-NEXT:    s_add_u32 s16, s1, s14
-; GFX6-NEXT:    v_mov_b32_e32 v0, s16
+; GFX6-NEXT:    s_add_u32 s14, s1, s14
+; GFX6-NEXT:    v_mov_b32_e32 v0, s14
 ; GFX6-NEXT:    v_mul_hi_u32 v0, s10, v0
-; GFX6-NEXT:    s_addc_u32 s17, 0, s4
+; GFX6-NEXT:    s_addc_u32 s15, 0, s4
 ; GFX6-NEXT:    s_mov_b32 s1, s5
-; GFX6-NEXT:    s_mul_i32 s4, s10, s17
+; GFX6-NEXT:    s_mul_i32 s4, s10, s15
 ; GFX6-NEXT:    v_readfirstlane_b32 s5, v0
 ; GFX6-NEXT:    s_add_i32 s4, s5, s4
-; GFX6-NEXT:    s_mul_i32 s5, s11, s16
-; GFX6-NEXT:    s_add_i32 s18, s4, s5
-; GFX6-NEXT:    s_sub_i32 s14, s7, s18
-; GFX6-NEXT:    s_mul_i32 s4, s10, s16
+; GFX6-NEXT:    s_mul_i32 s5, s11, s14
+; GFX6-NEXT:    s_add_i32 s16, s4, s5
+; GFX6-NEXT:    s_sub_i32 s17, s7, s16
+; GFX6-NEXT:    s_mul_i32 s4, s10, s14
 ; GFX6-NEXT:    s_sub_u32 s6, s6, s4
 ; GFX6-NEXT:    s_cselect_b64 s[4:5], -1, 0
-; GFX6-NEXT:    s_or_b32 s15, s4, s5
-; GFX6-NEXT:    s_subb_u32 s19, s14, s11
-; GFX6-NEXT:    s_sub_u32 s20, s6, s10
-; GFX6-NEXT:    s_cselect_b64 s[14:15], -1, 0
-; GFX6-NEXT:    s_or_b32 s14, s14, s15
-; GFX6-NEXT:    s_subb_u32 s14, s19, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s14, s11
-; GFX6-NEXT:    s_cselect_b32 s15, -1, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s20, s10
-; GFX6-NEXT:    s_cselect_b32 s19, -1, 0
-; GFX6-NEXT:    s_cmp_eq_u32 s14, s11
-; GFX6-NEXT:    s_cselect_b32 s14, s19, s15
-; GFX6-NEXT:    s_add_u32 s15, s16, 1
-; GFX6-NEXT:    s_addc_u32 s19, s17, 0
-; GFX6-NEXT:    s_add_u32 s20, s16, 2
-; GFX6-NEXT:    s_addc_u32 s21, s17, 0
-; GFX6-NEXT:    s_cmp_lg_u32 s14, 0
-; GFX6-NEXT:    s_cselect_b32 s14, s20, s15
-; GFX6-NEXT:    s_cselect_b32 s15, s21, s19
+; GFX6-NEXT:    s_or_b32 s18, s4, s5
+; GFX6-NEXT:    s_cmp_lg_u32 s18, 0
+; GFX6-NEXT:    s_subb_u32 s17, s17, s11
+; GFX6-NEXT:    s_sub_u32 s19, s6, s10
+; GFX6-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GFX6-NEXT:    s_or_b32 s4, s4, s5
-; GFX6-NEXT:    s_subb_u32 s4, s7, s18
+; GFX6-NEXT:    s_cmp_lg_u32 s4, 0
+; GFX6-NEXT:    s_subb_u32 s4, s17, 0
 ; GFX6-NEXT:    s_cmp_ge_u32 s4, s11
 ; GFX6-NEXT:    s_cselect_b32 s5, -1, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s6, s10
-; GFX6-NEXT:    s_cselect_b32 s6, -1, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s19, s10
+; GFX6-NEXT:    s_cselect_b32 s17, -1, 0
 ; GFX6-NEXT:    s_cmp_eq_u32 s4, s11
-; GFX6-NEXT:    s_cselect_b32 s4, s6, s5
+; GFX6-NEXT:    s_cselect_b32 s4, s17, s5
+; GFX6-NEXT:    s_add_u32 s5, s14, 1
+; GFX6-NEXT:    s_addc_u32 s17, s15, 0
+; GFX6-NEXT:    s_add_u32 s19, s14, 2
+; GFX6-NEXT:    s_addc_u32 s20, s15, 0
 ; GFX6-NEXT:    s_cmp_lg_u32 s4, 0
-; GFX6-NEXT:    s_cselect_b32 s5, s15, s17
-; GFX6-NEXT:    s_cselect_b32 s4, s14, s16
+; GFX6-NEXT:    s_cselect_b32 s4, s19, s5
+; GFX6-NEXT:    s_cselect_b32 s5, s20, s17
+; GFX6-NEXT:    s_cmp_lg_u32 s18, 0
+; GFX6-NEXT:    s_subb_u32 s7, s7, s16
+; GFX6-NEXT:    s_cmp_ge_u32 s7, s11
+; GFX6-NEXT:    s_cselect_b32 s16, -1, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s6, s10
+; GFX6-NEXT:    s_cselect_b32 s6, -1, 0
+; GFX6-NEXT:    s_cmp_eq_u32 s7, s11
+; GFX6-NEXT:    s_cselect_b32 s6, s6, s16
+; GFX6-NEXT:    s_cmp_lg_u32 s6, 0
+; GFX6-NEXT:    s_cselect_b32 s5, s5, s15
+; GFX6-NEXT:    s_cselect_b32 s4, s4, s14
 ; GFX6-NEXT:    s_xor_b64 s[6:7], s[12:13], s[8:9]
 ; GFX6-NEXT:    s_xor_b64 s[4:5], s[4:5], s[6:7]
 ; GFX6-NEXT:    s_sub_u32 s4, s4, s6
@@ -7945,8 +7949,8 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v0, s8
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v1, s9
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
-; GFX9-NEXT:    s_sub_u32 s4, 0, s8
-; GFX9-NEXT:    s_subb_u32 s5, 0, s9
+; GFX9-NEXT:    s_sub_u32 s10, 0, s8
+; GFX9-NEXT:    s_subb_u32 s11, 0, s9
 ; GFX9-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GFX9-NEXT:    v_rcp_f32_e32 v1, v0
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 0
@@ -7956,52 +7960,56 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX9-NEXT:    v_madmk_f32 v1, v2, 0xcf800000, v1
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v2, v2
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v1, v1
-; GFX9-NEXT:    v_readfirstlane_b32 s10, v2
-; GFX9-NEXT:    v_readfirstlane_b32 s11, v1
-; GFX9-NEXT:    s_mul_i32 s12, s4, s10
-; GFX9-NEXT:    s_mul_hi_u32 s14, s4, s11
-; GFX9-NEXT:    s_mul_i32 s13, s5, s11
-; GFX9-NEXT:    s_add_i32 s12, s14, s12
-; GFX9-NEXT:    s_mul_i32 s15, s4, s11
-; GFX9-NEXT:    s_add_i32 s12, s12, s13
-; GFX9-NEXT:    s_mul_hi_u32 s14, s11, s15
-; GFX9-NEXT:    s_mul_i32 s16, s11, s12
-; GFX9-NEXT:    s_mul_hi_u32 s13, s11, s12
+; GFX9-NEXT:    v_readfirstlane_b32 s12, v2
+; GFX9-NEXT:    v_readfirstlane_b32 s4, v1
+; GFX9-NEXT:    s_mul_i32 s5, s10, s12
+; GFX9-NEXT:    s_mul_hi_u32 s14, s10, s4
+; GFX9-NEXT:    s_mul_i32 s13, s11, s4
+; GFX9-NEXT:    s_add_i32 s5, s14, s5
+; GFX9-NEXT:    s_mul_i32 s15, s10, s4
+; GFX9-NEXT:    s_add_i32 s5, s5, s13
+; GFX9-NEXT:    s_mul_hi_u32 s14, s4, s15
+; GFX9-NEXT:    s_mul_i32 s16, s4, s5
+; GFX9-NEXT:    s_mul_hi_u32 s13, s4, s5
 ; GFX9-NEXT:    s_add_u32 s14, s14, s16
 ; GFX9-NEXT:    s_addc_u32 s13, 0, s13
-; GFX9-NEXT:    s_mul_hi_u32 s17, s10, s15
-; GFX9-NEXT:    s_mul_i32 s15, s10, s15
+; GFX9-NEXT:    s_mul_hi_u32 s17, s12, s15
+; GFX9-NEXT:    s_mul_i32 s15, s12, s15
 ; GFX9-NEXT:    s_add_u32 s14, s14, s15
-; GFX9-NEXT:    s_mul_hi_u32 s16, s10, s12
+; GFX9-NEXT:    s_mul_hi_u32 s16, s12, s5
 ; GFX9-NEXT:    s_addc_u32 s13, s13, s17
 ; GFX9-NEXT:    s_addc_u32 s14, s16, 0
-; GFX9-NEXT:    s_mul_i32 s12, s10, s12
-; GFX9-NEXT:    s_add_u32 s12, s13, s12
+; GFX9-NEXT:    s_mul_i32 s5, s12, s5
+; GFX9-NEXT:    s_add_u32 s5, s13, s5
 ; GFX9-NEXT:    s_addc_u32 s13, 0, s14
-; GFX9-NEXT:    s_add_u32 s11, s11, s12
-; GFX9-NEXT:    s_addc_u32 s10, s10, s13
-; GFX9-NEXT:    s_mul_i32 s12, s4, s10
-; GFX9-NEXT:    s_mul_hi_u32 s13, s4, s11
-; GFX9-NEXT:    s_add_i32 s12, s13, s12
-; GFX9-NEXT:    s_mul_i32 s5, s5, s11
-; GFX9-NEXT:    s_add_i32 s12, s12, s5
-; GFX9-NEXT:    s_mul_i32 s4, s4, s11
-; GFX9-NEXT:    s_mul_hi_u32 s13, s10, s4
-; GFX9-NEXT:    s_mul_i32 s14, s10, s4
-; GFX9-NEXT:    s_mul_i32 s16, s11, s12
-; GFX9-NEXT:    s_mul_hi_u32 s4, s11, s4
-; GFX9-NEXT:    s_mul_hi_u32 s15, s11, s12
-; GFX9-NEXT:    s_add_u32 s4, s4, s16
+; GFX9-NEXT:    s_add_u32 s14, s4, s5
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX9-NEXT:    s_addc_u32 s12, s12, s13
+; GFX9-NEXT:    s_mul_i32 s4, s10, s12
+; GFX9-NEXT:    s_mul_hi_u32 s5, s10, s14
+; GFX9-NEXT:    s_add_i32 s4, s5, s4
+; GFX9-NEXT:    s_mul_i32 s11, s11, s14
+; GFX9-NEXT:    s_add_i32 s4, s4, s11
+; GFX9-NEXT:    s_mul_i32 s10, s10, s14
+; GFX9-NEXT:    s_mul_hi_u32 s11, s12, s10
+; GFX9-NEXT:    s_mul_i32 s13, s12, s10
+; GFX9-NEXT:    s_mul_i32 s16, s14, s4
+; GFX9-NEXT:    s_mul_hi_u32 s10, s14, s10
+; GFX9-NEXT:    s_mul_hi_u32 s15, s14, s4
+; GFX9-NEXT:    s_add_u32 s10, s10, s16
 ; GFX9-NEXT:    s_addc_u32 s15, 0, s15
-; GFX9-NEXT:    s_add_u32 s4, s4, s14
-; GFX9-NEXT:    s_mul_hi_u32 s5, s10, s12
-; GFX9-NEXT:    s_addc_u32 s4, s15, s13
+; GFX9-NEXT:    s_add_u32 s10, s10, s13
+; GFX9-NEXT:    s_mul_hi_u32 s5, s12, s4
+; GFX9-NEXT:    s_addc_u32 s10, s15, s11
 ; GFX9-NEXT:    s_addc_u32 s5, s5, 0
-; GFX9-NEXT:    s_mul_i32 s12, s10, s12
-; GFX9-NEXT:    s_add_u32 s4, s4, s12
-; GFX9-NEXT:    s_addc_u32 s5, 0, s5
-; GFX9-NEXT:    s_add_u32 s11, s11, s4
-; GFX9-NEXT:    s_addc_u32 s10, s10, s5
+; GFX9-NEXT:    s_mul_i32 s4, s12, s4
+; GFX9-NEXT:    s_add_u32 s4, s10, s4
+; GFX9-NEXT:    s_addc_u32 s10, 0, s5
+; GFX9-NEXT:    s_add_u32 s11, s14, s4
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX9-NEXT:    s_addc_u32 s10, s12, s10
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_ashr_i32 s4, s3, 31
 ; GFX9-NEXT:    s_add_u32 s2, s2, s4
@@ -8020,35 +8028,38 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX9-NEXT:    s_addc_u32 s11, s12, s15
 ; GFX9-NEXT:    s_addc_u32 s12, s14, 0
 ; GFX9-NEXT:    s_mul_i32 s10, s3, s10
-; GFX9-NEXT:    s_add_u32 s13, s11, s10
-; GFX9-NEXT:    s_addc_u32 s12, 0, s12
-; GFX9-NEXT:    s_mul_i32 s10, s8, s12
-; GFX9-NEXT:    s_mul_hi_u32 s11, s8, s13
+; GFX9-NEXT:    s_add_u32 s14, s11, s10
+; GFX9-NEXT:    s_addc_u32 s15, 0, s12
+; GFX9-NEXT:    s_mul_i32 s10, s8, s15
+; GFX9-NEXT:    s_mul_hi_u32 s11, s8, s14
 ; GFX9-NEXT:    s_add_i32 s10, s11, s10
-; GFX9-NEXT:    s_mul_i32 s11, s9, s13
-; GFX9-NEXT:    s_add_i32 s14, s10, s11
-; GFX9-NEXT:    s_sub_i32 s15, s3, s14
-; GFX9-NEXT:    s_mul_i32 s10, s8, s13
+; GFX9-NEXT:    s_mul_i32 s11, s9, s14
+; GFX9-NEXT:    s_add_i32 s16, s10, s11
+; GFX9-NEXT:    s_sub_i32 s12, s3, s16
+; GFX9-NEXT:    s_mul_i32 s10, s8, s14
 ; GFX9-NEXT:    s_sub_u32 s2, s2, s10
 ; GFX9-NEXT:    s_cselect_b64 s[10:11], -1, 0
-; GFX9-NEXT:    s_subb_u32 s15, s15, s9
-; GFX9-NEXT:    s_sub_u32 s16, s2, s8
-; GFX9-NEXT:    s_subb_u32 s15, s15, 0
-; GFX9-NEXT:    s_cmp_ge_u32 s15, s9
+; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
+; GFX9-NEXT:    s_subb_u32 s17, s12, s9
+; GFX9-NEXT:    s_sub_u32 s18, s2, s8
+; GFX9-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[12:13], 0
+; GFX9-NEXT:    s_subb_u32 s12, s17, 0
+; GFX9-NEXT:    s_cmp_ge_u32 s12, s9
+; GFX9-NEXT:    s_cselect_b32 s13, -1, 0
+; GFX9-NEXT:    s_cmp_ge_u32 s18, s8
 ; GFX9-NEXT:    s_cselect_b32 s17, -1, 0
-; GFX9-NEXT:    s_cmp_ge_u32 s16, s8
-; GFX9-NEXT:    s_cselect_b32 s16, -1, 0
-; GFX9-NEXT:    s_cmp_eq_u32 s15, s9
-; GFX9-NEXT:    s_cselect_b32 s15, s16, s17
-; GFX9-NEXT:    s_add_u32 s16, s13, 1
-; GFX9-NEXT:    s_addc_u32 s17, s12, 0
-; GFX9-NEXT:    s_add_u32 s18, s13, 2
-; GFX9-NEXT:    s_addc_u32 s19, s12, 0
-; GFX9-NEXT:    s_cmp_lg_u32 s15, 0
-; GFX9-NEXT:    s_cselect_b32 s15, s18, s16
-; GFX9-NEXT:    s_cselect_b32 s16, s19, s17
+; GFX9-NEXT:    s_cmp_eq_u32 s12, s9
+; GFX9-NEXT:    s_cselect_b32 s12, s17, s13
+; GFX9-NEXT:    s_add_u32 s13, s14, 1
+; GFX9-NEXT:    s_addc_u32 s17, s15, 0
+; GFX9-NEXT:    s_add_u32 s18, s14, 2
+; GFX9-NEXT:    s_addc_u32 s19, s15, 0
+; GFX9-NEXT:    s_cmp_lg_u32 s12, 0
+; GFX9-NEXT:    s_cselect_b32 s12, s18, s13
+; GFX9-NEXT:    s_cselect_b32 s13, s19, s17
 ; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT:    s_subb_u32 s3, s3, s14
+; GFX9-NEXT:    s_subb_u32 s3, s3, s16
 ; GFX9-NEXT:    s_cmp_ge_u32 s3, s9
 ; GFX9-NEXT:    s_cselect_b32 s10, -1, 0
 ; GFX9-NEXT:    s_cmp_ge_u32 s2, s8
@@ -8056,8 +8067,8 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX9-NEXT:    s_cmp_eq_u32 s3, s9
 ; GFX9-NEXT:    s_cselect_b32 s2, s2, s10
 ; GFX9-NEXT:    s_cmp_lg_u32 s2, 0
-; GFX9-NEXT:    s_cselect_b32 s3, s16, s12
-; GFX9-NEXT:    s_cselect_b32 s2, s15, s13
+; GFX9-NEXT:    s_cselect_b32 s3, s13, s15
+; GFX9-NEXT:    s_cselect_b32 s2, s12, s14
 ; GFX9-NEXT:    s_xor_b64 s[4:5], s[4:5], s[6:7]
 ; GFX9-NEXT:    s_xor_b64 s[2:3], s[2:3], s[4:5]
 ; GFX9-NEXT:    s_sub_u32 s2, s2, s4
@@ -8317,9 +8328,10 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    s_addc_u32 s17, 0, s18
 ; GFX6-NEXT:    s_add_u32 s18, s12, s13
 ; GFX6-NEXT:    v_mov_b32_e32 v0, s18
-; GFX6-NEXT:    v_mul_hi_u32 v0, s14, v0
 ; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GFX6-NEXT:    v_mul_hi_u32 v0, s14, v0
 ; GFX6-NEXT:    s_or_b32 s12, s12, s13
+; GFX6-NEXT:    s_cmp_lg_u32 s12, 0
 ; GFX6-NEXT:    s_addc_u32 s16, s16, s17
 ; GFX6-NEXT:    s_mul_i32 s12, s14, s16
 ; GFX6-NEXT:    v_readfirstlane_b32 s13, v0
@@ -8350,6 +8362,7 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    s_add_u32 s15, s18, s12
 ; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
 ; GFX6-NEXT:    s_or_b32 s12, s12, s13
+; GFX6-NEXT:    s_cmp_lg_u32 s12, 0
 ; GFX6-NEXT:    s_addc_u32 s14, s16, s14
 ; GFX6-NEXT:    s_ashr_i32 s12, s9, 31
 ; GFX6-NEXT:    s_add_u32 s8, s8, s12
@@ -8374,53 +8387,55 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    v_readfirstlane_b32 s16, v0
 ; GFX6-NEXT:    s_addc_u32 s16, s16, 0
 ; GFX6-NEXT:    s_mul_i32 s14, s9, s14
-; GFX6-NEXT:    s_add_u32 s18, s15, s14
-; GFX6-NEXT:    v_mov_b32_e32 v0, s18
+; GFX6-NEXT:    s_add_u32 s17, s15, s14
+; GFX6-NEXT:    v_mov_b32_e32 v0, s17
 ; GFX6-NEXT:    v_mul_hi_u32 v0, s6, v0
-; GFX6-NEXT:    s_addc_u32 s19, 0, s16
-; GFX6-NEXT:    s_mul_i32 s14, s6, s19
+; GFX6-NEXT:    s_addc_u32 s16, 0, s16
+; GFX6-NEXT:    s_mul_i32 s14, s6, s16
 ; GFX6-NEXT:    v_readfirstlane_b32 s15, v0
 ; GFX6-NEXT:    s_add_i32 s14, s15, s14
-; GFX6-NEXT:    s_mul_i32 s15, s7, s18
-; GFX6-NEXT:    s_add_i32 s20, s14, s15
-; GFX6-NEXT:    s_sub_i32 s16, s9, s20
-; GFX6-NEXT:    s_mul_i32 s14, s6, s18
+; GFX6-NEXT:    s_mul_i32 s15, s7, s17
+; GFX6-NEXT:    s_add_i32 s18, s14, s15
+; GFX6-NEXT:    s_sub_i32 s19, s9, s18
+; GFX6-NEXT:    s_mul_i32 s14, s6, s17
 ; GFX6-NEXT:    s_sub_u32 s8, s8, s14
 ; GFX6-NEXT:    s_cselect_b64 s[14:15], -1, 0
-; GFX6-NEXT:    s_or_b32 s17, s14, s15
-; GFX6-NEXT:    s_subb_u32 s21, s16, s7
-; GFX6-NEXT:    s_sub_u32 s22, s8, s6
-; GFX6-NEXT:    s_cselect_b64 s[16:17], -1, 0
-; GFX6-NEXT:    s_or_b32 s16, s16, s17
-; GFX6-NEXT:    s_subb_u32 s16, s21, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s16, s7
-; GFX6-NEXT:    s_cselect_b32 s17, -1, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s22, s6
-; GFX6-NEXT:    s_cselect_b32 s21, -1, 0
-; GFX6-NEXT:    s_cmp_eq_u32 s16, s7
-; GFX6-NEXT:    s_cselect_b32 s16, s21, s17
-; GFX6-NEXT:    s_add_u32 s17, s18, 1
-; GFX6-NEXT:    s_addc_u32 s21, s19, 0
-; GFX6-NEXT:    s_add_u32 s22, s18, 2
-; GFX6-NEXT:    s_addc_u32 s23, s19, 0
-; GFX6-NEXT:    s_cmp_lg_u32 s16, 0
-; GFX6-NEXT:    s_cselect_b32 s16, s22, s17
-; GFX6-NEXT:    s_cselect_b32 s17, s23, s21
+; GFX6-NEXT:    s_or_b32 s20, s14, s15
+; GFX6-NEXT:    s_cmp_lg_u32 s20, 0
+; GFX6-NEXT:    s_subb_u32 s19, s19, s7
+; GFX6-NEXT:    s_sub_u32 s21, s8, s6
+; GFX6-NEXT:    s_cselect_b64 s[14:15], -1, 0
 ; GFX6-NEXT:    s_or_b32 s14, s14, s15
-; GFX6-NEXT:    s_subb_u32 s9, s9, s20
+; GFX6-NEXT:    s_cmp_lg_u32 s14, 0
+; GFX6-NEXT:    s_subb_u32 s14, s19, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s14, s7
+; GFX6-NEXT:    s_cselect_b32 s15, -1, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s21, s6
+; GFX6-NEXT:    s_cselect_b32 s19, -1, 0
+; GFX6-NEXT:    s_cmp_eq_u32 s14, s7
+; GFX6-NEXT:    s_cselect_b32 s14, s19, s15
+; GFX6-NEXT:    s_add_u32 s15, s17, 1
+; GFX6-NEXT:    s_addc_u32 s19, s16, 0
+; GFX6-NEXT:    s_add_u32 s21, s17, 2
+; GFX6-NEXT:    s_addc_u32 s22, s16, 0
+; GFX6-NEXT:    s_cmp_lg_u32 s14, 0
+; GFX6-NEXT:    s_cselect_b32 s14, s21, s15
+; GFX6-NEXT:    s_cselect_b32 s15, s22, s19
+; GFX6-NEXT:    s_cmp_lg_u32 s20, 0
+; GFX6-NEXT:    s_subb_u32 s9, s9, s18
 ; GFX6-NEXT:    s_cmp_ge_u32 s9, s7
-; GFX6-NEXT:    s_cselect_b32 s14, -1, 0
+; GFX6-NEXT:    s_cselect_b32 s18, -1, 0
 ; GFX6-NEXT:    s_cmp_ge_u32 s8, s6
 ; GFX6-NEXT:    s_cselect_b32 s6, -1, 0
 ; GFX6-NEXT:    s_cmp_eq_u32 s9, s7
-; GFX6-NEXT:    s_cselect_b32 s6, s6, s14
+; GFX6-NEXT:    s_cselect_b32 s6, s6, s18
 ; GFX6-NEXT:    s_cmp_lg_u32 s6, 0
-; GFX6-NEXT:    s_cselect_b32 s7, s17, s19
-; GFX6-NEXT:    s_cselect_b32 s6, s16, s18
+; GFX6-NEXT:    s_cselect_b32 s7, s15, s16
+; GFX6-NEXT:    s_cselect_b32 s6, s14, s17
 ; GFX6-NEXT:    s_xor_b64 s[2:3], s[12:13], s[2:3]
 ; GFX6-NEXT:    s_xor_b64 s[6:7], s[6:7], s[2:3]
-; GFX6-NEXT:    s_sub_u32 s16, s6, s2
-; GFX6-NEXT:    s_subb_u32 s17, s7, s3
+; GFX6-NEXT:    s_sub_u32 s14, s6, s2
+; GFX6-NEXT:    s_subb_u32 s15, s7, s3
 ; GFX6-NEXT:    s_ashr_i32 s6, s1, 31
 ; GFX6-NEXT:    s_add_u32 s0, s0, s6
 ; GFX6-NEXT:    s_mov_b32 s7, s6
@@ -8439,39 +8454,40 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    v_cvt_u32_f32_e32 v0, v0
 ; GFX6-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GFX6-NEXT:    v_mul_hi_u32 v2, s12, v0
-; GFX6-NEXT:    v_readfirstlane_b32 s14, v1
+; GFX6-NEXT:    v_readfirstlane_b32 s16, v1
 ; GFX6-NEXT:    v_readfirstlane_b32 s2, v0
-; GFX6-NEXT:    s_mul_i32 s1, s12, s14
+; GFX6-NEXT:    s_mul_i32 s1, s12, s16
 ; GFX6-NEXT:    v_readfirstlane_b32 s3, v2
 ; GFX6-NEXT:    s_mul_i32 s0, s13, s2
 ; GFX6-NEXT:    s_add_i32 s1, s3, s1
 ; GFX6-NEXT:    s_add_i32 s3, s1, s0
-; GFX6-NEXT:    s_mul_i32 s15, s12, s2
+; GFX6-NEXT:    s_mul_i32 s17, s12, s2
 ; GFX6-NEXT:    v_mul_hi_u32 v2, v0, s3
-; GFX6-NEXT:    v_mul_hi_u32 v0, v0, s15
+; GFX6-NEXT:    v_mul_hi_u32 v0, v0, s17
 ; GFX6-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
 ; GFX6-NEXT:    s_mul_i32 s4, s2, s3
 ; GFX6-NEXT:    v_readfirstlane_b32 s5, v2
 ; GFX6-NEXT:    v_readfirstlane_b32 s18, v0
-; GFX6-NEXT:    v_mul_hi_u32 v0, v1, s15
+; GFX6-NEXT:    v_mul_hi_u32 v0, v1, s17
 ; GFX6-NEXT:    v_mul_hi_u32 v1, v1, s3
 ; GFX6-NEXT:    s_add_u32 s4, s18, s4
 ; GFX6-NEXT:    s_addc_u32 s5, 0, s5
-; GFX6-NEXT:    s_mul_i32 s15, s14, s15
+; GFX6-NEXT:    s_mul_i32 s17, s16, s17
 ; GFX6-NEXT:    v_readfirstlane_b32 s18, v0
-; GFX6-NEXT:    s_add_u32 s4, s4, s15
+; GFX6-NEXT:    s_add_u32 s4, s4, s17
 ; GFX6-NEXT:    s_addc_u32 s4, s5, s18
 ; GFX6-NEXT:    v_readfirstlane_b32 s5, v1
 ; GFX6-NEXT:    s_addc_u32 s5, s5, 0
-; GFX6-NEXT:    s_mul_i32 s3, s14, s3
+; GFX6-NEXT:    s_mul_i32 s3, s16, s3
 ; GFX6-NEXT:    s_add_u32 s3, s4, s3
 ; GFX6-NEXT:    s_addc_u32 s4, 0, s5
 ; GFX6-NEXT:    s_add_u32 s5, s2, s3
 ; GFX6-NEXT:    v_mov_b32_e32 v0, s5
-; GFX6-NEXT:    v_mul_hi_u32 v0, s12, v0
 ; GFX6-NEXT:    s_cselect_b64 s[2:3], -1, 0
+; GFX6-NEXT:    v_mul_hi_u32 v0, s12, v0
 ; GFX6-NEXT:    s_or_b32 s2, s2, s3
-; GFX6-NEXT:    s_addc_u32 s4, s14, s4
+; GFX6-NEXT:    s_cmp_lg_u32 s2, 0
+; GFX6-NEXT:    s_addc_u32 s4, s16, s4
 ; GFX6-NEXT:    s_mul_i32 s2, s12, s4
 ; GFX6-NEXT:    v_readfirstlane_b32 s3, v0
 ; GFX6-NEXT:    s_add_i32 s2, s3, s2
@@ -8485,14 +8501,14 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    v_mul_hi_u32 v1, s4, v0
 ; GFX6-NEXT:    v_mul_hi_u32 v0, s5, v0
 ; GFX6-NEXT:    s_mul_i32 s13, s5, s2
-; GFX6-NEXT:    v_readfirstlane_b32 s15, v2
-; GFX6-NEXT:    s_add_u32 s13, s15, s13
-; GFX6-NEXT:    v_readfirstlane_b32 s14, v0
+; GFX6-NEXT:    v_readfirstlane_b32 s17, v2
+; GFX6-NEXT:    s_add_u32 s13, s17, s13
+; GFX6-NEXT:    v_readfirstlane_b32 s16, v0
 ; GFX6-NEXT:    s_mul_i32 s3, s4, s3
-; GFX6-NEXT:    s_addc_u32 s14, 0, s14
+; GFX6-NEXT:    s_addc_u32 s16, 0, s16
 ; GFX6-NEXT:    v_readfirstlane_b32 s12, v3
 ; GFX6-NEXT:    s_add_u32 s3, s13, s3
-; GFX6-NEXT:    s_addc_u32 s3, s14, s12
+; GFX6-NEXT:    s_addc_u32 s3, s16, s12
 ; GFX6-NEXT:    v_readfirstlane_b32 s12, v1
 ; GFX6-NEXT:    s_addc_u32 s12, s12, 0
 ; GFX6-NEXT:    s_mul_i32 s2, s4, s2
@@ -8501,6 +8517,7 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    s_add_u32 s13, s5, s2
 ; GFX6-NEXT:    s_cselect_b64 s[2:3], -1, 0
 ; GFX6-NEXT:    s_or_b32 s2, s2, s3
+; GFX6-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX6-NEXT:    s_addc_u32 s12, s4, s12
 ; GFX6-NEXT:    s_ashr_i32 s4, s11, 31
 ; GFX6-NEXT:    s_add_u32 s2, s10, s4
@@ -8512,70 +8529,72 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    v_mov_b32_e32 v2, s13
 ; GFX6-NEXT:    v_mul_hi_u32 v3, s10, v2
 ; GFX6-NEXT:    s_mul_i32 s2, s10, s12
-; GFX6-NEXT:    v_readfirstlane_b32 s14, v1
+; GFX6-NEXT:    v_readfirstlane_b32 s16, v1
 ; GFX6-NEXT:    v_mul_hi_u32 v1, s11, v2
-; GFX6-NEXT:    v_readfirstlane_b32 s15, v3
+; GFX6-NEXT:    v_readfirstlane_b32 s17, v3
 ; GFX6-NEXT:    v_mul_hi_u32 v0, s11, v0
-; GFX6-NEXT:    s_add_u32 s2, s15, s2
-; GFX6-NEXT:    s_addc_u32 s14, 0, s14
+; GFX6-NEXT:    s_add_u32 s2, s17, s2
+; GFX6-NEXT:    s_addc_u32 s16, 0, s16
 ; GFX6-NEXT:    s_mul_i32 s13, s11, s13
-; GFX6-NEXT:    v_readfirstlane_b32 s15, v1
+; GFX6-NEXT:    v_readfirstlane_b32 s17, v1
 ; GFX6-NEXT:    s_add_u32 s2, s2, s13
-; GFX6-NEXT:    s_addc_u32 s2, s14, s15
+; GFX6-NEXT:    s_addc_u32 s2, s16, s17
 ; GFX6-NEXT:    v_readfirstlane_b32 s13, v0
 ; GFX6-NEXT:    s_addc_u32 s13, s13, 0
 ; GFX6-NEXT:    s_mul_i32 s12, s11, s12
-; GFX6-NEXT:    s_add_u32 s18, s2, s12
-; GFX6-NEXT:    v_mov_b32_e32 v0, s18
+; GFX6-NEXT:    s_add_u32 s16, s2, s12
+; GFX6-NEXT:    v_mov_b32_e32 v0, s16
 ; GFX6-NEXT:    v_mul_hi_u32 v0, s8, v0
-; GFX6-NEXT:    s_addc_u32 s19, 0, s13
-; GFX6-NEXT:    s_mul_i32 s12, s8, s19
+; GFX6-NEXT:    s_addc_u32 s17, 0, s13
+; GFX6-NEXT:    s_mul_i32 s12, s8, s17
 ; GFX6-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX6-NEXT:    v_readfirstlane_b32 s13, v0
 ; GFX6-NEXT:    s_add_i32 s12, s13, s12
-; GFX6-NEXT:    s_mul_i32 s13, s9, s18
-; GFX6-NEXT:    s_add_i32 s20, s12, s13
-; GFX6-NEXT:    s_sub_i32 s14, s11, s20
-; GFX6-NEXT:    s_mul_i32 s12, s8, s18
+; GFX6-NEXT:    s_mul_i32 s13, s9, s16
+; GFX6-NEXT:    s_add_i32 s18, s12, s13
+; GFX6-NEXT:    s_sub_i32 s19, s11, s18
+; GFX6-NEXT:    s_mul_i32 s12, s8, s16
 ; GFX6-NEXT:    s_sub_u32 s10, s10, s12
 ; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
-; GFX6-NEXT:    s_or_b32 s15, s12, s13
-; GFX6-NEXT:    s_subb_u32 s21, s14, s9
-; GFX6-NEXT:    s_sub_u32 s22, s10, s8
-; GFX6-NEXT:    s_cselect_b64 s[14:15], -1, 0
-; GFX6-NEXT:    s_or_b32 s14, s14, s15
-; GFX6-NEXT:    s_subb_u32 s14, s21, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s14, s9
-; GFX6-NEXT:    s_cselect_b32 s15, -1, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s22, s8
-; GFX6-NEXT:    s_cselect_b32 s21, -1, 0
-; GFX6-NEXT:    s_cmp_eq_u32 s14, s9
-; GFX6-NEXT:    s_cselect_b32 s14, s21, s15
-; GFX6-NEXT:    s_add_u32 s15, s18, 1
-; GFX6-NEXT:    s_addc_u32 s21, s19, 0
-; GFX6-NEXT:    s_add_u32 s22, s18, 2
-; GFX6-NEXT:    s_addc_u32 s23, s19, 0
-; GFX6-NEXT:    s_cmp_lg_u32 s14, 0
-; GFX6-NEXT:    s_cselect_b32 s14, s22, s15
-; GFX6-NEXT:    s_cselect_b32 s15, s23, s21
+; GFX6-NEXT:    s_or_b32 s20, s12, s13
+; GFX6-NEXT:    s_cmp_lg_u32 s20, 0
+; GFX6-NEXT:    s_subb_u32 s19, s19, s9
+; GFX6-NEXT:    s_sub_u32 s21, s10, s8
+; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
 ; GFX6-NEXT:    s_or_b32 s12, s12, s13
-; GFX6-NEXT:    s_subb_u32 s11, s11, s20
+; GFX6-NEXT:    s_cmp_lg_u32 s12, 0
+; GFX6-NEXT:    s_subb_u32 s12, s19, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s12, s9
+; GFX6-NEXT:    s_cselect_b32 s13, -1, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s21, s8
+; GFX6-NEXT:    s_cselect_b32 s19, -1, 0
+; GFX6-NEXT:    s_cmp_eq_u32 s12, s9
+; GFX6-NEXT:    s_cselect_b32 s12, s19, s13
+; GFX6-NEXT:    s_add_u32 s13, s16, 1
+; GFX6-NEXT:    s_addc_u32 s19, s17, 0
+; GFX6-NEXT:    s_add_u32 s21, s16, 2
+; GFX6-NEXT:    s_addc_u32 s22, s17, 0
+; GFX6-NEXT:    s_cmp_lg_u32 s12, 0
+; GFX6-NEXT:    s_cselect_b32 s12, s21, s13
+; GFX6-NEXT:    s_cselect_b32 s13, s22, s19
+; GFX6-NEXT:    s_cmp_lg_u32 s20, 0
+; GFX6-NEXT:    s_subb_u32 s11, s11, s18
 ; GFX6-NEXT:    s_cmp_ge_u32 s11, s9
-; GFX6-NEXT:    s_cselect_b32 s12, -1, 0
+; GFX6-NEXT:    s_cselect_b32 s18, -1, 0
 ; GFX6-NEXT:    s_cmp_ge_u32 s10, s8
 ; GFX6-NEXT:    s_cselect_b32 s8, -1, 0
 ; GFX6-NEXT:    s_cmp_eq_u32 s11, s9
-; GFX6-NEXT:    s_cselect_b32 s8, s8, s12
+; GFX6-NEXT:    s_cselect_b32 s8, s8, s18
 ; GFX6-NEXT:    s_cmp_lg_u32 s8, 0
-; GFX6-NEXT:    s_cselect_b32 s9, s15, s19
-; GFX6-NEXT:    s_cselect_b32 s8, s14, s18
+; GFX6-NEXT:    s_cselect_b32 s9, s13, s17
+; GFX6-NEXT:    s_cselect_b32 s8, s12, s16
 ; GFX6-NEXT:    s_xor_b64 s[4:5], s[4:5], s[6:7]
 ; GFX6-NEXT:    s_xor_b64 s[6:7], s[8:9], s[4:5]
 ; GFX6-NEXT:    s_sub_u32 s4, s6, s4
 ; GFX6-NEXT:    s_subb_u32 s5, s7, s5
 ; GFX6-NEXT:    s_mov_b32 s2, -1
-; GFX6-NEXT:    v_mov_b32_e32 v0, s16
-; GFX6-NEXT:    v_mov_b32_e32 v1, s17
+; GFX6-NEXT:    v_mov_b32_e32 v0, s14
+; GFX6-NEXT:    v_mov_b32_e32 v1, s15
 ; GFX6-NEXT:    v_mov_b32_e32 v2, s4
 ; GFX6-NEXT:    v_mov_b32_e32 v3, s5
 ; GFX6-NEXT:    s_waitcnt lgkmcnt(0)
@@ -8595,8 +8614,8 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    s_xor_b64 s[6:7], s[6:7], s[2:3]
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v0, s6
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v1, s7
-; GFX9-NEXT:    s_sub_u32 s12, 0, s6
-; GFX9-NEXT:    s_subb_u32 s13, 0, s7
+; GFX9-NEXT:    s_sub_u32 s14, 0, s6
+; GFX9-NEXT:    s_subb_u32 s15, 0, s7
 ; GFX9-NEXT:    v_mac_f32_e32 v0, 0x4f800000, v1
 ; GFX9-NEXT:    v_rcp_f32_e32 v0, v0
 ; GFX9-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -8605,52 +8624,56 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    v_mac_f32_e32 v0, 0xcf800000, v1
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GFX9-NEXT:    v_readfirstlane_b32 s14, v1
-; GFX9-NEXT:    v_readfirstlane_b32 s15, v0
-; GFX9-NEXT:    s_mul_i32 s16, s12, s14
-; GFX9-NEXT:    s_mul_hi_u32 s18, s12, s15
-; GFX9-NEXT:    s_mul_i32 s17, s13, s15
-; GFX9-NEXT:    s_add_i32 s16, s18, s16
-; GFX9-NEXT:    s_mul_i32 s19, s12, s15
-; GFX9-NEXT:    s_add_i32 s16, s16, s17
-; GFX9-NEXT:    s_mul_hi_u32 s18, s15, s19
-; GFX9-NEXT:    s_mul_i32 s20, s15, s16
-; GFX9-NEXT:    s_mul_hi_u32 s17, s15, s16
+; GFX9-NEXT:    v_readfirstlane_b32 s16, v1
+; GFX9-NEXT:    v_readfirstlane_b32 s12, v0
+; GFX9-NEXT:    s_mul_i32 s13, s14, s16
+; GFX9-NEXT:    s_mul_hi_u32 s18, s14, s12
+; GFX9-NEXT:    s_mul_i32 s17, s15, s12
+; GFX9-NEXT:    s_add_i32 s13, s18, s13
+; GFX9-NEXT:    s_mul_i32 s19, s14, s12
+; GFX9-NEXT:    s_add_i32 s13, s13, s17
+; GFX9-NEXT:    s_mul_hi_u32 s18, s12, s19
+; GFX9-NEXT:    s_mul_i32 s20, s12, s13
+; GFX9-NEXT:    s_mul_hi_u32 s17, s12, s13
 ; GFX9-NEXT:    s_add_u32 s18, s18, s20
 ; GFX9-NEXT:    s_addc_u32 s17, 0, s17
-; GFX9-NEXT:    s_mul_hi_u32 s20, s14, s19
-; GFX9-NEXT:    s_mul_i32 s19, s14, s19
+; GFX9-NEXT:    s_mul_hi_u32 s20, s16, s19
+; GFX9-NEXT:    s_mul_i32 s19, s16, s19
 ; GFX9-NEXT:    s_add_u32 s18, s18, s19
-; GFX9-NEXT:    s_mul_hi_u32 s21, s14, s16
+; GFX9-NEXT:    s_mul_hi_u32 s21, s16, s13
 ; GFX9-NEXT:    s_addc_u32 s17, s17, s20
 ; GFX9-NEXT:    s_addc_u32 s18, s21, 0
-; GFX9-NEXT:    s_mul_i32 s16, s14, s16
-; GFX9-NEXT:    s_add_u32 s16, s17, s16
+; GFX9-NEXT:    s_mul_i32 s13, s16, s13
+; GFX9-NEXT:    s_add_u32 s13, s17, s13
 ; GFX9-NEXT:    s_addc_u32 s17, 0, s18
-; GFX9-NEXT:    s_add_u32 s15, s15, s16
-; GFX9-NEXT:    s_addc_u32 s14, s14, s17
-; GFX9-NEXT:    s_mul_i32 s16, s12, s14
-; GFX9-NEXT:    s_mul_hi_u32 s17, s12, s15
-; GFX9-NEXT:    s_add_i32 s16, s17, s16
-; GFX9-NEXT:    s_mul_i32 s13, s13, s15
-; GFX9-NEXT:    s_add_i32 s16, s16, s13
-; GFX9-NEXT:    s_mul_i32 s12, s12, s15
-; GFX9-NEXT:    s_mul_hi_u32 s17, s14, s12
-; GFX9-NEXT:    s_mul_i32 s18, s14, s12
-; GFX9-NEXT:    s_mul_i32 s20, s15, s16
-; GFX9-NEXT:    s_mul_hi_u32 s12, s15, s12
-; GFX9-NEXT:    s_mul_hi_u32 s19, s15, s16
-; GFX9-NEXT:    s_add_u32 s12, s12, s20
+; GFX9-NEXT:    s_add_u32 s18, s12, s13
+; GFX9-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[12:13], 0
+; GFX9-NEXT:    s_addc_u32 s16, s16, s17
+; GFX9-NEXT:    s_mul_i32 s12, s14, s16
+; GFX9-NEXT:    s_mul_hi_u32 s13, s14, s18
+; GFX9-NEXT:    s_add_i32 s12, s13, s12
+; GFX9-NEXT:    s_mul_i32 s15, s15, s18
+; GFX9-NEXT:    s_add_i32 s12, s12, s15
+; GFX9-NEXT:    s_mul_i32 s14, s14, s18
+; GFX9-NEXT:    s_mul_hi_u32 s15, s16, s14
+; GFX9-NEXT:    s_mul_i32 s17, s16, s14
+; GFX9-NEXT:    s_mul_i32 s20, s18, s12
+; GFX9-NEXT:    s_mul_hi_u32 s14, s18, s14
+; GFX9-NEXT:    s_mul_hi_u32 s19, s18, s12
+; GFX9-NEXT:    s_add_u32 s14, s14, s20
 ; GFX9-NEXT:    s_addc_u32 s19, 0, s19
-; GFX9-NEXT:    s_add_u32 s12, s12, s18
-; GFX9-NEXT:    s_mul_hi_u32 s13, s14, s16
-; GFX9-NEXT:    s_addc_u32 s12, s19, s17
+; GFX9-NEXT:    s_add_u32 s14, s14, s17
+; GFX9-NEXT:    s_mul_hi_u32 s13, s16, s12
+; GFX9-NEXT:    s_addc_u32 s14, s19, s15
 ; GFX9-NEXT:    s_addc_u32 s13, s13, 0
-; GFX9-NEXT:    s_mul_i32 s16, s14, s16
-; GFX9-NEXT:    s_add_u32 s12, s12, s16
-; GFX9-NEXT:    s_addc_u32 s13, 0, s13
-; GFX9-NEXT:    s_add_u32 s15, s15, s12
-; GFX9-NEXT:    s_addc_u32 s14, s14, s13
+; GFX9-NEXT:    s_mul_i32 s12, s16, s12
+; GFX9-NEXT:    s_add_u32 s12, s14, s12
+; GFX9-NEXT:    s_addc_u32 s14, 0, s13
+; GFX9-NEXT:    s_add_u32 s15, s18, s12
+; GFX9-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[12:13], 0
+; GFX9-NEXT:    s_addc_u32 s14, s16, s14
 ; GFX9-NEXT:    s_ashr_i32 s12, s9, 31
 ; GFX9-NEXT:    s_add_u32 s8, s8, s12
 ; GFX9-NEXT:    s_mov_b32 s13, s12
@@ -8668,35 +8691,38 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    s_addc_u32 s15, s16, s19
 ; GFX9-NEXT:    s_addc_u32 s16, s18, 0
 ; GFX9-NEXT:    s_mul_i32 s14, s9, s14
-; GFX9-NEXT:    s_add_u32 s17, s15, s14
-; GFX9-NEXT:    s_addc_u32 s16, 0, s16
-; GFX9-NEXT:    s_mul_i32 s14, s6, s16
-; GFX9-NEXT:    s_mul_hi_u32 s15, s6, s17
+; GFX9-NEXT:    s_add_u32 s18, s15, s14
+; GFX9-NEXT:    s_addc_u32 s19, 0, s16
+; GFX9-NEXT:    s_mul_i32 s14, s6, s19
+; GFX9-NEXT:    s_mul_hi_u32 s15, s6, s18
 ; GFX9-NEXT:    s_add_i32 s14, s15, s14
-; GFX9-NEXT:    s_mul_i32 s15, s7, s17
-; GFX9-NEXT:    s_add_i32 s18, s14, s15
-; GFX9-NEXT:    s_sub_i32 s19, s9, s18
-; GFX9-NEXT:    s_mul_i32 s14, s6, s17
+; GFX9-NEXT:    s_mul_i32 s15, s7, s18
+; GFX9-NEXT:    s_add_i32 s20, s14, s15
+; GFX9-NEXT:    s_sub_i32 s16, s9, s20
+; GFX9-NEXT:    s_mul_i32 s14, s6, s18
 ; GFX9-NEXT:    s_sub_u32 s8, s8, s14
 ; GFX9-NEXT:    s_cselect_b64 s[14:15], -1, 0
-; GFX9-NEXT:    s_subb_u32 s19, s19, s7
-; GFX9-NEXT:    s_sub_u32 s20, s8, s6
-; GFX9-NEXT:    s_subb_u32 s19, s19, 0
-; GFX9-NEXT:    s_cmp_ge_u32 s19, s7
+; GFX9-NEXT:    s_cmp_lg_u64 s[14:15], 0
+; GFX9-NEXT:    s_subb_u32 s21, s16, s7
+; GFX9-NEXT:    s_sub_u32 s22, s8, s6
+; GFX9-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[16:17], 0
+; GFX9-NEXT:    s_subb_u32 s16, s21, 0
+; GFX9-NEXT:    s_cmp_ge_u32 s16, s7
+; GFX9-NEXT:    s_cselect_b32 s17, -1, 0
+; GFX9-NEXT:    s_cmp_ge_u32 s22, s6
 ; GFX9-NEXT:    s_cselect_b32 s21, -1, 0
-; GFX9-NEXT:    s_cmp_ge_u32 s20, s6
-; GFX9-NEXT:    s_cselect_b32 s20, -1, 0
-; GFX9-NEXT:    s_cmp_eq_u32 s19, s7
-; GFX9-NEXT:    s_cselect_b32 s19, s20, s21
-; GFX9-NEXT:    s_add_u32 s20, s17, 1
-; GFX9-NEXT:    s_addc_u32 s21, s16, 0
-; GFX9-NEXT:    s_add_u32 s22, s17, 2
-; GFX9-NEXT:    s_addc_u32 s23, s16, 0
-; GFX9-NEXT:    s_cmp_lg_u32 s19, 0
-; GFX9-NEXT:    s_cselect_b32 s19, s22, s20
-; GFX9-NEXT:    s_cselect_b32 s20, s23, s21
+; GFX9-NEXT:    s_cmp_eq_u32 s16, s7
+; GFX9-NEXT:    s_cselect_b32 s16, s21, s17
+; GFX9-NEXT:    s_add_u32 s17, s18, 1
+; GFX9-NEXT:    s_addc_u32 s21, s19, 0
+; GFX9-NEXT:    s_add_u32 s22, s18, 2
+; GFX9-NEXT:    s_addc_u32 s23, s19, 0
+; GFX9-NEXT:    s_cmp_lg_u32 s16, 0
+; GFX9-NEXT:    s_cselect_b32 s16, s22, s17
+; GFX9-NEXT:    s_cselect_b32 s17, s23, s21
 ; GFX9-NEXT:    s_cmp_lg_u64 s[14:15], 0
-; GFX9-NEXT:    s_subb_u32 s9, s9, s18
+; GFX9-NEXT:    s_subb_u32 s9, s9, s20
 ; GFX9-NEXT:    s_cmp_ge_u32 s9, s7
 ; GFX9-NEXT:    s_cselect_b32 s14, -1, 0
 ; GFX9-NEXT:    s_cmp_ge_u32 s8, s6
@@ -8704,12 +8730,12 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    s_cmp_eq_u32 s9, s7
 ; GFX9-NEXT:    s_cselect_b32 s6, s6, s14
 ; GFX9-NEXT:    s_cmp_lg_u32 s6, 0
-; GFX9-NEXT:    s_cselect_b32 s7, s20, s16
-; GFX9-NEXT:    s_cselect_b32 s6, s19, s17
+; GFX9-NEXT:    s_cselect_b32 s7, s17, s19
+; GFX9-NEXT:    s_cselect_b32 s6, s16, s18
 ; GFX9-NEXT:    s_xor_b64 s[2:3], s[12:13], s[2:3]
 ; GFX9-NEXT:    s_xor_b64 s[6:7], s[6:7], s[2:3]
-; GFX9-NEXT:    s_sub_u32 s12, s6, s2
-; GFX9-NEXT:    s_subb_u32 s13, s7, s3
+; GFX9-NEXT:    s_sub_u32 s14, s6, s2
+; GFX9-NEXT:    s_subb_u32 s15, s7, s3
 ; GFX9-NEXT:    s_ashr_i32 s2, s1, 31
 ; GFX9-NEXT:    s_add_u32 s0, s0, s2
 ; GFX9-NEXT:    s_mov_b32 s3, s2
@@ -8718,8 +8744,8 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v0, s6
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v1, s7
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
-; GFX9-NEXT:    s_sub_u32 s4, 0, s6
-; GFX9-NEXT:    s_subb_u32 s5, 0, s7
+; GFX9-NEXT:    s_sub_u32 s8, 0, s6
+; GFX9-NEXT:    s_subb_u32 s9, 0, s7
 ; GFX9-NEXT:    v_mac_f32_e32 v0, 0x4f800000, v1
 ; GFX9-NEXT:    v_rcp_f32_e32 v1, v0
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 0
@@ -8729,98 +8755,105 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    v_mac_f32_e32 v1, 0xcf800000, v2
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v2, v2
-; GFX9-NEXT:    v_readfirstlane_b32 s8, v1
-; GFX9-NEXT:    v_readfirstlane_b32 s15, v2
-; GFX9-NEXT:    s_mul_hi_u32 s14, s4, s8
-; GFX9-NEXT:    s_mul_i32 s16, s4, s15
-; GFX9-NEXT:    s_mul_i32 s9, s5, s8
-; GFX9-NEXT:    s_add_i32 s14, s14, s16
-; GFX9-NEXT:    s_add_i32 s14, s14, s9
-; GFX9-NEXT:    s_mul_i32 s17, s4, s8
-; GFX9-NEXT:    s_mul_i32 s16, s8, s14
-; GFX9-NEXT:    s_mul_hi_u32 s18, s8, s17
-; GFX9-NEXT:    s_mul_hi_u32 s9, s8, s14
+; GFX9-NEXT:    v_readfirstlane_b32 s4, v1
+; GFX9-NEXT:    v_readfirstlane_b32 s13, v2
+; GFX9-NEXT:    s_mul_hi_u32 s12, s8, s4
+; GFX9-NEXT:    s_mul_i32 s16, s8, s13
+; GFX9-NEXT:    s_mul_i32 s5, s9, s4
+; GFX9-NEXT:    s_add_i32 s12, s12, s16
+; GFX9-NEXT:    s_add_i32 s12, s12, s5
+; GFX9-NEXT:    s_mul_i32 s17, s8, s4
+; GFX9-NEXT:    s_mul_i32 s16, s4, s12
+; GFX9-NEXT:    s_mul_hi_u32 s18, s4, s17
+; GFX9-NEXT:    s_mul_hi_u32 s5, s4, s12
 ; GFX9-NEXT:    s_add_u32 s16, s18, s16
-; GFX9-NEXT:    s_addc_u32 s9, 0, s9
-; GFX9-NEXT:    s_mul_hi_u32 s19, s15, s17
-; GFX9-NEXT:    s_mul_i32 s17, s15, s17
+; GFX9-NEXT:    s_addc_u32 s5, 0, s5
+; GFX9-NEXT:    s_mul_hi_u32 s19, s13, s17
+; GFX9-NEXT:    s_mul_i32 s17, s13, s17
 ; GFX9-NEXT:    s_add_u32 s16, s16, s17
-; GFX9-NEXT:    s_mul_hi_u32 s18, s15, s14
-; GFX9-NEXT:    s_addc_u32 s9, s9, s19
+; GFX9-NEXT:    s_mul_hi_u32 s18, s13, s12
+; GFX9-NEXT:    s_addc_u32 s5, s5, s19
 ; GFX9-NEXT:    s_addc_u32 s16, s18, 0
-; GFX9-NEXT:    s_mul_i32 s14, s15, s14
-; GFX9-NEXT:    s_add_u32 s9, s9, s14
-; GFX9-NEXT:    s_addc_u32 s14, 0, s16
-; GFX9-NEXT:    s_add_u32 s8, s8, s9
-; GFX9-NEXT:    s_addc_u32 s9, s15, s14
-; GFX9-NEXT:    s_mul_i32 s14, s4, s9
-; GFX9-NEXT:    s_mul_hi_u32 s15, s4, s8
-; GFX9-NEXT:    s_add_i32 s14, s15, s14
-; GFX9-NEXT:    s_mul_i32 s5, s5, s8
-; GFX9-NEXT:    s_add_i32 s14, s14, s5
-; GFX9-NEXT:    s_mul_i32 s4, s4, s8
-; GFX9-NEXT:    s_mul_hi_u32 s15, s9, s4
-; GFX9-NEXT:    s_mul_i32 s16, s9, s4
-; GFX9-NEXT:    s_mul_i32 s18, s8, s14
-; GFX9-NEXT:    s_mul_hi_u32 s4, s8, s4
-; GFX9-NEXT:    s_mul_hi_u32 s17, s8, s14
-; GFX9-NEXT:    s_add_u32 s4, s4, s18
+; GFX9-NEXT:    s_mul_i32 s12, s13, s12
+; GFX9-NEXT:    s_add_u32 s5, s5, s12
+; GFX9-NEXT:    s_addc_u32 s12, 0, s16
+; GFX9-NEXT:    s_add_u32 s16, s4, s5
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX9-NEXT:    s_addc_u32 s12, s13, s12
+; GFX9-NEXT:    s_mul_i32 s4, s8, s12
+; GFX9-NEXT:    s_mul_hi_u32 s5, s8, s16
+; GFX9-NEXT:    s_add_i32 s4, s5, s4
+; GFX9-NEXT:    s_mul_i32 s9, s9, s16
+; GFX9-NEXT:    s_add_i32 s4, s4, s9
+; GFX9-NEXT:    s_mul_i32 s8, s8, s16
+; GFX9-NEXT:    s_mul_hi_u32 s9, s12, s8
+; GFX9-NEXT:    s_mul_i32 s13, s12, s8
+; GFX9-NEXT:    s_mul_i32 s18, s16, s4
+; GFX9-NEXT:    s_mul_hi_u32 s8, s16, s8
+; GFX9-NEXT:    s_mul_hi_u32 s17, s16, s4
+; GFX9-NEXT:    s_add_u32 s8, s8, s18
 ; GFX9-NEXT:    s_addc_u32 s17, 0, s17
-; GFX9-NEXT:    s_add_u32 s4, s4, s16
-; GFX9-NEXT:    s_mul_hi_u32 s5, s9, s14
-; GFX9-NEXT:    s_addc_u32 s4, s17, s15
+; GFX9-NEXT:    s_add_u32 s8, s8, s13
+; GFX9-NEXT:    s_mul_hi_u32 s5, s12, s4
+; GFX9-NEXT:    s_addc_u32 s8, s17, s9
 ; GFX9-NEXT:    s_addc_u32 s5, s5, 0
-; GFX9-NEXT:    s_mul_i32 s14, s9, s14
-; GFX9-NEXT:    s_add_u32 s4, s4, s14
-; GFX9-NEXT:    s_addc_u32 s5, 0, s5
-; GFX9-NEXT:    s_add_u32 s14, s8, s4
-; GFX9-NEXT:    s_addc_u32 s15, s9, s5
+; GFX9-NEXT:    s_mul_i32 s4, s12, s4
+; GFX9-NEXT:    s_add_u32 s4, s8, s4
+; GFX9-NEXT:    s_addc_u32 s8, 0, s5
+; GFX9-NEXT:    s_add_u32 s13, s16, s4
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX9-NEXT:    s_addc_u32 s12, s12, s8
 ; GFX9-NEXT:    s_ashr_i32 s4, s11, 31
 ; GFX9-NEXT:    s_add_u32 s8, s10, s4
 ; GFX9-NEXT:    s_mov_b32 s5, s4
 ; GFX9-NEXT:    s_addc_u32 s9, s11, s4
 ; GFX9-NEXT:    s_xor_b64 s[8:9], s[8:9], s[4:5]
-; GFX9-NEXT:    s_mul_i32 s11, s8, s15
-; GFX9-NEXT:    s_mul_hi_u32 s16, s8, s14
-; GFX9-NEXT:    s_mul_hi_u32 s10, s8, s15
+; GFX9-NEXT:    s_mul_i32 s11, s8, s12
+; GFX9-NEXT:    s_mul_hi_u32 s16, s8, s13
+; GFX9-NEXT:    s_mul_hi_u32 s10, s8, s12
 ; GFX9-NEXT:    s_add_u32 s11, s16, s11
 ; GFX9-NEXT:    s_addc_u32 s10, 0, s10
-; GFX9-NEXT:    s_mul_hi_u32 s17, s9, s14
-; GFX9-NEXT:    s_mul_i32 s14, s9, s14
-; GFX9-NEXT:    s_add_u32 s11, s11, s14
-; GFX9-NEXT:    s_mul_hi_u32 s16, s9, s15
+; GFX9-NEXT:    s_mul_hi_u32 s17, s9, s13
+; GFX9-NEXT:    s_mul_i32 s13, s9, s13
+; GFX9-NEXT:    s_add_u32 s11, s11, s13
+; GFX9-NEXT:    s_mul_hi_u32 s16, s9, s12
 ; GFX9-NEXT:    s_addc_u32 s10, s10, s17
 ; GFX9-NEXT:    s_addc_u32 s11, s16, 0
-; GFX9-NEXT:    s_mul_i32 s14, s9, s15
-; GFX9-NEXT:    s_add_u32 s14, s10, s14
-; GFX9-NEXT:    s_addc_u32 s15, 0, s11
-; GFX9-NEXT:    s_mul_i32 s10, s6, s15
-; GFX9-NEXT:    s_mul_hi_u32 s11, s6, s14
+; GFX9-NEXT:    s_mul_i32 s12, s9, s12
+; GFX9-NEXT:    s_add_u32 s16, s10, s12
+; GFX9-NEXT:    s_addc_u32 s17, 0, s11
+; GFX9-NEXT:    s_mul_i32 s10, s6, s17
+; GFX9-NEXT:    s_mul_hi_u32 s11, s6, s16
 ; GFX9-NEXT:    s_add_i32 s10, s11, s10
-; GFX9-NEXT:    s_mul_i32 s11, s7, s14
-; GFX9-NEXT:    s_add_i32 s16, s10, s11
-; GFX9-NEXT:    s_sub_i32 s17, s9, s16
-; GFX9-NEXT:    s_mul_i32 s10, s6, s14
+; GFX9-NEXT:    s_mul_i32 s11, s7, s16
+; GFX9-NEXT:    s_add_i32 s18, s10, s11
+; GFX9-NEXT:    s_sub_i32 s12, s9, s18
+; GFX9-NEXT:    s_mul_i32 s10, s6, s16
 ; GFX9-NEXT:    s_sub_u32 s8, s8, s10
 ; GFX9-NEXT:    s_cselect_b64 s[10:11], -1, 0
-; GFX9-NEXT:    s_subb_u32 s17, s17, s7
-; GFX9-NEXT:    s_sub_u32 s18, s8, s6
-; GFX9-NEXT:    s_subb_u32 s17, s17, 0
-; GFX9-NEXT:    s_cmp_ge_u32 s17, s7
+; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
+; GFX9-NEXT:    s_subb_u32 s19, s12, s7
+; GFX9-NEXT:    s_sub_u32 s20, s8, s6
+; GFX9-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[12:13], 0
+; GFX9-NEXT:    s_subb_u32 s12, s19, 0
+; GFX9-NEXT:    s_cmp_ge_u32 s12, s7
+; GFX9-NEXT:    s_cselect_b32 s13, -1, 0
+; GFX9-NEXT:    s_cmp_ge_u32 s20, s6
 ; GFX9-NEXT:    s_cselect_b32 s19, -1, 0
-; GFX9-NEXT:    s_cmp_ge_u32 s18, s6
-; GFX9-NEXT:    s_cselect_b32 s18, -1, 0
-; GFX9-NEXT:    s_cmp_eq_u32 s17, s7
-; GFX9-NEXT:    s_cselect_b32 s17, s18, s19
-; GFX9-NEXT:    s_add_u32 s18, s14, 1
-; GFX9-NEXT:    s_addc_u32 s19, s15, 0
-; GFX9-NEXT:    s_add_u32 s20, s14, 2
-; GFX9-NEXT:    s_addc_u32 s21, s15, 0
-; GFX9-NEXT:    s_cmp_lg_u32 s17, 0
-; GFX9-NEXT:    s_cselect_b32 s17, s20, s18
-; GFX9-NEXT:    s_cselect_b32 s18, s21, s19
+; GFX9-NEXT:    s_cmp_eq_u32 s12, s7
+; GFX9-NEXT:    s_cselect_b32 s12, s19, s13
+; GFX9-NEXT:    s_add_u32 s13, s16, 1
+; GFX9-NEXT:    s_addc_u32 s19, s17, 0
+; GFX9-NEXT:    s_add_u32 s20, s16, 2
+; GFX9-NEXT:    s_addc_u32 s21, s17, 0
+; GFX9-NEXT:    s_cmp_lg_u32 s12, 0
+; GFX9-NEXT:    s_cselect_b32 s12, s20, s13
+; GFX9-NEXT:    s_cselect_b32 s13, s21, s19
 ; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT:    s_subb_u32 s9, s9, s16
+; GFX9-NEXT:    s_subb_u32 s9, s9, s18
 ; GFX9-NEXT:    s_cmp_ge_u32 s9, s7
 ; GFX9-NEXT:    s_cselect_b32 s10, -1, 0
 ; GFX9-NEXT:    s_cmp_ge_u32 s8, s6
@@ -8828,14 +8861,14 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    s_cmp_eq_u32 s9, s7
 ; GFX9-NEXT:    s_cselect_b32 s6, s6, s10
 ; GFX9-NEXT:    s_cmp_lg_u32 s6, 0
-; GFX9-NEXT:    s_cselect_b32 s7, s18, s15
-; GFX9-NEXT:    s_cselect_b32 s6, s17, s14
+; GFX9-NEXT:    s_cselect_b32 s7, s13, s17
+; GFX9-NEXT:    s_cselect_b32 s6, s12, s16
 ; GFX9-NEXT:    s_xor_b64 s[2:3], s[4:5], s[2:3]
 ; GFX9-NEXT:    s_xor_b64 s[4:5], s[6:7], s[2:3]
 ; GFX9-NEXT:    s_sub_u32 s2, s4, s2
 ; GFX9-NEXT:    s_subb_u32 s3, s5, s3
-; GFX9-NEXT:    v_mov_b32_e32 v1, s12
-; GFX9-NEXT:    v_mov_b32_e32 v2, s13
+; GFX9-NEXT:    v_mov_b32_e32 v1, s14
+; GFX9-NEXT:    v_mov_b32_e32 v2, s15
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-NEXT:    v_mov_b32_e32 v4, s3
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
@@ -9056,9 +9089,10 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX6-NEXT:    s_addc_u32 s13, 0, s14
 ; GFX6-NEXT:    s_add_u32 s14, s0, s1
 ; GFX6-NEXT:    v_mov_b32_e32 v0, s14
-; GFX6-NEXT:    v_mul_hi_u32 v0, s10, v0
 ; GFX6-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GFX6-NEXT:    v_mul_hi_u32 v0, s10, v0
 ; GFX6-NEXT:    s_or_b32 s0, s0, s1
+; GFX6-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX6-NEXT:    s_addc_u32 s12, s12, s13
 ; GFX6-NEXT:    s_mul_i32 s0, s10, s12
 ; GFX6-NEXT:    v_readfirstlane_b32 s1, v0
@@ -9089,6 +9123,7 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX6-NEXT:    s_add_u32 s13, s14, s0
 ; GFX6-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GFX6-NEXT:    s_or_b32 s0, s0, s1
+; GFX6-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX6-NEXT:    s_addc_u32 s12, s12, s10
 ; GFX6-NEXT:    s_ashr_i32 s10, s7, 31
 ; GFX6-NEXT:    s_add_u32 s0, s6, s10
@@ -9123,43 +9158,46 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX6-NEXT:    v_readfirstlane_b32 s5, v0
 ; GFX6-NEXT:    s_add_i32 s4, s5, s4
 ; GFX6-NEXT:    s_mul_i32 s5, s9, s12
-; GFX6-NEXT:    s_add_i32 s14, s4, s5
-; GFX6-NEXT:    s_sub_i32 s13, s7, s14
+; GFX6-NEXT:    s_add_i32 s13, s4, s5
+; GFX6-NEXT:    s_sub_i32 s14, s7, s13
 ; GFX6-NEXT:    s_mul_i32 s4, s8, s12
 ; GFX6-NEXT:    s_sub_u32 s6, s6, s4
 ; GFX6-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GFX6-NEXT:    s_or_b32 s12, s4, s5
-; GFX6-NEXT:    s_subb_u32 s15, s13, s9
-; GFX6-NEXT:    s_sub_u32 s16, s6, s8
-; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
-; GFX6-NEXT:    s_or_b32 s17, s12, s13
-; GFX6-NEXT:    s_subb_u32 s17, s15, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s17, s9
-; GFX6-NEXT:    s_cselect_b32 s18, -1, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s16, s8
-; GFX6-NEXT:    s_cselect_b32 s19, -1, 0
-; GFX6-NEXT:    s_cmp_eq_u32 s17, s9
-; GFX6-NEXT:    s_cselect_b32 s18, s19, s18
-; GFX6-NEXT:    s_or_b32 s12, s12, s13
-; GFX6-NEXT:    s_subb_u32 s15, s15, s9
-; GFX6-NEXT:    s_sub_u32 s19, s16, s8
-; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
-; GFX6-NEXT:    s_or_b32 s12, s12, s13
-; GFX6-NEXT:    s_subb_u32 s12, s15, 0
-; GFX6-NEXT:    s_cmp_lg_u32 s18, 0
-; GFX6-NEXT:    s_cselect_b32 s13, s19, s16
-; GFX6-NEXT:    s_cselect_b32 s12, s12, s17
+; GFX6-NEXT:    s_cmp_lg_u32 s12, 0
+; GFX6-NEXT:    s_subb_u32 s14, s14, s9
+; GFX6-NEXT:    s_sub_u32 s15, s6, s8
+; GFX6-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GFX6-NEXT:    s_or_b32 s4, s4, s5
-; GFX6-NEXT:    s_subb_u32 s4, s7, s14
-; GFX6-NEXT:    s_cmp_ge_u32 s4, s9
+; GFX6-NEXT:    s_cmp_lg_u32 s4, 0
+; GFX6-NEXT:    s_subb_u32 s16, s14, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s16, s9
 ; GFX6-NEXT:    s_cselect_b32 s5, -1, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s6, s8
+; GFX6-NEXT:    s_cmp_ge_u32 s15, s8
+; GFX6-NEXT:    s_cselect_b32 s17, -1, 0
+; GFX6-NEXT:    s_cmp_eq_u32 s16, s9
+; GFX6-NEXT:    s_cselect_b32 s17, s17, s5
+; GFX6-NEXT:    s_cmp_lg_u32 s4, 0
+; GFX6-NEXT:    s_subb_u32 s14, s14, s9
+; GFX6-NEXT:    s_sub_u32 s18, s15, s8
+; GFX6-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX6-NEXT:    s_or_b32 s4, s4, s5
+; GFX6-NEXT:    s_cmp_lg_u32 s4, 0
+; GFX6-NEXT:    s_subb_u32 s4, s14, 0
+; GFX6-NEXT:    s_cmp_lg_u32 s17, 0
+; GFX6-NEXT:    s_cselect_b32 s14, s18, s15
+; GFX6-NEXT:    s_cselect_b32 s4, s4, s16
+; GFX6-NEXT:    s_cmp_lg_u32 s12, 0
+; GFX6-NEXT:    s_subb_u32 s5, s7, s13
+; GFX6-NEXT:    s_cmp_ge_u32 s5, s9
 ; GFX6-NEXT:    s_cselect_b32 s7, -1, 0
-; GFX6-NEXT:    s_cmp_eq_u32 s4, s9
-; GFX6-NEXT:    s_cselect_b32 s5, s7, s5
-; GFX6-NEXT:    s_cmp_lg_u32 s5, 0
-; GFX6-NEXT:    s_cselect_b32 s5, s12, s4
-; GFX6-NEXT:    s_cselect_b32 s4, s13, s6
+; GFX6-NEXT:    s_cmp_ge_u32 s6, s8
+; GFX6-NEXT:    s_cselect_b32 s8, -1, 0
+; GFX6-NEXT:    s_cmp_eq_u32 s5, s9
+; GFX6-NEXT:    s_cselect_b32 s7, s8, s7
+; GFX6-NEXT:    s_cmp_lg_u32 s7, 0
+; GFX6-NEXT:    s_cselect_b32 s5, s4, s5
+; GFX6-NEXT:    s_cselect_b32 s4, s14, s6
 ; GFX6-NEXT:    s_xor_b64 s[4:5], s[4:5], s[10:11]
 ; GFX6-NEXT:    s_sub_u32 s4, s4, s10
 ; GFX6-NEXT:    s_subb_u32 s5, s5, s10
@@ -9181,8 +9219,8 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v0, s6
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v1, s7
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
-; GFX9-NEXT:    s_sub_u32 s4, 0, s6
-; GFX9-NEXT:    s_subb_u32 s5, 0, s7
+; GFX9-NEXT:    s_sub_u32 s8, 0, s6
+; GFX9-NEXT:    s_subb_u32 s9, 0, s7
 ; GFX9-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GFX9-NEXT:    v_rcp_f32_e32 v1, v0
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 0
@@ -9192,52 +9230,56 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX9-NEXT:    v_madmk_f32 v1, v2, 0xcf800000, v1
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v2, v2
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v1, v1
-; GFX9-NEXT:    v_readfirstlane_b32 s8, v2
-; GFX9-NEXT:    v_readfirstlane_b32 s9, v1
-; GFX9-NEXT:    s_mul_i32 s10, s4, s8
-; GFX9-NEXT:    s_mul_hi_u32 s12, s4, s9
-; GFX9-NEXT:    s_mul_i32 s11, s5, s9
-; GFX9-NEXT:    s_add_i32 s10, s12, s10
-; GFX9-NEXT:    s_mul_i32 s13, s4, s9
-; GFX9-NEXT:    s_add_i32 s10, s10, s11
-; GFX9-NEXT:    s_mul_hi_u32 s12, s9, s13
-; GFX9-NEXT:    s_mul_i32 s14, s9, s10
-; GFX9-NEXT:    s_mul_hi_u32 s11, s9, s10
+; GFX9-NEXT:    v_readfirstlane_b32 s10, v2
+; GFX9-NEXT:    v_readfirstlane_b32 s4, v1
+; GFX9-NEXT:    s_mul_i32 s5, s8, s10
+; GFX9-NEXT:    s_mul_hi_u32 s12, s8, s4
+; GFX9-NEXT:    s_mul_i32 s11, s9, s4
+; GFX9-NEXT:    s_add_i32 s5, s12, s5
+; GFX9-NEXT:    s_mul_i32 s13, s8, s4
+; GFX9-NEXT:    s_add_i32 s5, s5, s11
+; GFX9-NEXT:    s_mul_hi_u32 s12, s4, s13
+; GFX9-NEXT:    s_mul_i32 s14, s4, s5
+; GFX9-NEXT:    s_mul_hi_u32 s11, s4, s5
 ; GFX9-NEXT:    s_add_u32 s12, s12, s14
 ; GFX9-NEXT:    s_addc_u32 s11, 0, s11
-; GFX9-NEXT:    s_mul_hi_u32 s15, s8, s13
-; GFX9-NEXT:    s_mul_i32 s13, s8, s13
+; GFX9-NEXT:    s_mul_hi_u32 s15, s10, s13
+; GFX9-NEXT:    s_mul_i32 s13, s10, s13
 ; GFX9-NEXT:    s_add_u32 s12, s12, s13
-; GFX9-NEXT:    s_mul_hi_u32 s14, s8, s10
+; GFX9-NEXT:    s_mul_hi_u32 s14, s10, s5
 ; GFX9-NEXT:    s_addc_u32 s11, s11, s15
 ; GFX9-NEXT:    s_addc_u32 s12, s14, 0
-; GFX9-NEXT:    s_mul_i32 s10, s8, s10
-; GFX9-NEXT:    s_add_u32 s10, s11, s10
+; GFX9-NEXT:    s_mul_i32 s5, s10, s5
+; GFX9-NEXT:    s_add_u32 s5, s11, s5
 ; GFX9-NEXT:    s_addc_u32 s11, 0, s12
-; GFX9-NEXT:    s_add_u32 s9, s9, s10
-; GFX9-NEXT:    s_addc_u32 s8, s8, s11
-; GFX9-NEXT:    s_mul_i32 s10, s4, s8
-; GFX9-NEXT:    s_mul_hi_u32 s11, s4, s9
-; GFX9-NEXT:    s_add_i32 s10, s11, s10
-; GFX9-NEXT:    s_mul_i32 s5, s5, s9
-; GFX9-NEXT:    s_add_i32 s10, s10, s5
-; GFX9-NEXT:    s_mul_i32 s4, s4, s9
-; GFX9-NEXT:    s_mul_hi_u32 s11, s8, s4
-; GFX9-NEXT:    s_mul_i32 s12, s8, s4
-; GFX9-NEXT:    s_mul_i32 s14, s9, s10
-; GFX9-NEXT:    s_mul_hi_u32 s4, s9, s4
-; GFX9-NEXT:    s_mul_hi_u32 s13, s9, s10
-; GFX9-NEXT:    s_add_u32 s4, s4, s14
+; GFX9-NEXT:    s_add_u32 s12, s4, s5
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX9-NEXT:    s_addc_u32 s10, s10, s11
+; GFX9-NEXT:    s_mul_i32 s4, s8, s10
+; GFX9-NEXT:    s_mul_hi_u32 s5, s8, s12
+; GFX9-NEXT:    s_add_i32 s4, s5, s4
+; GFX9-NEXT:    s_mul_i32 s9, s9, s12
+; GFX9-NEXT:    s_add_i32 s4, s4, s9
+; GFX9-NEXT:    s_mul_i32 s8, s8, s12
+; GFX9-NEXT:    s_mul_hi_u32 s9, s10, s8
+; GFX9-NEXT:    s_mul_i32 s11, s10, s8
+; GFX9-NEXT:    s_mul_i32 s14, s12, s4
+; GFX9-NEXT:    s_mul_hi_u32 s8, s12, s8
+; GFX9-NEXT:    s_mul_hi_u32 s13, s12, s4
+; GFX9-NEXT:    s_add_u32 s8, s8, s14
 ; GFX9-NEXT:    s_addc_u32 s13, 0, s13
-; GFX9-NEXT:    s_add_u32 s4, s4, s12
-; GFX9-NEXT:    s_mul_hi_u32 s5, s8, s10
-; GFX9-NEXT:    s_addc_u32 s4, s13, s11
+; GFX9-NEXT:    s_add_u32 s8, s8, s11
+; GFX9-NEXT:    s_mul_hi_u32 s5, s10, s4
+; GFX9-NEXT:    s_addc_u32 s8, s13, s9
 ; GFX9-NEXT:    s_addc_u32 s5, s5, 0
-; GFX9-NEXT:    s_mul_i32 s10, s8, s10
-; GFX9-NEXT:    s_add_u32 s4, s4, s10
-; GFX9-NEXT:    s_addc_u32 s5, 0, s5
-; GFX9-NEXT:    s_add_u32 s9, s9, s4
-; GFX9-NEXT:    s_addc_u32 s8, s8, s5
+; GFX9-NEXT:    s_mul_i32 s4, s10, s4
+; GFX9-NEXT:    s_add_u32 s4, s8, s4
+; GFX9-NEXT:    s_addc_u32 s8, 0, s5
+; GFX9-NEXT:    s_add_u32 s9, s12, s4
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX9-NEXT:    s_addc_u32 s8, s10, s8
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_ashr_i32 s4, s3, 31
 ; GFX9-NEXT:    s_add_u32 s2, s2, s4
@@ -9267,9 +9309,11 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX9-NEXT:    s_mul_i32 s8, s6, s8
 ; GFX9-NEXT:    s_sub_u32 s2, s2, s8
 ; GFX9-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[8:9], 0
 ; GFX9-NEXT:    s_subb_u32 s13, s10, s7
 ; GFX9-NEXT:    s_sub_u32 s14, s2, s6
 ; GFX9-NEXT:    s_cselect_b64 s[10:11], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
 ; GFX9-NEXT:    s_subb_u32 s15, s13, 0
 ; GFX9-NEXT:    s_cmp_ge_u32 s15, s7
 ; GFX9-NEXT:    s_cselect_b32 s16, -1, 0
@@ -9278,11 +9322,13 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
 ; GFX9-NEXT:    s_cmp_eq_u32 s15, s7
 ; GFX9-NEXT:    s_cselect_b32 s16, s17, s16
 ; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT:    s_subb_u32 s10, s13, s7
-; GFX9-NEXT:    s_sub_u32 s11, s14, s6
-; GFX9-NEXT:    s_subb_u32 s10, s10, 0
+; GFX9-NEXT:    s_subb_u32 s13, s13, s7
+; GFX9-NEXT:    s_sub_u32 s17, s14, s6
+; GFX9-NEXT:    s_cselect_b64 s[10:11], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
+; GFX9-NEXT:    s_subb_u32 s10, s13, 0
 ; GFX9-NEXT:    s_cmp_lg_u32 s16, 0
-; GFX9-NEXT:    s_cselect_b32 s11, s11, s14
+; GFX9-NEXT:    s_cselect_b32 s11, s17, s14
 ; GFX9-NEXT:    s_cselect_b32 s10, s10, s15
 ; GFX9-NEXT:    s_cmp_lg_u64 s[8:9], 0
 ; GFX9-NEXT:    s_subb_u32 s3, s3, s12
@@ -9444,9 +9490,10 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    s_addc_u32 s15, 0, s16
 ; GFX6-NEXT:    s_add_u32 s16, s6, s7
 ; GFX6-NEXT:    v_mov_b32_e32 v0, s16
-; GFX6-NEXT:    v_mul_hi_u32 v0, s12, v0
 ; GFX6-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; GFX6-NEXT:    v_mul_hi_u32 v0, s12, v0
 ; GFX6-NEXT:    s_or_b32 s6, s6, s7
+; GFX6-NEXT:    s_cmp_lg_u32 s6, 0
 ; GFX6-NEXT:    s_addc_u32 s14, s14, s15
 ; GFX6-NEXT:    s_mul_i32 s6, s12, s14
 ; GFX6-NEXT:    v_readfirstlane_b32 s7, v0
@@ -9477,6 +9524,7 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    s_add_u32 s13, s16, s6
 ; GFX6-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GFX6-NEXT:    s_or_b32 s6, s6, s7
+; GFX6-NEXT:    s_cmp_lg_u32 s6, 0
 ; GFX6-NEXT:    s_addc_u32 s12, s14, s12
 ; GFX6-NEXT:    s_ashr_i32 s6, s9, 31
 ; GFX6-NEXT:    s_add_u32 s8, s8, s6
@@ -9509,46 +9557,49 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    v_readfirstlane_b32 s14, v0
 ; GFX6-NEXT:    s_add_i32 s13, s14, s13
 ; GFX6-NEXT:    s_mul_i32 s14, s3, s12
-; GFX6-NEXT:    s_add_i32 s16, s13, s14
-; GFX6-NEXT:    s_sub_i32 s14, s9, s16
+; GFX6-NEXT:    s_add_i32 s14, s13, s14
+; GFX6-NEXT:    s_sub_i32 s15, s9, s14
 ; GFX6-NEXT:    s_mul_i32 s12, s2, s12
 ; GFX6-NEXT:    s_sub_u32 s8, s8, s12
 ; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
-; GFX6-NEXT:    s_or_b32 s15, s12, s13
-; GFX6-NEXT:    s_subb_u32 s17, s14, s3
-; GFX6-NEXT:    s_sub_u32 s18, s8, s2
-; GFX6-NEXT:    s_cselect_b64 s[14:15], -1, 0
-; GFX6-NEXT:    s_or_b32 s19, s14, s15
-; GFX6-NEXT:    s_subb_u32 s19, s17, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s19, s3
-; GFX6-NEXT:    s_cselect_b32 s20, -1, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s18, s2
-; GFX6-NEXT:    s_cselect_b32 s21, -1, 0
-; GFX6-NEXT:    s_cmp_eq_u32 s19, s3
-; GFX6-NEXT:    s_cselect_b32 s20, s21, s20
-; GFX6-NEXT:    s_or_b32 s14, s14, s15
-; GFX6-NEXT:    s_subb_u32 s17, s17, s3
-; GFX6-NEXT:    s_sub_u32 s21, s18, s2
-; GFX6-NEXT:    s_cselect_b64 s[14:15], -1, 0
-; GFX6-NEXT:    s_or_b32 s14, s14, s15
-; GFX6-NEXT:    s_subb_u32 s14, s17, 0
-; GFX6-NEXT:    s_cmp_lg_u32 s20, 0
-; GFX6-NEXT:    s_cselect_b32 s15, s21, s18
-; GFX6-NEXT:    s_cselect_b32 s14, s14, s19
+; GFX6-NEXT:    s_or_b32 s16, s12, s13
+; GFX6-NEXT:    s_cmp_lg_u32 s16, 0
+; GFX6-NEXT:    s_subb_u32 s15, s15, s3
+; GFX6-NEXT:    s_sub_u32 s17, s8, s2
+; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GFX6-NEXT:    s_or_b32 s12, s12, s13
+; GFX6-NEXT:    s_cmp_lg_u32 s12, 0
+; GFX6-NEXT:    s_subb_u32 s18, s15, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s18, s3
+; GFX6-NEXT:    s_cselect_b32 s13, -1, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s17, s2
+; GFX6-NEXT:    s_cselect_b32 s19, -1, 0
+; GFX6-NEXT:    s_cmp_eq_u32 s18, s3
+; GFX6-NEXT:    s_cselect_b32 s19, s19, s13
+; GFX6-NEXT:    s_cmp_lg_u32 s12, 0
+; GFX6-NEXT:    s_subb_u32 s15, s15, s3
+; GFX6-NEXT:    s_sub_u32 s20, s17, s2
+; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
 ; GFX6-NEXT:    s_or_b32 s12, s12, s13
-; GFX6-NEXT:    s_subb_u32 s9, s9, s16
+; GFX6-NEXT:    s_cmp_lg_u32 s12, 0
+; GFX6-NEXT:    s_subb_u32 s12, s15, 0
+; GFX6-NEXT:    s_cmp_lg_u32 s19, 0
+; GFX6-NEXT:    s_cselect_b32 s13, s20, s17
+; GFX6-NEXT:    s_cselect_b32 s12, s12, s18
+; GFX6-NEXT:    s_cmp_lg_u32 s16, 0
+; GFX6-NEXT:    s_subb_u32 s9, s9, s14
 ; GFX6-NEXT:    s_cmp_ge_u32 s9, s3
-; GFX6-NEXT:    s_cselect_b32 s12, -1, 0
+; GFX6-NEXT:    s_cselect_b32 s14, -1, 0
 ; GFX6-NEXT:    s_cmp_ge_u32 s8, s2
 ; GFX6-NEXT:    s_cselect_b32 s2, -1, 0
 ; GFX6-NEXT:    s_cmp_eq_u32 s9, s3
-; GFX6-NEXT:    s_cselect_b32 s2, s2, s12
+; GFX6-NEXT:    s_cselect_b32 s2, s2, s14
 ; GFX6-NEXT:    s_cmp_lg_u32 s2, 0
-; GFX6-NEXT:    s_cselect_b32 s3, s14, s9
-; GFX6-NEXT:    s_cselect_b32 s2, s15, s8
+; GFX6-NEXT:    s_cselect_b32 s3, s12, s9
+; GFX6-NEXT:    s_cselect_b32 s2, s13, s8
 ; GFX6-NEXT:    s_xor_b64 s[2:3], s[2:3], s[6:7]
-; GFX6-NEXT:    s_sub_u32 s14, s2, s6
-; GFX6-NEXT:    s_subb_u32 s15, s3, s6
+; GFX6-NEXT:    s_sub_u32 s12, s2, s6
+; GFX6-NEXT:    s_subb_u32 s13, s3, s6
 ; GFX6-NEXT:    s_ashr_i32 s2, s1, 31
 ; GFX6-NEXT:    s_add_u32 s0, s0, s2
 ; GFX6-NEXT:    s_mov_b32 s3, s2
@@ -9567,39 +9618,40 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    v_cvt_u32_f32_e32 v0, v0
 ; GFX6-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GFX6-NEXT:    v_mul_hi_u32 v2, s8, v0
-; GFX6-NEXT:    v_readfirstlane_b32 s12, v1
+; GFX6-NEXT:    v_readfirstlane_b32 s14, v1
 ; GFX6-NEXT:    v_readfirstlane_b32 s2, v0
-; GFX6-NEXT:    s_mul_i32 s1, s8, s12
+; GFX6-NEXT:    s_mul_i32 s1, s8, s14
 ; GFX6-NEXT:    v_readfirstlane_b32 s3, v2
 ; GFX6-NEXT:    s_mul_i32 s0, s9, s2
 ; GFX6-NEXT:    s_add_i32 s1, s3, s1
 ; GFX6-NEXT:    s_add_i32 s3, s1, s0
-; GFX6-NEXT:    s_mul_i32 s13, s8, s2
+; GFX6-NEXT:    s_mul_i32 s15, s8, s2
 ; GFX6-NEXT:    v_mul_hi_u32 v2, v0, s3
-; GFX6-NEXT:    v_mul_hi_u32 v0, v0, s13
+; GFX6-NEXT:    v_mul_hi_u32 v0, v0, s15
 ; GFX6-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
 ; GFX6-NEXT:    s_mul_i32 s4, s2, s3
 ; GFX6-NEXT:    v_readfirstlane_b32 s5, v2
 ; GFX6-NEXT:    v_readfirstlane_b32 s16, v0
-; GFX6-NEXT:    v_mul_hi_u32 v0, v1, s13
+; GFX6-NEXT:    v_mul_hi_u32 v0, v1, s15
 ; GFX6-NEXT:    v_mul_hi_u32 v1, v1, s3
 ; GFX6-NEXT:    s_add_u32 s4, s16, s4
 ; GFX6-NEXT:    s_addc_u32 s5, 0, s5
-; GFX6-NEXT:    s_mul_i32 s13, s12, s13
+; GFX6-NEXT:    s_mul_i32 s15, s14, s15
 ; GFX6-NEXT:    v_readfirstlane_b32 s16, v0
-; GFX6-NEXT:    s_add_u32 s4, s4, s13
+; GFX6-NEXT:    s_add_u32 s4, s4, s15
 ; GFX6-NEXT:    s_addc_u32 s4, s5, s16
 ; GFX6-NEXT:    v_readfirstlane_b32 s5, v1
 ; GFX6-NEXT:    s_addc_u32 s5, s5, 0
-; GFX6-NEXT:    s_mul_i32 s3, s12, s3
+; GFX6-NEXT:    s_mul_i32 s3, s14, s3
 ; GFX6-NEXT:    s_add_u32 s3, s4, s3
 ; GFX6-NEXT:    s_addc_u32 s4, 0, s5
 ; GFX6-NEXT:    s_add_u32 s5, s2, s3
 ; GFX6-NEXT:    v_mov_b32_e32 v0, s5
-; GFX6-NEXT:    v_mul_hi_u32 v0, s8, v0
 ; GFX6-NEXT:    s_cselect_b64 s[2:3], -1, 0
+; GFX6-NEXT:    v_mul_hi_u32 v0, s8, v0
 ; GFX6-NEXT:    s_or_b32 s2, s2, s3
-; GFX6-NEXT:    s_addc_u32 s4, s12, s4
+; GFX6-NEXT:    s_cmp_lg_u32 s2, 0
+; GFX6-NEXT:    s_addc_u32 s4, s14, s4
 ; GFX6-NEXT:    s_mul_i32 s2, s8, s4
 ; GFX6-NEXT:    v_readfirstlane_b32 s3, v0
 ; GFX6-NEXT:    s_add_i32 s2, s3, s2
@@ -9613,98 +9665,102 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX6-NEXT:    v_mul_hi_u32 v1, s4, v0
 ; GFX6-NEXT:    v_mul_hi_u32 v0, s5, v0
 ; GFX6-NEXT:    s_mul_i32 s9, s5, s2
-; GFX6-NEXT:    v_readfirstlane_b32 s13, v2
-; GFX6-NEXT:    s_add_u32 s9, s13, s9
-; GFX6-NEXT:    v_readfirstlane_b32 s12, v0
+; GFX6-NEXT:    v_readfirstlane_b32 s15, v2
+; GFX6-NEXT:    s_add_u32 s9, s15, s9
+; GFX6-NEXT:    v_readfirstlane_b32 s14, v0
 ; GFX6-NEXT:    s_mul_i32 s3, s4, s3
-; GFX6-NEXT:    s_addc_u32 s12, 0, s12
+; GFX6-NEXT:    s_addc_u32 s14, 0, s14
 ; GFX6-NEXT:    v_readfirstlane_b32 s8, v3
 ; GFX6-NEXT:    s_add_u32 s3, s9, s3
-; GFX6-NEXT:    s_addc_u32 s3, s12, s8
+; GFX6-NEXT:    s_addc_u32 s3, s14, s8
 ; GFX6-NEXT:    v_readfirstlane_b32 s8, v1
 ; GFX6-NEXT:    s_addc_u32 s8, s8, 0
 ; GFX6-NEXT:    s_mul_i32 s2, s4, s2
 ; GFX6-NEXT:    s_add_u32 s2, s3, s2
 ; GFX6-NEXT:    s_addc_u32 s8, 0, s8
-; GFX6-NEXT:    s_add_u32 s12, s5, s2
+; GFX6-NEXT:    s_add_u32 s14, s5, s2
 ; GFX6-NEXT:    s_cselect_b64 s[2:3], -1, 0
 ; GFX6-NEXT:    s_or_b32 s2, s2, s3
-; GFX6-NEXT:    s_addc_u32 s13, s4, s8
+; GFX6-NEXT:    s_cmp_lg_u32 s2, 0
+; GFX6-NEXT:    s_addc_u32 s15, s4, s8
 ; GFX6-NEXT:    s_ashr_i32 s4, s11, 31
 ; GFX6-NEXT:    s_add_u32 s2, s10, s4
 ; GFX6-NEXT:    s_mov_b32 s5, s4
 ; GFX6-NEXT:    s_addc_u32 s3, s11, s4
 ; GFX6-NEXT:    s_xor_b64 s[8:9], s[2:3], s[4:5]
-; GFX6-NEXT:    v_mov_b32_e32 v0, s13
+; GFX6-NEXT:    v_mov_b32_e32 v0, s15
 ; GFX6-NEXT:    v_mul_hi_u32 v1, s8, v0
-; GFX6-NEXT:    v_mov_b32_e32 v2, s12
+; GFX6-NEXT:    v_mov_b32_e32 v2, s14
 ; GFX6-NEXT:    v_mul_hi_u32 v3, s8, v2
-; GFX6-NEXT:    s_mul_i32 s2, s8, s13
+; GFX6-NEXT:    s_mul_i32 s2, s8, s15
 ; GFX6-NEXT:    v_readfirstlane_b32 s10, v1
 ; GFX6-NEXT:    v_mul_hi_u32 v1, s9, v2
 ; GFX6-NEXT:    v_readfirstlane_b32 s11, v3
 ; GFX6-NEXT:    v_mul_hi_u32 v0, s9, v0
 ; GFX6-NEXT:    s_add_u32 s2, s11, s2
 ; GFX6-NEXT:    s_addc_u32 s10, 0, s10
-; GFX6-NEXT:    s_mul_i32 s11, s9, s12
-; GFX6-NEXT:    v_readfirstlane_b32 s12, v1
+; GFX6-NEXT:    s_mul_i32 s11, s9, s14
+; GFX6-NEXT:    v_readfirstlane_b32 s14, v1
 ; GFX6-NEXT:    s_add_u32 s2, s2, s11
-; GFX6-NEXT:    s_addc_u32 s2, s10, s12
+; GFX6-NEXT:    s_addc_u32 s2, s10, s14
 ; GFX6-NEXT:    v_readfirstlane_b32 s10, v0
 ; GFX6-NEXT:    s_addc_u32 s10, s10, 0
-; GFX6-NEXT:    s_mul_i32 s11, s9, s13
+; GFX6-NEXT:    s_mul_i32 s11, s9, s15
 ; GFX6-NEXT:    s_add_u32 s11, s2, s11
 ; GFX6-NEXT:    v_mov_b32_e32 v0, s11
 ; GFX6-NEXT:    v_mul_hi_u32 v0, s6, v0
 ; GFX6-NEXT:    s_addc_u32 s10, 0, s10
 ; GFX6-NEXT:    s_mul_i32 s10, s6, s10
 ; GFX6-NEXT:    s_mov_b32 s3, 0xf000
-; GFX6-NEXT:    v_readfirstlane_b32 s12, v0
-; GFX6-NEXT:    s_add_i32 s10, s12, s10
-; GFX6-NEXT:    s_mul_i32 s12, s7, s11
-; GFX6-NEXT:    s_add_i32 s16, s10, s12
-; GFX6-NEXT:    s_sub_i32 s12, s9, s16
+; GFX6-NEXT:    v_readfirstlane_b32 s14, v0
+; GFX6-NEXT:    s_add_i32 s10, s14, s10
+; GFX6-NEXT:    s_mul_i32 s14, s7, s11
+; GFX6-NEXT:    s_add_i32 s14, s10, s14
+; GFX6-NEXT:    s_sub_i32 s15, s9, s14
 ; GFX6-NEXT:    s_mul_i32 s10, s6, s11
 ; GFX6-NEXT:    s_sub_u32 s8, s8, s10
 ; GFX6-NEXT:    s_cselect_b64 s[10:11], -1, 0
-; GFX6-NEXT:    s_or_b32 s13, s10, s11
-; GFX6-NEXT:    s_subb_u32 s17, s12, s7
-; GFX6-NEXT:    s_sub_u32 s18, s8, s6
-; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
-; GFX6-NEXT:    s_or_b32 s19, s12, s13
-; GFX6-NEXT:    s_subb_u32 s19, s17, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s19, s7
-; GFX6-NEXT:    s_cselect_b32 s20, -1, 0
-; GFX6-NEXT:    s_cmp_ge_u32 s18, s6
-; GFX6-NEXT:    s_cselect_b32 s21, -1, 0
-; GFX6-NEXT:    s_cmp_eq_u32 s19, s7
-; GFX6-NEXT:    s_cselect_b32 s20, s21, s20
-; GFX6-NEXT:    s_or_b32 s12, s12, s13
-; GFX6-NEXT:    s_subb_u32 s17, s17, s7
-; GFX6-NEXT:    s_sub_u32 s21, s18, s6
-; GFX6-NEXT:    s_cselect_b64 s[12:13], -1, 0
-; GFX6-NEXT:    s_or_b32 s12, s12, s13
-; GFX6-NEXT:    s_subb_u32 s12, s17, 0
-; GFX6-NEXT:    s_cmp_lg_u32 s20, 0
-; GFX6-NEXT:    s_cselect_b32 s13, s21, s18
-; GFX6-NEXT:    s_cselect_b32 s12, s12, s19
+; GFX6-NEXT:    s_or_b32 s16, s10, s11
+; GFX6-NEXT:    s_cmp_lg_u32 s16, 0
+; GFX6-NEXT:    s_subb_u32 s15, s15, s7
+; GFX6-NEXT:    s_sub_u32 s17, s8, s6
+; GFX6-NEXT:    s_cselect_b64 s[10:11], -1, 0
 ; GFX6-NEXT:    s_or_b32 s10, s10, s11
-; GFX6-NEXT:    s_subb_u32 s9, s9, s16
+; GFX6-NEXT:    s_cmp_lg_u32 s10, 0
+; GFX6-NEXT:    s_subb_u32 s18, s15, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s18, s7
+; GFX6-NEXT:    s_cselect_b32 s11, -1, 0
+; GFX6-NEXT:    s_cmp_ge_u32 s17, s6
+; GFX6-NEXT:    s_cselect_b32 s19, -1, 0
+; GFX6-NEXT:    s_cmp_eq_u32 s18, s7
+; GFX6-NEXT:    s_cselect_b32 s19, s19, s11
+; GFX6-NEXT:    s_cmp_lg_u32 s10, 0
+; GFX6-NEXT:    s_subb_u32 s15, s15, s7
+; GFX6-NEXT:    s_sub_u32 s20, s17, s6
+; GFX6-NEXT:    s_cselect_b64 s[10:11], -1, 0
+; GFX6-NEXT:    s_or_b32 s10, s10, s11
+; GFX6-NEXT:    s_cmp_lg_u32 s10, 0
+; GFX6-NEXT:    s_subb_u32 s10, s15, 0
+; GFX6-NEXT:    s_cmp_lg_u32 s19, 0
+; GFX6-NEXT:    s_cselect_b32 s11, s20, s17
+; GFX6-NEXT:    s_cselect_b32 s10, s10, s18
+; GFX6-NEXT:    s_cmp_lg_u32 s16, 0
+; GFX6-NEXT:    s_subb_u32 s9, s9, s14
 ; GFX6-NEXT:    s_cmp_ge_u32 s9, s7
-; GFX6-NEXT:    s_cselect_b32 s10, -1, 0
+; GFX6-NEXT:    s_cselect_b32 s14, -1, 0
 ; GFX6-NEXT:    s_cmp_ge_u32 s8, s6
 ; GFX6-NEXT:    s_cselect_b32 s6, -1, 0
 ; GFX6-NEXT:    s_cmp_eq_u32 s9, s7
-; GFX6-NEXT:    s_cselect_b32 s6, s6, s10
+; GFX6-NEXT:    s_cselect_b32 s6, s6, s14
 ; GFX6-NEXT:    s_cmp_lg_u32 s6, 0
-; GFX6-NEXT:    s_cselect_b32 s7, s12, s9
-; GFX6-NEXT:    s_cselect_b32 s6, s13, s8
+; GFX6-NEXT:    s_cselect_b32 s7, s10, s9
+; GFX6-NEXT:    s_cselect_b32 s6, s11, s8
 ; GFX6-NEXT:    s_xor_b64 s[6:7], s[6:7], s[4:5]
 ; GFX6-NEXT:    s_sub_u32 s5, s6, s4
 ; GFX6-NEXT:    s_subb_u32 s4, s7, s4
 ; GFX6-NEXT:    s_mov_b32 s2, -1
-; GFX6-NEXT:    v_mov_b32_e32 v0, s14
-; GFX6-NEXT:    v_mov_b32_e32 v1, s15
+; GFX6-NEXT:    v_mov_b32_e32 v0, s12
+; GFX6-NEXT:    v_mov_b32_e32 v1, s13
 ; GFX6-NEXT:    v_mov_b32_e32 v2, s5
 ; GFX6-NEXT:    v_mov_b32_e32 v3, s4
 ; GFX6-NEXT:    s_waitcnt lgkmcnt(0)
@@ -9724,8 +9780,8 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    s_xor_b64 s[2:3], s[2:3], s[6:7]
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v0, s2
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v1, s3
-; GFX9-NEXT:    s_sub_u32 s6, 0, s2
-; GFX9-NEXT:    s_subb_u32 s7, 0, s3
+; GFX9-NEXT:    s_sub_u32 s12, 0, s2
+; GFX9-NEXT:    s_subb_u32 s13, 0, s3
 ; GFX9-NEXT:    v_mac_f32_e32 v0, 0x4f800000, v1
 ; GFX9-NEXT:    v_rcp_f32_e32 v0, v0
 ; GFX9-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -9734,52 +9790,56 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    v_mac_f32_e32 v0, 0xcf800000, v1
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GFX9-NEXT:    v_readfirstlane_b32 s12, v1
-; GFX9-NEXT:    v_readfirstlane_b32 s13, v0
-; GFX9-NEXT:    s_mul_i32 s14, s6, s12
-; GFX9-NEXT:    s_mul_hi_u32 s16, s6, s13
-; GFX9-NEXT:    s_mul_i32 s15, s7, s13
-; GFX9-NEXT:    s_add_i32 s14, s16, s14
-; GFX9-NEXT:    s_mul_i32 s17, s6, s13
-; GFX9-NEXT:    s_add_i32 s14, s14, s15
-; GFX9-NEXT:    s_mul_hi_u32 s16, s13, s17
-; GFX9-NEXT:    s_mul_i32 s18, s13, s14
-; GFX9-NEXT:    s_mul_hi_u32 s15, s13, s14
+; GFX9-NEXT:    v_readfirstlane_b32 s14, v1
+; GFX9-NEXT:    v_readfirstlane_b32 s6, v0
+; GFX9-NEXT:    s_mul_i32 s7, s12, s14
+; GFX9-NEXT:    s_mul_hi_u32 s16, s12, s6
+; GFX9-NEXT:    s_mul_i32 s15, s13, s6
+; GFX9-NEXT:    s_add_i32 s7, s16, s7
+; GFX9-NEXT:    s_mul_i32 s17, s12, s6
+; GFX9-NEXT:    s_add_i32 s7, s7, s15
+; GFX9-NEXT:    s_mul_hi_u32 s16, s6, s17
+; GFX9-NEXT:    s_mul_i32 s18, s6, s7
+; GFX9-NEXT:    s_mul_hi_u32 s15, s6, s7
 ; GFX9-NEXT:    s_add_u32 s16, s16, s18
 ; GFX9-NEXT:    s_addc_u32 s15, 0, s15
-; GFX9-NEXT:    s_mul_hi_u32 s18, s12, s17
-; GFX9-NEXT:    s_mul_i32 s17, s12, s17
+; GFX9-NEXT:    s_mul_hi_u32 s18, s14, s17
+; GFX9-NEXT:    s_mul_i32 s17, s14, s17
 ; GFX9-NEXT:    s_add_u32 s16, s16, s17
-; GFX9-NEXT:    s_mul_hi_u32 s19, s12, s14
+; GFX9-NEXT:    s_mul_hi_u32 s19, s14, s7
 ; GFX9-NEXT:    s_addc_u32 s15, s15, s18
 ; GFX9-NEXT:    s_addc_u32 s16, s19, 0
-; GFX9-NEXT:    s_mul_i32 s14, s12, s14
-; GFX9-NEXT:    s_add_u32 s14, s15, s14
+; GFX9-NEXT:    s_mul_i32 s7, s14, s7
+; GFX9-NEXT:    s_add_u32 s7, s15, s7
 ; GFX9-NEXT:    s_addc_u32 s15, 0, s16
-; GFX9-NEXT:    s_add_u32 s13, s13, s14
-; GFX9-NEXT:    s_addc_u32 s12, s12, s15
-; GFX9-NEXT:    s_mul_i32 s14, s6, s12
-; GFX9-NEXT:    s_mul_hi_u32 s15, s6, s13
-; GFX9-NEXT:    s_add_i32 s14, s15, s14
-; GFX9-NEXT:    s_mul_i32 s7, s7, s13
-; GFX9-NEXT:    s_add_i32 s14, s14, s7
-; GFX9-NEXT:    s_mul_i32 s6, s6, s13
-; GFX9-NEXT:    s_mul_hi_u32 s15, s12, s6
-; GFX9-NEXT:    s_mul_i32 s16, s12, s6
-; GFX9-NEXT:    s_mul_i32 s18, s13, s14
-; GFX9-NEXT:    s_mul_hi_u32 s6, s13, s6
-; GFX9-NEXT:    s_mul_hi_u32 s17, s13, s14
-; GFX9-NEXT:    s_add_u32 s6, s6, s18
+; GFX9-NEXT:    s_add_u32 s16, s6, s7
+; GFX9-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[6:7], 0
+; GFX9-NEXT:    s_addc_u32 s14, s14, s15
+; GFX9-NEXT:    s_mul_i32 s6, s12, s14
+; GFX9-NEXT:    s_mul_hi_u32 s7, s12, s16
+; GFX9-NEXT:    s_add_i32 s6, s7, s6
+; GFX9-NEXT:    s_mul_i32 s13, s13, s16
+; GFX9-NEXT:    s_add_i32 s6, s6, s13
+; GFX9-NEXT:    s_mul_i32 s12, s12, s16
+; GFX9-NEXT:    s_mul_hi_u32 s13, s14, s12
+; GFX9-NEXT:    s_mul_i32 s15, s14, s12
+; GFX9-NEXT:    s_mul_i32 s18, s16, s6
+; GFX9-NEXT:    s_mul_hi_u32 s12, s16, s12
+; GFX9-NEXT:    s_mul_hi_u32 s17, s16, s6
+; GFX9-NEXT:    s_add_u32 s12, s12, s18
 ; GFX9-NEXT:    s_addc_u32 s17, 0, s17
-; GFX9-NEXT:    s_add_u32 s6, s6, s16
-; GFX9-NEXT:    s_mul_hi_u32 s7, s12, s14
-; GFX9-NEXT:    s_addc_u32 s6, s17, s15
+; GFX9-NEXT:    s_add_u32 s12, s12, s15
+; GFX9-NEXT:    s_mul_hi_u32 s7, s14, s6
+; GFX9-NEXT:    s_addc_u32 s12, s17, s13
 ; GFX9-NEXT:    s_addc_u32 s7, s7, 0
-; GFX9-NEXT:    s_mul_i32 s14, s12, s14
-; GFX9-NEXT:    s_add_u32 s6, s6, s14
-; GFX9-NEXT:    s_addc_u32 s7, 0, s7
-; GFX9-NEXT:    s_add_u32 s13, s13, s6
-; GFX9-NEXT:    s_addc_u32 s12, s12, s7
+; GFX9-NEXT:    s_mul_i32 s6, s14, s6
+; GFX9-NEXT:    s_add_u32 s6, s12, s6
+; GFX9-NEXT:    s_addc_u32 s12, 0, s7
+; GFX9-NEXT:    s_add_u32 s13, s16, s6
+; GFX9-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[6:7], 0
+; GFX9-NEXT:    s_addc_u32 s12, s14, s12
 ; GFX9-NEXT:    s_ashr_i32 s6, s9, 31
 ; GFX9-NEXT:    s_add_u32 s8, s8, s6
 ; GFX9-NEXT:    s_mov_b32 s7, s6
@@ -9808,9 +9868,11 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    s_mul_i32 s12, s2, s12
 ; GFX9-NEXT:    s_sub_u32 s8, s8, s12
 ; GFX9-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[12:13], 0
 ; GFX9-NEXT:    s_subb_u32 s17, s14, s3
 ; GFX9-NEXT:    s_sub_u32 s18, s8, s2
 ; GFX9-NEXT:    s_cselect_b64 s[14:15], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[14:15], 0
 ; GFX9-NEXT:    s_subb_u32 s19, s17, 0
 ; GFX9-NEXT:    s_cmp_ge_u32 s19, s3
 ; GFX9-NEXT:    s_cselect_b32 s20, -1, 0
@@ -9819,11 +9881,13 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    s_cmp_eq_u32 s19, s3
 ; GFX9-NEXT:    s_cselect_b32 s20, s21, s20
 ; GFX9-NEXT:    s_cmp_lg_u64 s[14:15], 0
-; GFX9-NEXT:    s_subb_u32 s14, s17, s3
-; GFX9-NEXT:    s_sub_u32 s15, s18, s2
-; GFX9-NEXT:    s_subb_u32 s14, s14, 0
+; GFX9-NEXT:    s_subb_u32 s17, s17, s3
+; GFX9-NEXT:    s_sub_u32 s21, s18, s2
+; GFX9-NEXT:    s_cselect_b64 s[14:15], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[14:15], 0
+; GFX9-NEXT:    s_subb_u32 s14, s17, 0
 ; GFX9-NEXT:    s_cmp_lg_u32 s20, 0
-; GFX9-NEXT:    s_cselect_b32 s15, s15, s18
+; GFX9-NEXT:    s_cselect_b32 s15, s21, s18
 ; GFX9-NEXT:    s_cselect_b32 s14, s14, s19
 ; GFX9-NEXT:    s_cmp_lg_u64 s[12:13], 0
 ; GFX9-NEXT:    s_subb_u32 s9, s9, s16
@@ -9847,8 +9911,8 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v0, s2
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v1, s3
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
-; GFX9-NEXT:    s_sub_u32 s4, 0, s2
-; GFX9-NEXT:    s_subb_u32 s5, 0, s3
+; GFX9-NEXT:    s_sub_u32 s6, 0, s2
+; GFX9-NEXT:    s_subb_u32 s7, 0, s3
 ; GFX9-NEXT:    v_mac_f32_e32 v0, 0x4f800000, v1
 ; GFX9-NEXT:    v_rcp_f32_e32 v1, v0
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 0
@@ -9858,70 +9922,74 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    v_mac_f32_e32 v1, 0xcf800000, v2
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v2, v2
-; GFX9-NEXT:    v_readfirstlane_b32 s6, v1
+; GFX9-NEXT:    v_readfirstlane_b32 s4, v1
 ; GFX9-NEXT:    v_readfirstlane_b32 s9, v2
-; GFX9-NEXT:    s_mul_hi_u32 s8, s4, s6
-; GFX9-NEXT:    s_mul_i32 s14, s4, s9
-; GFX9-NEXT:    s_mul_i32 s7, s5, s6
+; GFX9-NEXT:    s_mul_hi_u32 s8, s6, s4
+; GFX9-NEXT:    s_mul_i32 s14, s6, s9
+; GFX9-NEXT:    s_mul_i32 s5, s7, s4
 ; GFX9-NEXT:    s_add_i32 s8, s8, s14
-; GFX9-NEXT:    s_add_i32 s8, s8, s7
-; GFX9-NEXT:    s_mul_i32 s15, s4, s6
-; GFX9-NEXT:    s_mul_i32 s14, s6, s8
-; GFX9-NEXT:    s_mul_hi_u32 s16, s6, s15
-; GFX9-NEXT:    s_mul_hi_u32 s7, s6, s8
+; GFX9-NEXT:    s_add_i32 s8, s8, s5
+; GFX9-NEXT:    s_mul_i32 s15, s6, s4
+; GFX9-NEXT:    s_mul_i32 s14, s4, s8
+; GFX9-NEXT:    s_mul_hi_u32 s16, s4, s15
+; GFX9-NEXT:    s_mul_hi_u32 s5, s4, s8
 ; GFX9-NEXT:    s_add_u32 s14, s16, s14
-; GFX9-NEXT:    s_addc_u32 s7, 0, s7
+; GFX9-NEXT:    s_addc_u32 s5, 0, s5
 ; GFX9-NEXT:    s_mul_hi_u32 s17, s9, s15
 ; GFX9-NEXT:    s_mul_i32 s15, s9, s15
 ; GFX9-NEXT:    s_add_u32 s14, s14, s15
 ; GFX9-NEXT:    s_mul_hi_u32 s16, s9, s8
-; GFX9-NEXT:    s_addc_u32 s7, s7, s17
+; GFX9-NEXT:    s_addc_u32 s5, s5, s17
 ; GFX9-NEXT:    s_addc_u32 s14, s16, 0
 ; GFX9-NEXT:    s_mul_i32 s8, s9, s8
-; GFX9-NEXT:    s_add_u32 s7, s7, s8
+; GFX9-NEXT:    s_add_u32 s5, s5, s8
 ; GFX9-NEXT:    s_addc_u32 s8, 0, s14
-; GFX9-NEXT:    s_add_u32 s6, s6, s7
-; GFX9-NEXT:    s_addc_u32 s7, s9, s8
-; GFX9-NEXT:    s_mul_i32 s8, s4, s7
-; GFX9-NEXT:    s_mul_hi_u32 s9, s4, s6
-; GFX9-NEXT:    s_add_i32 s8, s9, s8
-; GFX9-NEXT:    s_mul_i32 s5, s5, s6
-; GFX9-NEXT:    s_add_i32 s8, s8, s5
-; GFX9-NEXT:    s_mul_i32 s4, s4, s6
-; GFX9-NEXT:    s_mul_hi_u32 s9, s7, s4
-; GFX9-NEXT:    s_mul_i32 s14, s7, s4
-; GFX9-NEXT:    s_mul_i32 s16, s6, s8
-; GFX9-NEXT:    s_mul_hi_u32 s4, s6, s4
-; GFX9-NEXT:    s_mul_hi_u32 s15, s6, s8
-; GFX9-NEXT:    s_add_u32 s4, s4, s16
+; GFX9-NEXT:    s_add_u32 s14, s4, s5
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX9-NEXT:    s_addc_u32 s8, s9, s8
+; GFX9-NEXT:    s_mul_i32 s4, s6, s8
+; GFX9-NEXT:    s_mul_hi_u32 s5, s6, s14
+; GFX9-NEXT:    s_add_i32 s4, s5, s4
+; GFX9-NEXT:    s_mul_i32 s7, s7, s14
+; GFX9-NEXT:    s_add_i32 s4, s4, s7
+; GFX9-NEXT:    s_mul_i32 s6, s6, s14
+; GFX9-NEXT:    s_mul_hi_u32 s7, s8, s6
+; GFX9-NEXT:    s_mul_i32 s9, s8, s6
+; GFX9-NEXT:    s_mul_i32 s16, s14, s4
+; GFX9-NEXT:    s_mul_hi_u32 s6, s14, s6
+; GFX9-NEXT:    s_mul_hi_u32 s15, s14, s4
+; GFX9-NEXT:    s_add_u32 s6, s6, s16
 ; GFX9-NEXT:    s_addc_u32 s15, 0, s15
-; GFX9-NEXT:    s_add_u32 s4, s4, s14
-; GFX9-NEXT:    s_mul_hi_u32 s5, s7, s8
-; GFX9-NEXT:    s_addc_u32 s4, s15, s9
+; GFX9-NEXT:    s_add_u32 s6, s6, s9
+; GFX9-NEXT:    s_mul_hi_u32 s5, s8, s4
+; GFX9-NEXT:    s_addc_u32 s6, s15, s7
 ; GFX9-NEXT:    s_addc_u32 s5, s5, 0
-; GFX9-NEXT:    s_mul_i32 s8, s7, s8
-; GFX9-NEXT:    s_add_u32 s4, s4, s8
-; GFX9-NEXT:    s_addc_u32 s5, 0, s5
-; GFX9-NEXT:    s_add_u32 s8, s6, s4
-; GFX9-NEXT:    s_addc_u32 s9, s7, s5
+; GFX9-NEXT:    s_mul_i32 s4, s8, s4
+; GFX9-NEXT:    s_add_u32 s4, s6, s4
+; GFX9-NEXT:    s_addc_u32 s6, 0, s5
+; GFX9-NEXT:    s_add_u32 s9, s14, s4
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX9-NEXT:    s_addc_u32 s8, s8, s6
 ; GFX9-NEXT:    s_ashr_i32 s4, s11, 31
 ; GFX9-NEXT:    s_add_u32 s6, s10, s4
 ; GFX9-NEXT:    s_mov_b32 s5, s4
 ; GFX9-NEXT:    s_addc_u32 s7, s11, s4
 ; GFX9-NEXT:    s_xor_b64 s[6:7], s[6:7], s[4:5]
-; GFX9-NEXT:    s_mul_i32 s11, s6, s9
-; GFX9-NEXT:    s_mul_hi_u32 s14, s6, s8
-; GFX9-NEXT:    s_mul_hi_u32 s10, s6, s9
+; GFX9-NEXT:    s_mul_i32 s11, s6, s8
+; GFX9-NEXT:    s_mul_hi_u32 s14, s6, s9
+; GFX9-NEXT:    s_mul_hi_u32 s10, s6, s8
 ; GFX9-NEXT:    s_add_u32 s11, s14, s11
 ; GFX9-NEXT:    s_addc_u32 s10, 0, s10
-; GFX9-NEXT:    s_mul_hi_u32 s15, s7, s8
-; GFX9-NEXT:    s_mul_i32 s8, s7, s8
-; GFX9-NEXT:    s_add_u32 s8, s11, s8
-; GFX9-NEXT:    s_mul_hi_u32 s14, s7, s9
-; GFX9-NEXT:    s_addc_u32 s8, s10, s15
-; GFX9-NEXT:    s_addc_u32 s10, s14, 0
+; GFX9-NEXT:    s_mul_hi_u32 s15, s7, s9
 ; GFX9-NEXT:    s_mul_i32 s9, s7, s9
-; GFX9-NEXT:    s_add_u32 s8, s8, s9
+; GFX9-NEXT:    s_add_u32 s9, s11, s9
+; GFX9-NEXT:    s_mul_hi_u32 s14, s7, s8
+; GFX9-NEXT:    s_addc_u32 s9, s10, s15
+; GFX9-NEXT:    s_addc_u32 s10, s14, 0
+; GFX9-NEXT:    s_mul_i32 s8, s7, s8
+; GFX9-NEXT:    s_add_u32 s8, s9, s8
 ; GFX9-NEXT:    s_addc_u32 s9, 0, s10
 ; GFX9-NEXT:    s_mul_i32 s9, s2, s9
 ; GFX9-NEXT:    s_mul_hi_u32 s10, s2, s8
@@ -9932,9 +10000,11 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    s_mul_i32 s8, s2, s8
 ; GFX9-NEXT:    s_sub_u32 s6, s6, s8
 ; GFX9-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[8:9], 0
 ; GFX9-NEXT:    s_subb_u32 s15, s10, s3
 ; GFX9-NEXT:    s_sub_u32 s16, s6, s2
 ; GFX9-NEXT:    s_cselect_b64 s[10:11], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
 ; GFX9-NEXT:    s_subb_u32 s17, s15, 0
 ; GFX9-NEXT:    s_cmp_ge_u32 s17, s3
 ; GFX9-NEXT:    s_cselect_b32 s18, -1, 0
@@ -9943,11 +10013,13 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
 ; GFX9-NEXT:    s_cmp_eq_u32 s17, s3
 ; GFX9-NEXT:    s_cselect_b32 s18, s19, s18
 ; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT:    s_subb_u32 s10, s15, s3
-; GFX9-NEXT:    s_sub_u32 s11, s16, s2
-; GFX9-NEXT:    s_subb_u32 s10, s10, 0
+; GFX9-NEXT:    s_subb_u32 s15, s15, s3
+; GFX9-NEXT:    s_sub_u32 s19, s16, s2
+; GFX9-NEXT:    s_cselect_b64 s[10:11], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
+; GFX9-NEXT:    s_subb_u32 s10, s15, 0
 ; GFX9-NEXT:    s_cmp_lg_u32 s18, 0
-; GFX9-NEXT:    s_cselect_b32 s11, s11, s16
+; GFX9-NEXT:    s_cselect_b32 s11, s19, s16
 ; GFX9-NEXT:    s_cselect_b32 s10, s10, s17
 ; GFX9-NEXT:    s_cmp_lg_u64 s[8:9], 0
 ; GFX9-NEXT:    s_subb_u32 s7, s7, s14
diff --git a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll
index 01f4414..394727c 100644
--- a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll
@@ -612,11 +612,12 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX8-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8-NEXT:    s_mov_b32 m0, s3
-; GFX8-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8-NEXT:    s_add_i32 s2, s2, s6
+; GFX8-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8-NEXT:    s_add_i32 s2, s2, s8
 ; GFX8-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX8-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -652,11 +653,12 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9-NEXT:    s_mov_b32 m0, s3
-; GFX9-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9-NEXT:    s_add_i32 s2, s2, s6
+; GFX9-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9-NEXT:    s_add_i32 s2, s2, s8
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -691,10 +693,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W64-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W64-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX10W64-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX10W64-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX10W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W64-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -730,10 +733,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W32-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W32-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX10W32-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX10W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX10W32-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX10W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX10W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX10W32-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX10W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W32-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -770,10 +774,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX11W64-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX11W64-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX11W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -813,10 +818,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX11W32-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX11W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX11W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX11W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX11W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX11W32-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX11W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -853,10 +859,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX12W64-NEXT:    s_wait_alu 0xfffe
 ; GFX12W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX12W64-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX12W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -894,15 +901,15 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W32-NEXT:    ; implicit-def: $vgpr0
 ; GFX12W32-NEXT:  .LBB2_1: ; %ComputeLoop
 ; GFX12W32-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX12W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX12W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX12W32-NEXT:    s_add_co_i32 s0, s0, s3
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
-; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX12W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX12W32-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX12W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -992,11 +999,12 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
 ; GFX8-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8-NEXT:    s_mov_b32 m0, s3
-; GFX8-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8-NEXT:    s_add_i32 s2, s2, s6
+; GFX8-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8-NEXT:    s_add_i32 s2, s2, s8
 ; GFX8-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX8-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1034,11 +1042,12 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9-NEXT:    s_mov_b32 m0, s3
-; GFX9-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9-NEXT:    s_add_i32 s2, s2, s6
+; GFX9-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9-NEXT:    s_add_i32 s2, s2, s8
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1075,10 +1084,11 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
 ; GFX10W64-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W64-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX10W64-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX10W64-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX10W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W64-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1117,10 +1127,11 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
 ; GFX10W32-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W32-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX10W32-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX10W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX10W32-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX10W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX10W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX10W32-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX10W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W32-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1160,10 +1171,11 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
 ; GFX11W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX11W64-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX11W64-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX11W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1206,10 +1218,11 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
 ; GFX11W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX11W32-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX11W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX11W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX11W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX11W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX11W32-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX11W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1248,10 +1261,11 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
 ; GFX12W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX12W64-NEXT:    s_wait_alu 0xfffe
 ; GFX12W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX12W64-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX12W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1292,15 +1306,15 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
 ; GFX12W32-NEXT:    ; implicit-def: $vgpr0
 ; GFX12W32-NEXT:  .LBB3_1: ; %ComputeLoop
 ; GFX12W32-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX12W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX12W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX12W32-NEXT:    s_add_co_i32 s0, s0, s3
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
-; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX12W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX12W32-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX12W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2059,11 +2073,12 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX8-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8-NEXT:    s_mov_b32 m0, s3
-; GFX8-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8-NEXT:    s_add_i32 s2, s2, s6
+; GFX8-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8-NEXT:    s_add_i32 s2, s2, s8
 ; GFX8-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX8-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2099,11 +2114,12 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9-NEXT:    s_mov_b32 m0, s3
-; GFX9-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9-NEXT:    s_add_i32 s2, s2, s6
+; GFX9-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9-NEXT:    s_add_i32 s2, s2, s8
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2138,10 +2154,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W64-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W64-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX10W64-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX10W64-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX10W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W64-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2177,10 +2194,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W32-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W32-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX10W32-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX10W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX10W32-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX10W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX10W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX10W32-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX10W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W32-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2217,10 +2235,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX11W64-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX11W64-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX11W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2260,10 +2279,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX11W32-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX11W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX11W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX11W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX11W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX11W32-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX11W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2301,10 +2321,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX12W64-NEXT:    s_wait_alu 0xfffe
 ; GFX12W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX12W64-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX12W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2342,15 +2363,15 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W32-NEXT:    ; implicit-def: $vgpr0
 ; GFX12W32-NEXT:  .LBB7_1: ; %ComputeLoop
 ; GFX12W32-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX12W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX12W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX12W32-NEXT:    s_add_co_i32 s0, s0, s3
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
-; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX12W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX12W32-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX12W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll
index 9db6d70..258bc295 100644
--- a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll
@@ -717,11 +717,12 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s2, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s2
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s6, m0
-; GFX8_ITERATIVE-NEXT:    s_add_i32 s6, s6, s3
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s7, v0, s2
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s6, m0
+; GFX8_ITERATIVE-NEXT:    s_add_i32 s6, s6, s7
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -761,11 +762,12 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s2, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s2
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s6, m0
-; GFX9_ITERATIVE-NEXT:    s_add_i32 s6, s6, s3
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s7, v0, s2
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s6, m0
+; GFX9_ITERATIVE-NEXT:    s_add_i32 s6, s6, s7
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -803,12 +805,13 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1064_ITERATIVE-NEXT:    ; implicit-def: $vgpr1
 ; GFX1064_ITERATIVE-NEXT:  .LBB2_1: ; %ComputeLoop
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s7, v0, s2
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s6, s2
-; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
-; GFX1064_ITERATIVE-NEXT:    s_add_i32 s6, s6, s7
+; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s7, s[0:1]
+; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s7
+; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s7
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s6, s7
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064_ITERATIVE-NEXT:    s_add_i32 s6, s6, s8
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -850,10 +853,11 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s1, s0
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s2, v0, s1
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s1
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s6, s1
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s0, s0, s3
 ; GFX1032_ITERATIVE-NEXT:    s_add_i32 s6, s6, s2
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -893,13 +897,14 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1164_ITERATIVE-NEXT:    ; implicit-def: $vgpr0
 ; GFX1164_ITERATIVE-NEXT:  .LBB2_1: ; %ComputeLoop
 ; GFX1164_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s2, s[0:1]
+; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s7, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s7, v1, s2
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s2
-; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
-; GFX1164_ITERATIVE-NEXT:    s_add_i32 s6, s6, s7
+; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s7
+; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s7
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s7
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164_ITERATIVE-NEXT:    s_add_i32 s6, s6, s8
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -944,10 +949,11 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s2, v1, s1
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s1
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s1
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s3
 ; GFX1132_ITERATIVE-NEXT:    s_add_i32 s6, s6, s2
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -987,14 +993,14 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1264_ITERATIVE-NEXT:    ; implicit-def: $vgpr0
 ; GFX1264_ITERATIVE-NEXT:  .LBB2_1: ; %ComputeLoop
 ; GFX1264_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1264_ITERATIVE-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1264_ITERATIVE-NEXT:    s_wait_alu 0xfffe
-; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s7, v1, s2
-; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s2
-; GFX1264_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
-; GFX1264_ITERATIVE-NEXT:    s_add_co_i32 s6, s6, s7
+; GFX1264_ITERATIVE-NEXT:    s_ctz_i32_b64 s7, s[0:1]
 ; GFX1264_ITERATIVE-NEXT:    s_wait_alu 0xfffe
+; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s7
+; GFX1264_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s7
+; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s7
 ; GFX1264_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1264_ITERATIVE-NEXT:    s_add_co_i32 s6, s6, s8
+; GFX1264_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1264_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX1264_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1264_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -1022,7 +1028,6 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1264_ITERATIVE-NEXT:    s_wait_kmcnt 0x0
 ; GFX1264_ITERATIVE-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX1264_ITERATIVE-NEXT:    s_mov_b32 s3, 0x31016000
-; GFX1264_ITERATIVE-NEXT:    s_wait_alu 0xf1ff
 ; GFX1264_ITERATIVE-NEXT:    v_add_nc_u32_e32 v0, s2, v0
 ; GFX1264_ITERATIVE-NEXT:    s_mov_b32 s2, -1
 ; GFX1264_ITERATIVE-NEXT:    buffer_store_b32 v0, off, s[0:3], null
@@ -1036,15 +1041,15 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1232_ITERATIVE-NEXT:    ; implicit-def: $vgpr0
 ; GFX1232_ITERATIVE-NEXT:  .LBB2_1: ; %ComputeLoop
 ; GFX1232_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
 ; GFX1232_ITERATIVE-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
 ; GFX1232_ITERATIVE-NEXT:    v_readlane_b32 s2, v1, s1
+; GFX1232_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s1
 ; GFX1232_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s1
-; GFX1232_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
+; GFX1232_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s3
 ; GFX1232_ITERATIVE-NEXT:    s_add_co_i32 s6, s6, s2
 ; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
-; GFX1232_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1232_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1232_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX1232_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1232_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -2358,6 +2363,7 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX8_ITERATIVE-NEXT:    s_addc_u32 s7, s7, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -2410,6 +2416,7 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX9_ITERATIVE-NEXT:    s_addc_u32 s7, s7, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -2455,12 +2462,13 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s2, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v3, s2
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s7, s2
 ; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s6, s2
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s7, s2
 ; GFX1064_ITERATIVE-NEXT:    s_add_u32 s6, s6, s3
 ; GFX1064_ITERATIVE-NEXT:    s_addc_u32 s7, s7, s8
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -2507,12 +2515,13 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s1, s0
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v3, s1
-; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v2, s7, s1
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s6, s1
+; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v2, s7, s1
 ; GFX1032_ITERATIVE-NEXT:    s_add_u32 s6, s6, s2
 ; GFX1032_ITERATIVE-NEXT:    s_addc_u32 s7, s7, s3
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -2560,13 +2569,14 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s3, v2, s2
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v3, s2
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s7, s2
 ; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s2
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s7, s2
 ; GFX1164_ITERATIVE-NEXT:    s_add_u32 s6, s6, s3
 ; GFX1164_ITERATIVE-NEXT:    s_addc_u32 s7, s7, s8
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
-; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -2616,13 +2626,14 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s2, v2, s1
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v3, s1
-; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v1, s7, s1
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s1
+; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v1, s7, s1
 ; GFX1132_ITERATIVE-NEXT:    s_add_u32 s6, s6, s2
 ; GFX1132_ITERATIVE-NEXT:    s_addc_u32 s7, s7, s3
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -2666,16 +2677,16 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1264_ITERATIVE-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1264_ITERATIVE-NEXT:  .LBB5_1: ; %ComputeLoop
 ; GFX1264_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1264_ITERATIVE-NEXT:    s_ctz_i32_b64 s8, s[0:1]
-; GFX1264_ITERATIVE-NEXT:    s_wait_alu 0xfffe
-; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s3, v3, s8
-; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s2, v2, s8
-; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v1, s7, s8
-; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s8
-; GFX1264_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
+; GFX1264_ITERATIVE-NEXT:    s_ctz_i32_b64 s10, s[0:1]
 ; GFX1264_ITERATIVE-NEXT:    s_wait_alu 0xfffe
+; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s3, v3, s10
+; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s2, v2, s10
+; GFX1264_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s10
+; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v1, s7, s10
+; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s10
 ; GFX1264_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[8:9]
 ; GFX1264_ITERATIVE-NEXT:    s_add_nc_u64 s[6:7], s[6:7], s[2:3]
+; GFX1264_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1264_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1264_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1264_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -2720,17 +2731,17 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1232_ITERATIVE-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1232_ITERATIVE-NEXT:  .LBB5_1: ; %ComputeLoop
 ; GFX1232_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
 ; GFX1232_ITERATIVE-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
 ; GFX1232_ITERATIVE-NEXT:    v_readlane_b32 s3, v3, s1
 ; GFX1232_ITERATIVE-NEXT:    v_readlane_b32 s2, v2, s1
+; GFX1232_ITERATIVE-NEXT:    s_lshl_b32 s8, 1, s1
 ; GFX1232_ITERATIVE-NEXT:    v_writelane_b32 v1, s7, s1
 ; GFX1232_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s1
-; GFX1232_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
-; GFX1232_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1232_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s8
 ; GFX1232_ITERATIVE-NEXT:    s_add_nc_u64 s[6:7], s[6:7], s[2:3]
+; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
+; GFX1232_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1232_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1232_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1232_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -4479,11 +4490,12 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s2, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s2
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v2, s12, m0
-; GFX8_ITERATIVE-NEXT:    s_add_i32 s12, s12, s3
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s2
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v2, s12, m0
+; GFX8_ITERATIVE-NEXT:    s_add_i32 s12, s12, s6
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -4538,11 +4550,12 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s2, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s2
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v2, s12, m0
-; GFX9_ITERATIVE-NEXT:    s_add_i32 s12, s12, s3
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s2
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v2, s12, m0
+; GFX9_ITERATIVE-NEXT:    s_add_i32 s12, s12, s6
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -4595,12 +4608,13 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1064_ITERATIVE-NEXT:    ; implicit-def: $vgpr2
 ; GFX1064_ITERATIVE-NEXT:  .LBB8_1: ; %ComputeLoop
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s2
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s12, s2
-; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
-; GFX1064_ITERATIVE-NEXT:    s_add_i32 s12, s12, s6
+; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s6, s[0:1]
+; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s7, v0, s6
+; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s6
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s12, s6
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064_ITERATIVE-NEXT:    s_add_i32 s12, s12, s7
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -4656,10 +4670,11 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s1, s0
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s2, v0, s1
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s1
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v2, s8, s1
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s0, s0, s3
 ; GFX1032_ITERATIVE-NEXT:    s_add_i32 s8, s8, s2
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -4713,13 +4728,14 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1164_ITERATIVE-NEXT:    ; implicit-def: $vgpr2
 ; GFX1164_ITERATIVE-NEXT:  .LBB8_1: ; %ComputeLoop
 ; GFX1164_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s2, s[0:1]
+; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s6, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s2
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v2, s12, s2
-; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
-; GFX1164_ITERATIVE-NEXT:    s_add_i32 s12, s12, s6
+; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s7, v0, s6
+; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s6
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v2, s12, s6
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164_ITERATIVE-NEXT:    s_add_i32 s12, s12, s7
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -4783,10 +4799,11 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s2, v0, s1
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s1
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v2, s8, s1
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s3
 ; GFX1132_ITERATIVE-NEXT:    s_add_i32 s8, s8, s2
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -4844,14 +4861,14 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1264_ITERATIVE-NEXT:    ; implicit-def: $vgpr0
 ; GFX1264_ITERATIVE-NEXT:  .LBB8_1: ; %ComputeLoop
 ; GFX1264_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1264_ITERATIVE-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1264_ITERATIVE-NEXT:    s_wait_alu 0xfffe
-; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s7, v1, s2
-; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s2
-; GFX1264_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
-; GFX1264_ITERATIVE-NEXT:    s_add_co_i32 s6, s6, s7
+; GFX1264_ITERATIVE-NEXT:    s_ctz_i32_b64 s7, s[0:1]
 ; GFX1264_ITERATIVE-NEXT:    s_wait_alu 0xfffe
+; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s7
+; GFX1264_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s7
+; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s7
 ; GFX1264_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1264_ITERATIVE-NEXT:    s_add_co_i32 s6, s6, s8
+; GFX1264_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1264_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1264_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1264_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -4879,7 +4896,6 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1264_ITERATIVE-NEXT:    s_wait_kmcnt 0x0
 ; GFX1264_ITERATIVE-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX1264_ITERATIVE-NEXT:    s_mov_b32 s3, 0x31016000
-; GFX1264_ITERATIVE-NEXT:    s_wait_alu 0xf1ff
 ; GFX1264_ITERATIVE-NEXT:    v_sub_nc_u32_e32 v0, s2, v0
 ; GFX1264_ITERATIVE-NEXT:    s_mov_b32 s2, -1
 ; GFX1264_ITERATIVE-NEXT:    buffer_store_b32 v0, off, s[0:3], null
@@ -4893,15 +4909,15 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1232_ITERATIVE-NEXT:    ; implicit-def: $vgpr0
 ; GFX1232_ITERATIVE-NEXT:  .LBB8_1: ; %ComputeLoop
 ; GFX1232_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
 ; GFX1232_ITERATIVE-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
 ; GFX1232_ITERATIVE-NEXT:    v_readlane_b32 s2, v1, s1
+; GFX1232_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s1
 ; GFX1232_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s1
-; GFX1232_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
+; GFX1232_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s3
 ; GFX1232_ITERATIVE-NEXT:    s_add_co_i32 s6, s6, s2
 ; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
-; GFX1232_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1232_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1232_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1232_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1232_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -6657,6 +6673,7 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX8_ITERATIVE-NEXT:    s_addc_u32 s9, s9, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -6729,6 +6746,7 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX9_ITERATIVE-NEXT:    s_addc_u32 s9, s9, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -6794,12 +6812,13 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s2, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s6, v1, s2
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v5, s9, s2
 ; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v4, s8, s2
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v5, s9, s2
 ; GFX1064_ITERATIVE-NEXT:    s_add_u32 s8, s8, s3
 ; GFX1064_ITERATIVE-NEXT:    s_addc_u32 s9, s9, s6
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -6864,12 +6883,13 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s1, s0
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s1
-; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v5, s9, s1
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v4, s8, s1
+; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v5, s9, s1
 ; GFX1032_ITERATIVE-NEXT:    s_add_u32 s8, s8, s2
 ; GFX1032_ITERATIVE-NEXT:    s_addc_u32 s9, s9, s3
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -6935,13 +6955,14 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s6, v1, s2
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v5, s9, s2
 ; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v4, s8, s2
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v5, s9, s2
 ; GFX1164_ITERATIVE-NEXT:    s_add_u32 s8, s8, s3
 ; GFX1164_ITERATIVE-NEXT:    s_addc_u32 s9, s9, s6
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[2:3], 1, s2
-; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -7015,13 +7036,14 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s1
-; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v5, s9, s1
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v4, s8, s1
+; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v5, s9, s1
 ; GFX1132_ITERATIVE-NEXT:    s_add_u32 s8, s8, s2
 ; GFX1132_ITERATIVE-NEXT:    s_addc_u32 s9, s9, s3
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -7087,16 +7109,16 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1264_ITERATIVE-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1264_ITERATIVE-NEXT:  .LBB11_1: ; %ComputeLoop
 ; GFX1264_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1264_ITERATIVE-NEXT:    s_ctz_i32_b64 s8, s[0:1]
-; GFX1264_ITERATIVE-NEXT:    s_wait_alu 0xfffe
-; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s3, v3, s8
-; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s2, v2, s8
-; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v1, s7, s8
-; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s8
-; GFX1264_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
+; GFX1264_ITERATIVE-NEXT:    s_ctz_i32_b64 s10, s[0:1]
 ; GFX1264_ITERATIVE-NEXT:    s_wait_alu 0xfffe
+; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s3, v3, s10
+; GFX1264_ITERATIVE-NEXT:    v_readlane_b32 s2, v2, s10
+; GFX1264_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s10
+; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v1, s7, s10
+; GFX1264_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s10
 ; GFX1264_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[8:9]
 ; GFX1264_ITERATIVE-NEXT:    s_add_nc_u64 s[6:7], s[6:7], s[2:3]
+; GFX1264_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1264_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1264_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1264_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
@@ -7141,17 +7163,17 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
 ; GFX1232_ITERATIVE-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1232_ITERATIVE-NEXT:  .LBB11_1: ; %ComputeLoop
 ; GFX1232_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
 ; GFX1232_ITERATIVE-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
 ; GFX1232_ITERATIVE-NEXT:    v_readlane_b32 s3, v3, s1
 ; GFX1232_ITERATIVE-NEXT:    v_readlane_b32 s2, v2, s1
+; GFX1232_ITERATIVE-NEXT:    s_lshl_b32 s8, 1, s1
 ; GFX1232_ITERATIVE-NEXT:    v_writelane_b32 v1, s7, s1
 ; GFX1232_ITERATIVE-NEXT:    v_writelane_b32 v0, s6, s1
-; GFX1232_ITERATIVE-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
-; GFX1232_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1232_ITERATIVE-NEXT:    s_and_not1_b32 s0, s0, s8
 ; GFX1232_ITERATIVE-NEXT:    s_add_nc_u64 s[6:7], s[6:7], s[2:3]
+; GFX1232_ITERATIVE-NEXT:    s_wait_alu 0xfffe
+; GFX1232_ITERATIVE-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1232_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1232_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1232_ITERATIVE-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
diff --git a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll
index 6167a84..23c5f4f 100644
--- a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll
@@ -499,11 +499,12 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -539,11 +540,12 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -578,10 +580,11 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -618,10 +621,11 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX1032_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -659,10 +663,11 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -702,10 +707,11 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX1132_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1082,10 +1088,11 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
 ; GFX8_ITERATIVE-NEXT:  .LBB3_1: ; %ComputeLoop
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s4, v0, s3
-; GFX8_ITERATIVE-NEXT:    s_add_i32 s2, s2, s4
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s3
+; GFX8_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[4:5]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1110,10 +1117,11 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
 ; GFX9_ITERATIVE-NEXT:  .LBB3_1: ; %ComputeLoop
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s4, v0, s3
-; GFX9_ITERATIVE-NEXT:    s_add_i32 s2, s2, s4
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s3
+; GFX9_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[4:5]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1139,8 +1147,9 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[4:5]
+; GFX1064_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1167,8 +1176,9 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
-; GFX1032_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1196,8 +1206,10 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[4:5]
+; GFX1164_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1227,8 +1239,10 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
-; GFX1132_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2008,6 +2022,7 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s7
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s6
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2056,6 +2071,7 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s7
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s6
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2096,12 +2112,13 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s6, s[2:3]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s7, v0, s6
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v3, s6
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s6
 ; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s6
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s6
 ; GFX1064_ITERATIVE-NEXT:    s_add_u32 s0, s0, s7
 ; GFX1064_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s8
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s6
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2143,12 +2160,13 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s3, s2
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s3
-; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s3
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s3
+; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s3
 ; GFX1032_ITERATIVE-NEXT:    s_add_u32 s0, s0, s6
 ; GFX1032_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s7
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2191,13 +2209,14 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s7, v2, s6
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v3, s6
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s6
 ; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s6
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s6
 ; GFX1164_ITERATIVE-NEXT:    s_add_u32 s0, s0, s7
 ; GFX1164_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s8
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s6
-; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[2:3], s[2:3], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -2242,13 +2261,14 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s3
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s3
-; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s3
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s3
+; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s3
 ; GFX1132_ITERATIVE-NEXT:    s_add_u32 s0, s0, s6
 ; GFX1132_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s7
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
-; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s3
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -2861,6 +2881,7 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
 ; GFX8_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s5
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s4
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[4:5]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2893,6 +2914,7 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
 ; GFX9_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s5
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s4
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[4:5]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2924,6 +2946,7 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
 ; GFX1064_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s6
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s4
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[4:5]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2956,6 +2979,7 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
 ; GFX1032_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s5
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2989,6 +3013,8 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
 ; GFX1164_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s6
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s4
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[2:3], s[2:3], s[4:5]
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3022,8 +3048,9 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
 ; GFX1132_ITERATIVE-NEXT:    s_add_u32 s0, s0, s4
 ; GFX1132_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s5
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
-; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s3
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3879,11 +3906,12 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3919,11 +3947,12 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3958,10 +3987,11 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3998,10 +4028,11 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX1032_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4039,10 +4070,11 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_add_i32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -4082,10 +4114,11 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX1132_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -4462,10 +4495,11 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
 ; GFX8_ITERATIVE-NEXT:  .LBB11_1: ; %ComputeLoop
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s4, v0, s3
-; GFX8_ITERATIVE-NEXT:    s_add_i32 s2, s2, s4
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s3
+; GFX8_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[4:5]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4490,10 +4524,11 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
 ; GFX9_ITERATIVE-NEXT:  .LBB11_1: ; %ComputeLoop
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s4, v0, s3
-; GFX9_ITERATIVE-NEXT:    s_add_i32 s2, s2, s4
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s3
+; GFX9_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[4:5]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4519,8 +4554,9 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[4:5]
+; GFX1064_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4547,8 +4583,9 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
-; GFX1032_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4576,8 +4613,10 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[4:5], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[4:5]
+; GFX1164_ITERATIVE-NEXT:    s_add_i32 s2, s2, s6
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4607,8 +4646,10 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
-; GFX1132_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_add_i32 s0, s0, s3
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5411,6 +5452,7 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s7
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s6
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5459,6 +5501,7 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s7
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s6
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5499,12 +5542,13 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s6, s[2:3]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s7, v0, s6
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v3, s6
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s6
 ; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s6
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s6
 ; GFX1064_ITERATIVE-NEXT:    s_add_u32 s0, s0, s7
 ; GFX1064_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s8
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s6
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5546,12 +5590,13 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s3, s2
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s3
-; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s3
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s3
+; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s3
 ; GFX1032_ITERATIVE-NEXT:    s_add_u32 s0, s0, s6
 ; GFX1032_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s7
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5594,13 +5639,14 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s7, v2, s6
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v3, s6
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s6
 ; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s6
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s6
 ; GFX1164_ITERATIVE-NEXT:    s_add_u32 s0, s0, s7
 ; GFX1164_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s8
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s6
-; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[2:3], s[2:3], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -5645,13 +5691,14 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s3
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s3
-; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s3
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s3
+; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s3
 ; GFX1132_ITERATIVE-NEXT:    s_add_u32 s0, s0, s6
 ; GFX1132_ITERATIVE-NEXT:    s_addc_u32 s1, s1, s7
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
-; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s3
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -6266,11 +6313,12 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT:    s_and_b32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8_ITERATIVE-NEXT:    s_and_b32 s2, s2, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6306,11 +6354,12 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT:    s_and_b32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9_ITERATIVE-NEXT:    s_and_b32 s2, s2, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6345,10 +6394,11 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_and_b32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_and_b32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6385,10 +6435,11 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX1032_ITERATIVE-NEXT:    s_and_b32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6426,10 +6477,11 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_and_b32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_and_b32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -6469,10 +6521,11 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX1132_ITERATIVE-NEXT:    s_and_b32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -6873,11 +6926,12 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s8
 ; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
 ; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s8
+; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
 ; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, m0
 ; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, m0
 ; GFX8_ITERATIVE-NEXT:    s_and_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
-; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB16_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6919,11 +6973,12 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s8
 ; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
 ; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s8
+; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
 ; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, m0
 ; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, m0
 ; GFX9_ITERATIVE-NEXT:    s_and_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
-; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB16_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6960,14 +7015,15 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1064_ITERATIVE-NEXT:  .LBB16_1: ; %ComputeLoop
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s8, s[2:3]
-; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
-; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s8
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s8
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s8
-; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
-; GFX1064_ITERATIVE-NEXT:    s_and_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s10, s[2:3]
+; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s10
+; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s10
+; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s10
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s10
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s10
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX1064_ITERATIVE-NEXT:    s_and_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB16_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7009,11 +7065,12 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s3, s2
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s3
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s8, 1, s3
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s3
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s8
 ; GFX1032_ITERATIVE-NEXT:    s_and_b64 s[0:1], s[0:1], s[6:7]
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB16_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7052,15 +7109,16 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1164_ITERATIVE-NEXT:  .LBB16_1: ; %ComputeLoop
 ; GFX1164_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s8, s[2:3]
+; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s10, s[2:3]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
-; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s8
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s8
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s8
-; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
-; GFX1164_ITERATIVE-NEXT:    s_and_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s10
+; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s10
+; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s10
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s10
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s10
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[2:3], s[2:3], s[8:9]
+; GFX1164_ITERATIVE-NEXT:    s_and_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB16_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -7105,11 +7163,12 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s3
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s3
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s8, 1, s3
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s3
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s8
 ; GFX1132_ITERATIVE-NEXT:    s_and_b64 s[0:1], s[0:1], s[6:7]
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s3
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB16_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -7613,11 +7672,12 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT:    s_or_b32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8_ITERATIVE-NEXT:    s_or_b32 s2, s2, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7653,11 +7713,12 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT:    s_or_b32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9_ITERATIVE-NEXT:    s_or_b32 s2, s2, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7692,10 +7753,11 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_or_b32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_or_b32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7732,10 +7794,11 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX1032_ITERATIVE-NEXT:    s_or_b32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7773,10 +7836,11 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_or_b32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_or_b32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -7816,10 +7880,11 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX1132_ITERATIVE-NEXT:    s_or_b32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -8219,11 +8284,12 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s8
 ; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
 ; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s8
+; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
 ; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, m0
 ; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, m0
 ; GFX8_ITERATIVE-NEXT:    s_or_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
-; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB18_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8265,11 +8331,12 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s8
 ; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
 ; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s8
+; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
 ; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, m0
 ; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, m0
 ; GFX9_ITERATIVE-NEXT:    s_or_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
-; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB18_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8306,14 +8373,15 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1064_ITERATIVE-NEXT:  .LBB18_1: ; %ComputeLoop
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s8, s[2:3]
-; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
-; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s8
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s8
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s8
-; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
-; GFX1064_ITERATIVE-NEXT:    s_or_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s10, s[2:3]
+; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s10
+; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s10
+; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s10
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s10
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s10
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX1064_ITERATIVE-NEXT:    s_or_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB18_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8355,11 +8423,12 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s3, s2
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s3
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s8, 1, s3
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s3
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s8
 ; GFX1032_ITERATIVE-NEXT:    s_or_b64 s[0:1], s[0:1], s[6:7]
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB18_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8398,15 +8467,16 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1164_ITERATIVE-NEXT:  .LBB18_1: ; %ComputeLoop
 ; GFX1164_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s8, s[2:3]
+; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s10, s[2:3]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
-; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s8
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s8
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s8
-; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
-; GFX1164_ITERATIVE-NEXT:    s_or_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s10
+; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s10
+; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s10
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s10
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s10
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[2:3], s[2:3], s[8:9]
+; GFX1164_ITERATIVE-NEXT:    s_or_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB18_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8451,11 +8521,12 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s3
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s3
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s8, 1, s3
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s3
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s8
 ; GFX1132_ITERATIVE-NEXT:    s_or_b64 s[0:1], s[0:1], s[6:7]
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s3
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB18_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8959,11 +9030,12 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT:    s_xor_b32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8_ITERATIVE-NEXT:    s_xor_b32 s2, s2, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB19_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8999,11 +9071,12 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT:    s_xor_b32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9_ITERATIVE-NEXT:    s_xor_b32 s2, s2, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB19_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9038,10 +9111,11 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_xor_b32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_xor_b32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB19_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9078,10 +9152,11 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX1032_ITERATIVE-NEXT:    s_xor_b32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB19_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9119,10 +9194,11 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_xor_b32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_xor_b32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB19_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -9162,10 +9238,11 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX1132_ITERATIVE-NEXT:    s_xor_b32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB19_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -9565,11 +9642,12 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s8
 ; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
 ; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s8
+; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
 ; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, m0
 ; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, m0
 ; GFX8_ITERATIVE-NEXT:    s_xor_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
-; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB20_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9611,11 +9689,12 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s8
 ; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
 ; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s8
+; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
 ; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, m0
 ; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, m0
 ; GFX9_ITERATIVE-NEXT:    s_xor_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
-; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB20_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9652,14 +9731,15 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1064_ITERATIVE-NEXT:  .LBB20_1: ; %ComputeLoop
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s8, s[2:3]
-; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
-; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s8
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s8
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s8
-; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
-; GFX1064_ITERATIVE-NEXT:    s_xor_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s10, s[2:3]
+; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s10
+; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s10
+; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s10
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s10
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s10
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX1064_ITERATIVE-NEXT:    s_xor_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB20_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9701,11 +9781,12 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s3, s2
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s3
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s8, 1, s3
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v2, s1, s3
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s8
 ; GFX1032_ITERATIVE-NEXT:    s_xor_b64 s[0:1], s[0:1], s[6:7]
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB20_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9744,15 +9825,16 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1164_ITERATIVE-NEXT:  .LBB20_1: ; %ComputeLoop
 ; GFX1164_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s8, s[2:3]
+; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s10, s[2:3]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s8
-; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s8
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s8
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s8
-; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s8
-; GFX1164_ITERATIVE-NEXT:    s_xor_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s10
+; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s10
+; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[8:9], 1, s10
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s10
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s10
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[2:3], s[2:3], s[8:9]
+; GFX1164_ITERATIVE-NEXT:    s_xor_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB20_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9797,11 +9879,12 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s7, v3, s3
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s6, v2, s3
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s8, 1, s3
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v1, s1, s3
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s8
 ; GFX1132_ITERATIVE-NEXT:    s_xor_b64 s[0:1], s[0:1], s[6:7]
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s3
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB20_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -10305,11 +10388,12 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT:    s_max_i32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8_ITERATIVE-NEXT:    s_max_i32 s2, s2, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB21_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10345,11 +10429,12 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT:    s_max_i32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9_ITERATIVE-NEXT:    s_max_i32 s2, s2, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB21_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10384,10 +10469,11 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_max_i32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_max_i32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB21_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10424,10 +10510,11 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX1032_ITERATIVE-NEXT:    s_max_i32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB21_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10465,10 +10552,11 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_max_i32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_max_i32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB21_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -10508,10 +10596,11 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX1132_ITERATIVE-NEXT:    s_max_i32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB21_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -11166,6 +11255,7 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s10
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB23_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11221,6 +11311,7 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s10
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB23_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11272,6 +11363,7 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s10
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB23_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11323,6 +11415,7 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB23_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11375,8 +11468,9 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_cselect_b32 s1, s1, s7
 ; GFX1164_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s10
-; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[2:3], s[2:3], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB23_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -11431,8 +11525,9 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_cselect_b32 s1, s1, s7
 ; GFX1132_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
-; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s3
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB23_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -12119,11 +12214,12 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT:    s_min_i32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8_ITERATIVE-NEXT:    s_min_i32 s2, s2, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB24_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12159,11 +12255,12 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT:    s_min_i32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9_ITERATIVE-NEXT:    s_min_i32 s2, s2, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB24_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12198,10 +12295,11 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_min_i32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_min_i32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB24_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12238,10 +12336,11 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX1032_ITERATIVE-NEXT:    s_min_i32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB24_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12279,10 +12378,11 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_min_i32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_min_i32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB24_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -12322,10 +12422,11 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX1132_ITERATIVE-NEXT:    s_min_i32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB24_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -12980,6 +13081,7 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s10
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB26_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -13035,6 +13137,7 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s10
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB26_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -13086,6 +13189,7 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s10
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB26_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -13137,6 +13241,7 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB26_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -13189,8 +13294,9 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_cselect_b32 s1, s1, s7
 ; GFX1164_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s10
-; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[2:3], s[2:3], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB26_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -13245,8 +13351,9 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_cselect_b32 s1, s1, s7
 ; GFX1132_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
-; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s3
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB26_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -13933,11 +14040,12 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT:    s_max_u32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8_ITERATIVE-NEXT:    s_max_u32 s2, s2, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB27_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -13973,11 +14081,12 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT:    s_max_u32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9_ITERATIVE-NEXT:    s_max_u32 s2, s2, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB27_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14012,10 +14121,11 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_max_u32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_max_u32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB27_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14052,10 +14162,11 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX1032_ITERATIVE-NEXT:    s_max_u32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB27_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14093,10 +14204,11 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_max_u32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_max_u32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB27_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -14136,10 +14248,11 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX1132_ITERATIVE-NEXT:    s_max_u32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB27_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -14788,6 +14901,7 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s10
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB29_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14842,6 +14956,7 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s10
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB29_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14892,6 +15007,7 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s10
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB29_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14942,6 +15058,7 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB29_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14995,6 +15112,8 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s10
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[2:3], s[2:3], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB29_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -15050,6 +15169,8 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s3
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB29_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -15732,11 +15853,12 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT:    s_min_u32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8_ITERATIVE-NEXT:    s_min_u32 s2, s2, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB30_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -15772,11 +15894,12 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9_ITERATIVE-NEXT:    s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT:    s_min_u32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9_ITERATIVE-NEXT:    s_min_u32 s2, s2, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB30_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -15811,10 +15934,11 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064_ITERATIVE-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX1064_ITERATIVE-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1064_ITERATIVE-NEXT:    s_min_u32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_min_u32 s2, s2, s8
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB30_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -15851,10 +15975,11 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1032_ITERATIVE-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX1032_ITERATIVE-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1032_ITERATIVE-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX1032_ITERATIVE-NEXT:    s_min_u32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB30_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -15892,10 +16017,11 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1164_ITERATIVE-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX1164_ITERATIVE-NEXT:    s_min_u32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_min_u32 s2, s2, s8
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB30_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -15935,10 +16061,11 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX1132_ITERATIVE-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX1132_ITERATIVE-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX1132_ITERATIVE-NEXT:    s_min_u32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB30_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -16588,6 +16715,7 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
 ; GFX8_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s10
 ; GFX8_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
 ; GFX8_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX8_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX8_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB32_1
 ; GFX8_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -16642,6 +16770,7 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
 ; GFX9_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s10
 ; GFX9_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s8
 ; GFX9_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX9_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX9_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB32_1
 ; GFX9_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -16692,6 +16821,7 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
 ; GFX1064_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1064_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s10
 ; GFX1064_ITERATIVE-NEXT:    s_andn2_b64 s[2:3], s[2:3], s[6:7]
+; GFX1064_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1064_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB32_1
 ; GFX1064_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -16742,6 +16872,7 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
 ; GFX1032_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1032_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX1032_ITERATIVE-NEXT:    s_andn2_b32 s2, s2, s3
+; GFX1032_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1032_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB32_1
 ; GFX1032_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -16795,6 +16926,8 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
 ; GFX1164_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1164_ITERATIVE-NEXT:    s_lshl_b64 s[6:7], 1, s10
 ; GFX1164_ITERATIVE-NEXT:    s_and_not1_b64 s[2:3], s[2:3], s[6:7]
+; GFX1164_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; GFX1164_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB32_1
 ; GFX1164_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -16850,6 +16983,8 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
 ; GFX1132_ITERATIVE-NEXT:    s_cselect_b32 s0, s0, s6
 ; GFX1132_ITERATIVE-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX1132_ITERATIVE-NEXT:    s_and_not1_b32 s2, s2, s3
+; GFX1132_ITERATIVE-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1132_ITERATIVE-NEXT:    s_cbranch_scc1 .LBB32_1
 ; GFX1132_ITERATIVE-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132_ITERATIVE-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll
index 9afc0c6..e4def28 100644
--- a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll
@@ -611,11 +611,12 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX8-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8-NEXT:    s_mov_b32 m0, s3
-; GFX8-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8-NEXT:    s_add_i32 s2, s2, s6
+; GFX8-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8-NEXT:    s_add_i32 s2, s2, s8
 ; GFX8-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX8-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -651,11 +652,12 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9-NEXT:    s_mov_b32 m0, s3
-; GFX9-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9-NEXT:    s_add_i32 s2, s2, s6
+; GFX9-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9-NEXT:    s_add_i32 s2, s2, s8
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -690,10 +692,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W64-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W64-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX10W64-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX10W64-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX10W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W64-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -729,10 +732,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W32-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W32-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX10W32-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX10W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX10W32-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX10W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX10W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX10W32-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX10W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W32-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -769,10 +773,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX11W64-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX11W64-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX11W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -812,10 +817,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX11W32-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX11W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX11W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX11W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX11W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX11W32-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX11W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -852,10 +858,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX12W64-NEXT:    s_wait_alu 0xfffe
 ; GFX12W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX12W64-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX12W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -893,15 +900,15 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W32-NEXT:    ; implicit-def: $vgpr0
 ; GFX12W32-NEXT:  .LBB2_1: ; %ComputeLoop
 ; GFX12W32-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX12W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX12W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX12W32-NEXT:    s_add_co_i32 s0, s0, s3
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
-; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX12W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX12W32-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX12W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1658,11 +1665,12 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX8-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8-NEXT:    s_mov_b32 m0, s3
-; GFX8-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8-NEXT:    s_add_i32 s2, s2, s6
+; GFX8-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8-NEXT:    s_add_i32 s2, s2, s8
 ; GFX8-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX8-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1698,11 +1706,12 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9-NEXT:    s_mov_b32 m0, s3
-; GFX9-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9-NEXT:    s_add_i32 s2, s2, s6
+; GFX9-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9-NEXT:    s_add_i32 s2, s2, s8
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1737,10 +1746,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W64-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W64-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX10W64-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX10W64-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX10W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W64-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1776,10 +1786,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W32-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W32-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX10W32-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX10W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX10W32-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX10W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX10W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX10W32-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX10W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W32-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1816,10 +1827,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX11W64-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX11W64-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX11W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1859,10 +1871,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX11W32-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX11W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX11W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX11W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX11W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX11W32-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX11W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1900,10 +1913,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX12W64-NEXT:    s_wait_alu 0xfffe
 ; GFX12W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX12W64-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX12W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1941,15 +1955,15 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W32-NEXT:    ; implicit-def: $vgpr0
 ; GFX12W32-NEXT:  .LBB6_1: ; %ComputeLoop
 ; GFX12W32-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX12W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX12W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX12W32-NEXT:    s_add_co_i32 s0, s0, s3
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
-; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX12W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX12W32-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX12W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll
index 10fd34f..39a3c9a 100644
--- a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll
@@ -628,11 +628,12 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX8-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8-NEXT:    s_mov_b32 m0, s3
-; GFX8-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8-NEXT:    s_add_i32 s2, s2, s6
+; GFX8-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8-NEXT:    s_add_i32 s2, s2, s8
 ; GFX8-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX8-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -669,11 +670,12 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9-NEXT:    s_mov_b32 m0, s3
-; GFX9-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9-NEXT:    s_add_i32 s2, s2, s6
+; GFX9-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9-NEXT:    s_add_i32 s2, s2, s8
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -709,10 +711,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W64-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W64-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX10W64-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX10W64-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX10W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W64-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -749,10 +752,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W32-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W32-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX10W32-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX10W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX10W32-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX10W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX10W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX10W32-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX10W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W32-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -790,10 +794,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX11W64-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX11W64-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX11W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -834,10 +839,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX11W32-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX11W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX11W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX11W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX11W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX11W32-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX11W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -874,10 +880,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX12W64-NEXT:    s_wait_alu 0xfffe
 ; GFX12W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX12W64-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX12W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -916,15 +923,15 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W32-NEXT:    ; implicit-def: $vgpr0
 ; GFX12W32-NEXT:  .LBB2_1: ; %ComputeLoop
 ; GFX12W32-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX12W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX12W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX12W32-NEXT:    s_add_co_i32 s0, s0, s3
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
-; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX12W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX12W32-NEXT:    s_cbranch_scc1 .LBB2_1
 ; GFX12W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1826,11 +1833,12 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX8-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX8-NEXT:    s_mov_b32 m0, s3
-; GFX8-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX8-NEXT:    s_add_i32 s2, s2, s6
+; GFX8-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX8-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX8-NEXT:    s_add_i32 s2, s2, s8
 ; GFX8-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX8-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX8-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX8-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1867,11 +1875,12 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX9-NEXT:    s_mov_b32 m0, s3
-; GFX9-NEXT:    v_readlane_b32 s6, v0, s3
-; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
-; GFX9-NEXT:    s_add_i32 s2, s2, s6
+; GFX9-NEXT:    v_readlane_b32 s8, v0, s3
 ; GFX9-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT:    v_writelane_b32 v1, s2, m0
+; GFX9-NEXT:    s_add_i32 s2, s2, s8
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1907,10 +1916,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W64-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W64-NEXT:    s_ff1_i32_b64 s3, s[0:1]
 ; GFX10W64-NEXT:    v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    v_writelane_b32 v1, s2, s3
 ; GFX10W64-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX10W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX10W64-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX10W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W64-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1947,10 +1957,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX10W32-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX10W32-NEXT:    s_ff1_i32_b32 s2, s1
 ; GFX10W32-NEXT:    v_readlane_b32 s3, v0, s2
+; GFX10W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX10W32-NEXT:    v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s6
 ; GFX10W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX10W32-NEXT:    s_andn2_b32 s1, s1, s2
+; GFX10W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX10W32-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX10W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX10W32-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1988,10 +1999,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX11W64-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX11W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT:    s_add_i32 s2, s2, s8
+; GFX11W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX11W64-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX11W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2032,10 +2044,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX11W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX11W32-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
 ; GFX11W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX11W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX11W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX11W32-NEXT:    s_add_i32 s0, s0, s3
-; GFX11W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX11W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX11W32-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX11W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX11W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2073,10 +2086,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W64-NEXT:    s_ctz_i32_b64 s3, s[0:1]
 ; GFX12W64-NEXT:    s_wait_alu 0xfffe
 ; GFX12W64-NEXT:    v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_lshl_b64 s[6:7], 1, s3
-; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    v_writelane_b32 v0, s2, s3
 ; GFX12W64-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT:    s_add_co_i32 s2, s2, s8
+; GFX12W64-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX12W64-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX12W64-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W64-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2115,15 +2129,15 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
 ; GFX12W32-NEXT:    ; implicit-def: $vgpr0
 ; GFX12W32-NEXT:  .LBB7_1: ; %ComputeLoop
 ; GFX12W32-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    s_ctz_i32_b32 s2, s1
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
 ; GFX12W32-NEXT:    v_readlane_b32 s3, v1, s2
+; GFX12W32-NEXT:    s_lshl_b32 s6, 1, s2
 ; GFX12W32-NEXT:    v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT:    s_lshl_b32 s2, 1, s2
+; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s6
 ; GFX12W32-NEXT:    s_add_co_i32 s0, s0, s3
 ; GFX12W32-NEXT:    s_wait_alu 0xfffe
-; GFX12W32-NEXT:    s_and_not1_b32 s1, s1, s2
+; GFX12W32-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX12W32-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX12W32-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX12W32-NEXT:    v_mbcnt_lo_u32_b32 v1, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/carryout-selection.ll b/llvm/test/CodeGen/AMDGPU/carryout-selection.ll
index b96de17..4a6fa4f 100644
--- a/llvm/test/CodeGen/AMDGPU/carryout-selection.ll
+++ b/llvm/test/CodeGen/AMDGPU/carryout-selection.ll
@@ -704,6 +704,7 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; CISI-NEXT:    s_add_u32 s4, s4, s6
 ; CISI-NEXT:    s_cselect_b64 s[12:13], -1, 0
 ; CISI-NEXT:    s_or_b32 s6, s12, s13
+; CISI-NEXT:    s_cmp_lg_u32 s6, 0
 ; CISI-NEXT:    s_addc_u32 s5, s5, s7
 ; CISI-NEXT:    s_mov_b32 s8, s0
 ; CISI-NEXT:    s_mov_b32 s9, s1
@@ -724,14 +725,16 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx8 s[0:7], s[4:5], 0x24
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v2, s2
+; VI-NEXT:    s_add_u32 s2, s4, s6
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
-; VI-NEXT:    s_add_u32 s0, s4, s6
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_addc_u32 s1, s5, s7
-; VI-NEXT:    v_mov_b32_e32 v4, s0
-; VI-NEXT:    v_mov_b32_e32 v5, s1
 ; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
-; VI-NEXT:    v_mov_b32_e32 v2, s2
+; VI-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; VI-NEXT:    s_addc_u32 s0, s5, s7
+; VI-NEXT:    v_mov_b32_e32 v4, s2
+; VI-NEXT:    v_mov_b32_e32 v5, s0
+; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; VI-NEXT:    v_mov_b32_e32 v3, s3
 ; VI-NEXT:    flat_store_dwordx2 v[0:1], v[4:5]
 ; VI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -743,10 +746,12 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX9-NEXT:    s_load_dwordx8 s[8:15], s[4:5], 0x24
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_add_u32 s0, s12, s14
-; GFX9-NEXT:    s_addc_u32 s1, s13, s15
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    s_add_u32 s2, s12, s14
+; GFX9-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    s_addc_u32 s0, s13, s15
+; GFX9-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s0
 ; GFX9-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GFX9-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s[0:1]
 ; GFX9-NEXT:    global_store_dwordx2 v2, v[0:1], s[8:9]
@@ -759,8 +764,10 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX1010-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1010-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1010-NEXT:    s_add_u32 s0, s12, s14
-; GFX1010-NEXT:    s_addc_u32 s1, s13, s15
+; GFX1010-NEXT:    s_cselect_b32 s1, -1, 0
 ; GFX1010-NEXT:    v_mov_b32_e32 v0, s0
+; GFX1010-NEXT:    s_cmp_lg_u32 s1, 0
+; GFX1010-NEXT:    s_addc_u32 s1, s13, s15
 ; GFX1010-NEXT:    s_cselect_b32 s0, -1, 0
 ; GFX1010-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX1010-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s0
@@ -774,8 +781,10 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX1030W32-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1030W32-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1030W32-NEXT:    s_add_u32 s4, s4, s6
-; GFX1030W32-NEXT:    s_addc_u32 s5, s5, s7
+; GFX1030W32-NEXT:    s_cselect_b32 s6, -1, 0
 ; GFX1030W32-NEXT:    v_mov_b32_e32 v0, s4
+; GFX1030W32-NEXT:    s_cmp_lg_u32 s6, 0
+; GFX1030W32-NEXT:    s_addc_u32 s5, s5, s7
 ; GFX1030W32-NEXT:    s_cselect_b32 s4, -1, 0
 ; GFX1030W32-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX1030W32-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s4
@@ -789,8 +798,10 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX1030W64-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1030W64-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1030W64-NEXT:    s_add_u32 s4, s4, s6
-; GFX1030W64-NEXT:    s_addc_u32 s5, s5, s7
+; GFX1030W64-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GFX1030W64-NEXT:    v_mov_b32_e32 v0, s4
+; GFX1030W64-NEXT:    s_cmp_lg_u64 s[8:9], 0
+; GFX1030W64-NEXT:    s_addc_u32 s5, s5, s7
 ; GFX1030W64-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX1030W64-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GFX1030W64-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s[4:5]
@@ -803,8 +814,10 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX11-NEXT:    s_load_b256 s[0:7], s[4:5], 0x24
 ; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-NEXT:    s_add_u32 s4, s4, s6
-; GFX11-NEXT:    s_addc_u32 s5, s5, s7
+; GFX11-NEXT:    s_cselect_b32 s6, -1, 0
 ; GFX11-NEXT:    v_mov_b32_e32 v0, s4
+; GFX11-NEXT:    s_cmp_lg_u32 s6, 0
+; GFX11-NEXT:    s_addc_u32 s5, s5, s7
 ; GFX11-NEXT:    s_cselect_b32 s4, -1, 0
 ; GFX11-NEXT:    v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s5
 ; GFX11-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s4
@@ -818,8 +831,10 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX1250-NEXT:    s_load_b256 s[8:15], s[4:5], 0x24
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    s_add_co_u32 s0, s12, s14
-; GFX1250-NEXT:    s_add_co_ci_u32 s1, s13, s15
+; GFX1250-NEXT:    s_cselect_b32 s1, -1, 0
 ; GFX1250-NEXT:    v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v0, s0
+; GFX1250-NEXT:    s_cmp_lg_u32 s1, 0
+; GFX1250-NEXT:    s_add_co_ci_u32 s1, s13, s15
 ; GFX1250-NEXT:    s_cselect_b32 s0, -1, 0
 ; GFX1250-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX1250-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s0
@@ -1676,6 +1691,7 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; CISI-NEXT:    s_sub_u32 s4, s4, s6
 ; CISI-NEXT:    s_cselect_b64 s[12:13], -1, 0
 ; CISI-NEXT:    s_or_b32 s6, s12, s13
+; CISI-NEXT:    s_cmp_lg_u32 s6, 0
 ; CISI-NEXT:    s_subb_u32 s5, s5, s7
 ; CISI-NEXT:    s_mov_b32 s8, s0
 ; CISI-NEXT:    s_mov_b32 s9, s1
@@ -1696,14 +1712,16 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx8 s[0:7], s[4:5], 0x24
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v2, s2
+; VI-NEXT:    s_sub_u32 s2, s4, s6
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
-; VI-NEXT:    s_sub_u32 s0, s4, s6
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_subb_u32 s1, s5, s7
-; VI-NEXT:    v_mov_b32_e32 v4, s0
-; VI-NEXT:    v_mov_b32_e32 v5, s1
 ; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
-; VI-NEXT:    v_mov_b32_e32 v2, s2
+; VI-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; VI-NEXT:    s_subb_u32 s0, s5, s7
+; VI-NEXT:    v_mov_b32_e32 v4, s2
+; VI-NEXT:    v_mov_b32_e32 v5, s0
+; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; VI-NEXT:    v_mov_b32_e32 v3, s3
 ; VI-NEXT:    flat_store_dwordx2 v[0:1], v[4:5]
 ; VI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -1715,10 +1733,12 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX9-NEXT:    s_load_dwordx8 s[8:15], s[4:5], 0x24
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_sub_u32 s0, s12, s14
-; GFX9-NEXT:    s_subb_u32 s1, s13, s15
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    s_sub_u32 s2, s12, s14
+; GFX9-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    s_subb_u32 s0, s13, s15
+; GFX9-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s0
 ; GFX9-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GFX9-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s[0:1]
 ; GFX9-NEXT:    global_store_dwordx2 v2, v[0:1], s[8:9]
@@ -1731,8 +1751,10 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX1010-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1010-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1010-NEXT:    s_sub_u32 s0, s12, s14
-; GFX1010-NEXT:    s_subb_u32 s1, s13, s15
+; GFX1010-NEXT:    s_cselect_b32 s1, -1, 0
 ; GFX1010-NEXT:    v_mov_b32_e32 v0, s0
+; GFX1010-NEXT:    s_cmp_lg_u32 s1, 0
+; GFX1010-NEXT:    s_subb_u32 s1, s13, s15
 ; GFX1010-NEXT:    s_cselect_b32 s0, -1, 0
 ; GFX1010-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX1010-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s0
@@ -1746,8 +1768,10 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX1030W32-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1030W32-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1030W32-NEXT:    s_sub_u32 s4, s4, s6
-; GFX1030W32-NEXT:    s_subb_u32 s5, s5, s7
+; GFX1030W32-NEXT:    s_cselect_b32 s6, -1, 0
 ; GFX1030W32-NEXT:    v_mov_b32_e32 v0, s4
+; GFX1030W32-NEXT:    s_cmp_lg_u32 s6, 0
+; GFX1030W32-NEXT:    s_subb_u32 s5, s5, s7
 ; GFX1030W32-NEXT:    s_cselect_b32 s4, -1, 0
 ; GFX1030W32-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX1030W32-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s4
@@ -1761,8 +1785,10 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX1030W64-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1030W64-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1030W64-NEXT:    s_sub_u32 s4, s4, s6
-; GFX1030W64-NEXT:    s_subb_u32 s5, s5, s7
+; GFX1030W64-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GFX1030W64-NEXT:    v_mov_b32_e32 v0, s4
+; GFX1030W64-NEXT:    s_cmp_lg_u64 s[8:9], 0
+; GFX1030W64-NEXT:    s_subb_u32 s5, s5, s7
 ; GFX1030W64-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX1030W64-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GFX1030W64-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s[4:5]
@@ -1775,8 +1801,10 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX11-NEXT:    s_load_b256 s[0:7], s[4:5], 0x24
 ; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-NEXT:    s_sub_u32 s4, s4, s6
-; GFX11-NEXT:    s_subb_u32 s5, s5, s7
+; GFX11-NEXT:    s_cselect_b32 s6, -1, 0
 ; GFX11-NEXT:    v_mov_b32_e32 v0, s4
+; GFX11-NEXT:    s_cmp_lg_u32 s6, 0
+; GFX11-NEXT:    s_subb_u32 s5, s5, s7
 ; GFX11-NEXT:    s_cselect_b32 s4, -1, 0
 ; GFX11-NEXT:    v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s5
 ; GFX11-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s4
@@ -1790,8 +1818,10 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
 ; GFX1250-NEXT:    s_load_b256 s[8:15], s[4:5], 0x24
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    s_sub_co_u32 s0, s12, s14
-; GFX1250-NEXT:    s_sub_co_ci_u32 s1, s13, s15
+; GFX1250-NEXT:    s_cselect_b32 s1, -1, 0
 ; GFX1250-NEXT:    v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v0, s0
+; GFX1250-NEXT:    s_cmp_lg_u32 s1, 0
+; GFX1250-NEXT:    s_sub_co_ci_u32 s1, s13, s15
 ; GFX1250-NEXT:    s_cselect_b32 s0, -1, 0
 ; GFX1250-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX1250-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s0
@@ -2188,46 +2218,49 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; VI-NEXT:    s_addc_u32 s6, s7, s9
 ; VI-NEXT:    s_addc_u32 s8, s8, 0
 ; VI-NEXT:    v_readfirstlane_b32 s7, v0
-; VI-NEXT:    s_add_u32 s10, s6, s7
-; VI-NEXT:    v_mov_b32_e32 v0, s10
+; VI-NEXT:    s_add_u32 s12, s6, s7
+; VI-NEXT:    v_mov_b32_e32 v0, s12
 ; VI-NEXT:    v_mad_u64_u32 v[0:1], s[6:7], s4, v0, 0
-; VI-NEXT:    s_addc_u32 s11, 0, s8
-; VI-NEXT:    s_mul_i32 s8, s4, s11
+; VI-NEXT:    s_addc_u32 s13, 0, s8
+; VI-NEXT:    s_mul_i32 s8, s4, s13
 ; VI-NEXT:    v_readfirstlane_b32 s9, v1
 ; VI-NEXT:    s_add_i32 s8, s9, s8
-; VI-NEXT:    s_mul_i32 s9, s5, s10
-; VI-NEXT:    s_add_i32 s12, s8, s9
-; VI-NEXT:    s_sub_i32 s13, s3, s12
+; VI-NEXT:    s_mul_i32 s9, s5, s12
+; VI-NEXT:    s_add_i32 s14, s8, s9
+; VI-NEXT:    s_sub_i32 s10, s3, s14
 ; VI-NEXT:    v_readfirstlane_b32 s8, v0
-; VI-NEXT:    s_sub_u32 s14, s2, s8
+; VI-NEXT:    s_sub_u32 s15, s2, s8
 ; VI-NEXT:    s_cselect_b64 s[8:9], -1, 0
-; VI-NEXT:    s_subb_u32 s13, s13, s5
-; VI-NEXT:    s_sub_u32 s15, s14, s4
-; VI-NEXT:    s_subb_u32 s13, s13, 0
-; VI-NEXT:    s_cmp_ge_u32 s13, s5
+; VI-NEXT:    s_cmp_lg_u64 s[8:9], 0
+; VI-NEXT:    s_subb_u32 s16, s10, s5
+; VI-NEXT:    s_sub_u32 s17, s15, s4
+; VI-NEXT:    s_cselect_b64 s[10:11], -1, 0
+; VI-NEXT:    s_cmp_lg_u64 s[10:11], 0
+; VI-NEXT:    s_subb_u32 s10, s16, 0
+; VI-NEXT:    s_cmp_ge_u32 s10, s5
+; VI-NEXT:    s_cselect_b32 s11, -1, 0
+; VI-NEXT:    s_cmp_ge_u32 s17, s4
 ; VI-NEXT:    s_cselect_b32 s16, -1, 0
-; VI-NEXT:    s_cmp_ge_u32 s15, s4
-; VI-NEXT:    s_cselect_b32 s15, -1, 0
-; VI-NEXT:    s_cmp_eq_u32 s13, s5
-; VI-NEXT:    s_cselect_b32 s13, s15, s16
-; VI-NEXT:    s_add_u32 s15, s10, 1
-; VI-NEXT:    s_addc_u32 s16, s11, 0
-; VI-NEXT:    s_add_u32 s17, s10, 2
-; VI-NEXT:    s_addc_u32 s18, s11, 0
-; VI-NEXT:    s_cmp_lg_u32 s13, 0
-; VI-NEXT:    s_cselect_b32 s13, s17, s15
-; VI-NEXT:    s_cselect_b32 s15, s18, s16
+; VI-NEXT:    s_cmp_eq_u32 s10, s5
+; VI-NEXT:    s_cselect_b32 s10, s16, s11
+; VI-NEXT:    s_add_u32 s11, s12, 1
+; VI-NEXT:    s_addc_u32 s16, s13, 0
+; VI-NEXT:    s_add_u32 s17, s12, 2
+; VI-NEXT:    s_addc_u32 s18, s13, 0
+; VI-NEXT:    s_cmp_lg_u32 s10, 0
+; VI-NEXT:    s_cselect_b32 s10, s17, s11
+; VI-NEXT:    s_cselect_b32 s11, s18, s16
 ; VI-NEXT:    s_cmp_lg_u64 s[8:9], 0
-; VI-NEXT:    s_subb_u32 s3, s3, s12
+; VI-NEXT:    s_subb_u32 s3, s3, s14
 ; VI-NEXT:    s_cmp_ge_u32 s3, s5
 ; VI-NEXT:    s_cselect_b32 s8, -1, 0
-; VI-NEXT:    s_cmp_ge_u32 s14, s4
+; VI-NEXT:    s_cmp_ge_u32 s15, s4
 ; VI-NEXT:    s_cselect_b32 s9, -1, 0
 ; VI-NEXT:    s_cmp_eq_u32 s3, s5
 ; VI-NEXT:    s_cselect_b32 s3, s9, s8
 ; VI-NEXT:    s_cmp_lg_u32 s3, 0
-; VI-NEXT:    s_cselect_b32 s9, s15, s11
-; VI-NEXT:    s_cselect_b32 s8, s13, s10
+; VI-NEXT:    s_cselect_b32 s9, s11, s13
+; VI-NEXT:    s_cselect_b32 s8, s10, s12
 ; VI-NEXT:    s_cbranch_execnz .LBB16_4
 ; VI-NEXT:  .LBB16_2:
 ; VI-NEXT:    v_cvt_f32_u32_e32 v0, s4
@@ -2278,8 +2311,8 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v0, s6
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v1, s7
-; GFX9-NEXT:    s_sub_u32 s8, 0, s6
-; GFX9-NEXT:    s_subb_u32 s9, 0, s7
+; GFX9-NEXT:    s_sub_u32 s10, 0, s6
+; GFX9-NEXT:    s_subb_u32 s11, 0, s7
 ; GFX9-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GFX9-NEXT:    v_rcp_f32_e32 v0, v0
 ; GFX9-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -2288,102 +2321,109 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX9-NEXT:    v_madmk_f32 v0, v1, 0xcf800000, v0
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GFX9-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GFX9-NEXT:    v_readfirstlane_b32 s10, v1
-; GFX9-NEXT:    v_readfirstlane_b32 s11, v0
-; GFX9-NEXT:    s_mul_i32 s12, s8, s10
-; GFX9-NEXT:    s_mul_hi_u32 s14, s8, s11
-; GFX9-NEXT:    s_mul_i32 s13, s9, s11
-; GFX9-NEXT:    s_add_i32 s12, s14, s12
-; GFX9-NEXT:    s_add_i32 s12, s12, s13
-; GFX9-NEXT:    s_mul_i32 s15, s8, s11
-; GFX9-NEXT:    s_mul_i32 s14, s11, s12
-; GFX9-NEXT:    s_mul_hi_u32 s16, s11, s15
-; GFX9-NEXT:    s_mul_hi_u32 s13, s11, s12
+; GFX9-NEXT:    v_readfirstlane_b32 s12, v1
+; GFX9-NEXT:    v_readfirstlane_b32 s8, v0
+; GFX9-NEXT:    s_mul_i32 s9, s10, s12
+; GFX9-NEXT:    s_mul_hi_u32 s14, s10, s8
+; GFX9-NEXT:    s_mul_i32 s13, s11, s8
+; GFX9-NEXT:    s_add_i32 s9, s14, s9
+; GFX9-NEXT:    s_add_i32 s9, s9, s13
+; GFX9-NEXT:    s_mul_i32 s15, s10, s8
+; GFX9-NEXT:    s_mul_i32 s14, s8, s9
+; GFX9-NEXT:    s_mul_hi_u32 s16, s8, s15
+; GFX9-NEXT:    s_mul_hi_u32 s13, s8, s9
 ; GFX9-NEXT:    s_add_u32 s14, s16, s14
 ; GFX9-NEXT:    s_addc_u32 s13, 0, s13
-; GFX9-NEXT:    s_mul_hi_u32 s17, s10, s15
-; GFX9-NEXT:    s_mul_i32 s15, s10, s15
+; GFX9-NEXT:    s_mul_hi_u32 s17, s12, s15
+; GFX9-NEXT:    s_mul_i32 s15, s12, s15
 ; GFX9-NEXT:    s_add_u32 s14, s14, s15
-; GFX9-NEXT:    s_mul_hi_u32 s16, s10, s12
+; GFX9-NEXT:    s_mul_hi_u32 s16, s12, s9
 ; GFX9-NEXT:    s_addc_u32 s13, s13, s17
 ; GFX9-NEXT:    s_addc_u32 s14, s16, 0
-; GFX9-NEXT:    s_mul_i32 s12, s10, s12
-; GFX9-NEXT:    s_add_u32 s12, s13, s12
+; GFX9-NEXT:    s_mul_i32 s9, s12, s9
+; GFX9-NEXT:    s_add_u32 s9, s13, s9
 ; GFX9-NEXT:    s_addc_u32 s13, 0, s14
-; GFX9-NEXT:    s_add_u32 s11, s11, s12
-; GFX9-NEXT:    s_addc_u32 s10, s10, s13
-; GFX9-NEXT:    s_mul_i32 s12, s8, s10
-; GFX9-NEXT:    s_mul_hi_u32 s13, s8, s11
-; GFX9-NEXT:    s_add_i32 s12, s13, s12
-; GFX9-NEXT:    s_mul_i32 s9, s9, s11
-; GFX9-NEXT:    s_add_i32 s12, s12, s9
-; GFX9-NEXT:    s_mul_i32 s8, s8, s11
-; GFX9-NEXT:    s_mul_hi_u32 s13, s10, s8
-; GFX9-NEXT:    s_mul_i32 s14, s10, s8
-; GFX9-NEXT:    s_mul_i32 s16, s11, s12
-; GFX9-NEXT:    s_mul_hi_u32 s8, s11, s8
-; GFX9-NEXT:    s_mul_hi_u32 s15, s11, s12
-; GFX9-NEXT:    s_add_u32 s8, s8, s16
+; GFX9-NEXT:    s_add_u32 s14, s8, s9
+; GFX9-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[8:9], 0
+; GFX9-NEXT:    s_addc_u32 s12, s12, s13
+; GFX9-NEXT:    s_mul_i32 s8, s10, s12
+; GFX9-NEXT:    s_mul_hi_u32 s9, s10, s14
+; GFX9-NEXT:    s_add_i32 s8, s9, s8
+; GFX9-NEXT:    s_mul_i32 s11, s11, s14
+; GFX9-NEXT:    s_add_i32 s8, s8, s11
+; GFX9-NEXT:    s_mul_i32 s10, s10, s14
+; GFX9-NEXT:    s_mul_hi_u32 s11, s12, s10
+; GFX9-NEXT:    s_mul_i32 s13, s12, s10
+; GFX9-NEXT:    s_mul_i32 s16, s14, s8
+; GFX9-NEXT:    s_mul_hi_u32 s10, s14, s10
+; GFX9-NEXT:    s_mul_hi_u32 s15, s14, s8
+; GFX9-NEXT:    s_add_u32 s10, s10, s16
 ; GFX9-NEXT:    s_addc_u32 s15, 0, s15
-; GFX9-NEXT:    s_add_u32 s8, s8, s14
-; GFX9-NEXT:    s_mul_hi_u32 s9, s10, s12
-; GFX9-NEXT:    s_addc_u32 s8, s15, s13
+; GFX9-NEXT:    s_add_u32 s10, s10, s13
+; GFX9-NEXT:    s_mul_hi_u32 s9, s12, s8
+; GFX9-NEXT:    s_addc_u32 s10, s15, s11
 ; GFX9-NEXT:    s_addc_u32 s9, s9, 0
-; GFX9-NEXT:    s_mul_i32 s12, s10, s12
-; GFX9-NEXT:    s_add_u32 s8, s8, s12
+; GFX9-NEXT:    s_mul_i32 s8, s12, s8
+; GFX9-NEXT:    s_add_u32 s8, s10, s8
+; GFX9-NEXT:    s_addc_u32 s10, 0, s9
+; GFX9-NEXT:    s_add_u32 s11, s14, s8
+; GFX9-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[8:9], 0
+; GFX9-NEXT:    s_addc_u32 s8, s12, s10
+; GFX9-NEXT:    s_mul_i32 s10, s2, s8
+; GFX9-NEXT:    s_mul_hi_u32 s12, s2, s11
+; GFX9-NEXT:    s_mul_hi_u32 s9, s2, s8
+; GFX9-NEXT:    s_add_u32 s10, s12, s10
 ; GFX9-NEXT:    s_addc_u32 s9, 0, s9
-; GFX9-NEXT:    s_add_u32 s8, s11, s8
-; GFX9-NEXT:    s_addc_u32 s9, s10, s9
-; GFX9-NEXT:    s_mul_i32 s11, s2, s9
-; GFX9-NEXT:    s_mul_hi_u32 s12, s2, s8
-; GFX9-NEXT:    s_mul_hi_u32 s10, s2, s9
-; GFX9-NEXT:    s_add_u32 s11, s12, s11
-; GFX9-NEXT:    s_addc_u32 s10, 0, s10
-; GFX9-NEXT:    s_mul_hi_u32 s13, s3, s8
-; GFX9-NEXT:    s_mul_i32 s8, s3, s8
-; GFX9-NEXT:    s_add_u32 s8, s11, s8
-; GFX9-NEXT:    s_mul_hi_u32 s12, s3, s9
-; GFX9-NEXT:    s_addc_u32 s8, s10, s13
+; GFX9-NEXT:    s_mul_hi_u32 s13, s3, s11
+; GFX9-NEXT:    s_mul_i32 s11, s3, s11
+; GFX9-NEXT:    s_add_u32 s10, s10, s11
+; GFX9-NEXT:    s_mul_hi_u32 s12, s3, s8
+; GFX9-NEXT:    s_addc_u32 s9, s9, s13
 ; GFX9-NEXT:    s_addc_u32 s10, s12, 0
-; GFX9-NEXT:    s_mul_i32 s9, s3, s9
-; GFX9-NEXT:    s_add_u32 s11, s8, s9
-; GFX9-NEXT:    s_addc_u32 s10, 0, s10
-; GFX9-NEXT:    s_mul_i32 s8, s6, s10
-; GFX9-NEXT:    s_mul_hi_u32 s9, s6, s11
+; GFX9-NEXT:    s_mul_i32 s8, s3, s8
+; GFX9-NEXT:    s_add_u32 s12, s9, s8
+; GFX9-NEXT:    s_addc_u32 s13, 0, s10
+; GFX9-NEXT:    s_mul_i32 s8, s6, s13
+; GFX9-NEXT:    s_mul_hi_u32 s9, s6, s12
 ; GFX9-NEXT:    s_add_i32 s8, s9, s8
-; GFX9-NEXT:    s_mul_i32 s9, s7, s11
-; GFX9-NEXT:    s_add_i32 s12, s8, s9
-; GFX9-NEXT:    s_sub_i32 s13, s3, s12
-; GFX9-NEXT:    s_mul_i32 s8, s6, s11
-; GFX9-NEXT:    s_sub_u32 s14, s2, s8
+; GFX9-NEXT:    s_mul_i32 s9, s7, s12
+; GFX9-NEXT:    s_add_i32 s14, s8, s9
+; GFX9-NEXT:    s_sub_i32 s10, s3, s14
+; GFX9-NEXT:    s_mul_i32 s8, s6, s12
+; GFX9-NEXT:    s_sub_u32 s15, s2, s8
 ; GFX9-NEXT:    s_cselect_b64 s[8:9], -1, 0
-; GFX9-NEXT:    s_subb_u32 s13, s13, s7
-; GFX9-NEXT:    s_sub_u32 s15, s14, s6
-; GFX9-NEXT:    s_subb_u32 s13, s13, 0
-; GFX9-NEXT:    s_cmp_ge_u32 s13, s7
+; GFX9-NEXT:    s_cmp_lg_u64 s[8:9], 0
+; GFX9-NEXT:    s_subb_u32 s16, s10, s7
+; GFX9-NEXT:    s_sub_u32 s17, s15, s6
+; GFX9-NEXT:    s_cselect_b64 s[10:11], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[10:11], 0
+; GFX9-NEXT:    s_subb_u32 s10, s16, 0
+; GFX9-NEXT:    s_cmp_ge_u32 s10, s7
+; GFX9-NEXT:    s_cselect_b32 s11, -1, 0
+; GFX9-NEXT:    s_cmp_ge_u32 s17, s6
 ; GFX9-NEXT:    s_cselect_b32 s16, -1, 0
-; GFX9-NEXT:    s_cmp_ge_u32 s15, s6
-; GFX9-NEXT:    s_cselect_b32 s15, -1, 0
-; GFX9-NEXT:    s_cmp_eq_u32 s13, s7
-; GFX9-NEXT:    s_cselect_b32 s13, s15, s16
-; GFX9-NEXT:    s_add_u32 s15, s11, 1
-; GFX9-NEXT:    s_addc_u32 s16, s10, 0
-; GFX9-NEXT:    s_add_u32 s17, s11, 2
-; GFX9-NEXT:    s_addc_u32 s18, s10, 0
-; GFX9-NEXT:    s_cmp_lg_u32 s13, 0
-; GFX9-NEXT:    s_cselect_b32 s13, s17, s15
-; GFX9-NEXT:    s_cselect_b32 s15, s18, s16
+; GFX9-NEXT:    s_cmp_eq_u32 s10, s7
+; GFX9-NEXT:    s_cselect_b32 s10, s16, s11
+; GFX9-NEXT:    s_add_u32 s11, s12, 1
+; GFX9-NEXT:    s_addc_u32 s16, s13, 0
+; GFX9-NEXT:    s_add_u32 s17, s12, 2
+; GFX9-NEXT:    s_addc_u32 s18, s13, 0
+; GFX9-NEXT:    s_cmp_lg_u32 s10, 0
+; GFX9-NEXT:    s_cselect_b32 s10, s17, s11
+; GFX9-NEXT:    s_cselect_b32 s11, s18, s16
 ; GFX9-NEXT:    s_cmp_lg_u64 s[8:9], 0
-; GFX9-NEXT:    s_subb_u32 s3, s3, s12
+; GFX9-NEXT:    s_subb_u32 s3, s3, s14
 ; GFX9-NEXT:    s_cmp_ge_u32 s3, s7
 ; GFX9-NEXT:    s_cselect_b32 s8, -1, 0
-; GFX9-NEXT:    s_cmp_ge_u32 s14, s6
+; GFX9-NEXT:    s_cmp_ge_u32 s15, s6
 ; GFX9-NEXT:    s_cselect_b32 s9, -1, 0
 ; GFX9-NEXT:    s_cmp_eq_u32 s3, s7
 ; GFX9-NEXT:    s_cselect_b32 s3, s9, s8
 ; GFX9-NEXT:    s_cmp_lg_u32 s3, 0
-; GFX9-NEXT:    s_cselect_b32 s9, s15, s10
-; GFX9-NEXT:    s_cselect_b32 s8, s13, s11
+; GFX9-NEXT:    s_cselect_b32 s9, s11, s13
+; GFX9-NEXT:    s_cselect_b32 s8, s10, s12
 ; GFX9-NEXT:    s_cbranch_execnz .LBB16_3
 ; GFX9-NEXT:  .LBB16_2:
 ; GFX9-NEXT:    v_cvt_f32_u32_e32 v0, s6
@@ -2463,40 +2503,44 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX1010-NEXT:    s_add_u32 s11, s12, s11
 ; GFX1010-NEXT:    s_addc_u32 s12, 0, s13
 ; GFX1010-NEXT:    s_add_u32 s8, s8, s11
+; GFX1010-NEXT:    s_cselect_b32 s11, -1, 0
+; GFX1010-NEXT:    s_mul_hi_u32 s13, s9, s8
+; GFX1010-NEXT:    s_cmp_lg_u32 s11, 0
+; GFX1010-NEXT:    s_mul_i32 s11, s9, s8
 ; GFX1010-NEXT:    s_addc_u32 s5, s5, s12
-; GFX1010-NEXT:    s_mul_hi_u32 s11, s9, s8
-; GFX1010-NEXT:    s_mul_i32 s12, s9, s8
-; GFX1010-NEXT:    s_mul_i32 s9, s9, s5
 ; GFX1010-NEXT:    s_mul_i32 s10, s10, s8
-; GFX1010-NEXT:    s_add_i32 s9, s11, s9
-; GFX1010-NEXT:    s_mul_i32 s11, s5, s12
+; GFX1010-NEXT:    s_mul_i32 s9, s9, s5
+; GFX1010-NEXT:    s_mul_hi_u32 s12, s8, s11
+; GFX1010-NEXT:    s_add_i32 s9, s13, s9
+; GFX1010-NEXT:    s_mul_hi_u32 s13, s5, s11
 ; GFX1010-NEXT:    s_add_i32 s9, s9, s10
-; GFX1010-NEXT:    s_mul_hi_u32 s10, s8, s12
+; GFX1010-NEXT:    s_mul_i32 s10, s5, s11
 ; GFX1010-NEXT:    s_mul_i32 s15, s8, s9
 ; GFX1010-NEXT:    s_mul_hi_u32 s14, s8, s9
-; GFX1010-NEXT:    s_add_u32 s10, s10, s15
-; GFX1010-NEXT:    s_mul_hi_u32 s13, s5, s12
+; GFX1010-NEXT:    s_add_u32 s12, s12, s15
 ; GFX1010-NEXT:    s_addc_u32 s14, 0, s14
-; GFX1010-NEXT:    s_mul_hi_u32 s12, s5, s9
-; GFX1010-NEXT:    s_add_u32 s10, s10, s11
+; GFX1010-NEXT:    s_mul_hi_u32 s11, s5, s9
+; GFX1010-NEXT:    s_add_u32 s10, s12, s10
 ; GFX1010-NEXT:    s_mul_i32 s9, s5, s9
 ; GFX1010-NEXT:    s_addc_u32 s10, s14, s13
-; GFX1010-NEXT:    s_addc_u32 s11, s12, 0
+; GFX1010-NEXT:    s_addc_u32 s11, s11, 0
 ; GFX1010-NEXT:    s_add_u32 s9, s10, s9
 ; GFX1010-NEXT:    s_addc_u32 s10, 0, s11
 ; GFX1010-NEXT:    s_add_u32 s8, s8, s9
+; GFX1010-NEXT:    s_cselect_b32 s9, -1, 0
+; GFX1010-NEXT:    s_mul_hi_u32 s11, s2, s8
+; GFX1010-NEXT:    s_cmp_lg_u32 s9, 0
+; GFX1010-NEXT:    s_mul_hi_u32 s9, s3, s8
 ; GFX1010-NEXT:    s_addc_u32 s5, s5, s10
-; GFX1010-NEXT:    s_mul_hi_u32 s9, s2, s8
-; GFX1010-NEXT:    s_mul_i32 s12, s2, s5
-; GFX1010-NEXT:    s_mul_hi_u32 s11, s2, s5
-; GFX1010-NEXT:    s_mul_hi_u32 s10, s3, s8
 ; GFX1010-NEXT:    s_mul_i32 s8, s3, s8
-; GFX1010-NEXT:    s_add_u32 s9, s9, s12
-; GFX1010-NEXT:    s_addc_u32 s11, 0, s11
+; GFX1010-NEXT:    s_mul_i32 s12, s2, s5
+; GFX1010-NEXT:    s_mul_hi_u32 s10, s2, s5
+; GFX1010-NEXT:    s_add_u32 s11, s11, s12
+; GFX1010-NEXT:    s_addc_u32 s10, 0, s10
 ; GFX1010-NEXT:    s_mul_hi_u32 s13, s3, s5
-; GFX1010-NEXT:    s_add_u32 s8, s9, s8
+; GFX1010-NEXT:    s_add_u32 s8, s11, s8
 ; GFX1010-NEXT:    s_mul_i32 s5, s3, s5
-; GFX1010-NEXT:    s_addc_u32 s8, s11, s10
+; GFX1010-NEXT:    s_addc_u32 s8, s10, s9
 ; GFX1010-NEXT:    s_addc_u32 s9, s13, 0
 ; GFX1010-NEXT:    s_add_u32 s5, s8, s5
 ; GFX1010-NEXT:    s_addc_u32 s8, 0, s9
@@ -2509,8 +2553,11 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX1010-NEXT:    s_sub_i32 s11, s3, s9
 ; GFX1010-NEXT:    s_sub_u32 s10, s2, s10
 ; GFX1010-NEXT:    s_cselect_b32 s12, -1, 0
+; GFX1010-NEXT:    s_cmp_lg_u32 s12, 0
 ; GFX1010-NEXT:    s_subb_u32 s11, s11, s7
 ; GFX1010-NEXT:    s_sub_u32 s13, s10, s6
+; GFX1010-NEXT:    s_cselect_b32 s14, -1, 0
+; GFX1010-NEXT:    s_cmp_lg_u32 s14, 0
 ; GFX1010-NEXT:    s_subb_u32 s11, s11, 0
 ; GFX1010-NEXT:    s_cmp_ge_u32 s11, s7
 ; GFX1010-NEXT:    s_cselect_b32 s14, -1, 0
@@ -2616,40 +2663,44 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX1030W32-NEXT:    s_add_u32 s11, s12, s11
 ; GFX1030W32-NEXT:    s_addc_u32 s12, 0, s13
 ; GFX1030W32-NEXT:    s_add_u32 s8, s8, s11
+; GFX1030W32-NEXT:    s_cselect_b32 s11, -1, 0
+; GFX1030W32-NEXT:    s_mul_hi_u32 s13, s9, s8
+; GFX1030W32-NEXT:    s_cmp_lg_u32 s11, 0
+; GFX1030W32-NEXT:    s_mul_i32 s11, s9, s8
 ; GFX1030W32-NEXT:    s_addc_u32 s7, s7, s12
-; GFX1030W32-NEXT:    s_mul_hi_u32 s11, s9, s8
-; GFX1030W32-NEXT:    s_mul_i32 s12, s9, s8
-; GFX1030W32-NEXT:    s_mul_i32 s9, s9, s7
 ; GFX1030W32-NEXT:    s_mul_i32 s10, s10, s8
-; GFX1030W32-NEXT:    s_add_i32 s9, s11, s9
-; GFX1030W32-NEXT:    s_mul_i32 s11, s7, s12
+; GFX1030W32-NEXT:    s_mul_i32 s9, s9, s7
+; GFX1030W32-NEXT:    s_mul_hi_u32 s12, s8, s11
+; GFX1030W32-NEXT:    s_add_i32 s9, s13, s9
+; GFX1030W32-NEXT:    s_mul_hi_u32 s13, s7, s11
 ; GFX1030W32-NEXT:    s_add_i32 s9, s9, s10
-; GFX1030W32-NEXT:    s_mul_hi_u32 s10, s8, s12
+; GFX1030W32-NEXT:    s_mul_i32 s10, s7, s11
 ; GFX1030W32-NEXT:    s_mul_i32 s15, s8, s9
 ; GFX1030W32-NEXT:    s_mul_hi_u32 s14, s8, s9
-; GFX1030W32-NEXT:    s_add_u32 s10, s10, s15
-; GFX1030W32-NEXT:    s_mul_hi_u32 s13, s7, s12
+; GFX1030W32-NEXT:    s_add_u32 s12, s12, s15
 ; GFX1030W32-NEXT:    s_addc_u32 s14, 0, s14
-; GFX1030W32-NEXT:    s_mul_hi_u32 s12, s7, s9
-; GFX1030W32-NEXT:    s_add_u32 s10, s10, s11
+; GFX1030W32-NEXT:    s_mul_hi_u32 s11, s7, s9
+; GFX1030W32-NEXT:    s_add_u32 s10, s12, s10
 ; GFX1030W32-NEXT:    s_mul_i32 s9, s7, s9
 ; GFX1030W32-NEXT:    s_addc_u32 s10, s14, s13
-; GFX1030W32-NEXT:    s_addc_u32 s11, s12, 0
+; GFX1030W32-NEXT:    s_addc_u32 s11, s11, 0
 ; GFX1030W32-NEXT:    s_add_u32 s9, s10, s9
 ; GFX1030W32-NEXT:    s_addc_u32 s10, 0, s11
 ; GFX1030W32-NEXT:    s_add_u32 s8, s8, s9
+; GFX1030W32-NEXT:    s_cselect_b32 s9, -1, 0
+; GFX1030W32-NEXT:    s_mul_hi_u32 s11, s2, s8
+; GFX1030W32-NEXT:    s_cmp_lg_u32 s9, 0
+; GFX1030W32-NEXT:    s_mul_hi_u32 s9, s3, s8
 ; GFX1030W32-NEXT:    s_addc_u32 s7, s7, s10
-; GFX1030W32-NEXT:    s_mul_hi_u32 s9, s2, s8
-; GFX1030W32-NEXT:    s_mul_i32 s12, s2, s7
-; GFX1030W32-NEXT:    s_mul_hi_u32 s11, s2, s7
-; GFX1030W32-NEXT:    s_mul_hi_u32 s10, s3, s8
 ; GFX1030W32-NEXT:    s_mul_i32 s8, s3, s8
-; GFX1030W32-NEXT:    s_add_u32 s9, s9, s12
-; GFX1030W32-NEXT:    s_addc_u32 s11, 0, s11
+; GFX1030W32-NEXT:    s_mul_i32 s12, s2, s7
+; GFX1030W32-NEXT:    s_mul_hi_u32 s10, s2, s7
+; GFX1030W32-NEXT:    s_add_u32 s11, s11, s12
+; GFX1030W32-NEXT:    s_addc_u32 s10, 0, s10
 ; GFX1030W32-NEXT:    s_mul_hi_u32 s13, s3, s7
-; GFX1030W32-NEXT:    s_add_u32 s8, s9, s8
+; GFX1030W32-NEXT:    s_add_u32 s8, s11, s8
 ; GFX1030W32-NEXT:    s_mul_i32 s7, s3, s7
-; GFX1030W32-NEXT:    s_addc_u32 s8, s11, s10
+; GFX1030W32-NEXT:    s_addc_u32 s8, s10, s9
 ; GFX1030W32-NEXT:    s_addc_u32 s9, s13, 0
 ; GFX1030W32-NEXT:    s_add_u32 s7, s8, s7
 ; GFX1030W32-NEXT:    s_addc_u32 s8, 0, s9
@@ -2662,8 +2713,11 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX1030W32-NEXT:    s_sub_i32 s11, s3, s9
 ; GFX1030W32-NEXT:    s_sub_u32 s10, s2, s10
 ; GFX1030W32-NEXT:    s_cselect_b32 s12, -1, 0
+; GFX1030W32-NEXT:    s_cmp_lg_u32 s12, 0
 ; GFX1030W32-NEXT:    s_subb_u32 s11, s11, s5
 ; GFX1030W32-NEXT:    s_sub_u32 s13, s10, s4
+; GFX1030W32-NEXT:    s_cselect_b32 s14, -1, 0
+; GFX1030W32-NEXT:    s_cmp_lg_u32 s14, 0
 ; GFX1030W32-NEXT:    s_subb_u32 s11, s11, 0
 ; GFX1030W32-NEXT:    s_cmp_ge_u32 s11, s5
 ; GFX1030W32-NEXT:    s_cselect_b32 s14, -1, 0
@@ -2736,8 +2790,8 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX1030W64-NEXT:  ; %bb.1:
 ; GFX1030W64-NEXT:    v_cvt_f32_u32_e32 v0, s4
 ; GFX1030W64-NEXT:    v_cvt_f32_u32_e32 v1, s5
-; GFX1030W64-NEXT:    s_sub_u32 s8, 0, s4
-; GFX1030W64-NEXT:    s_subb_u32 s9, 0, s5
+; GFX1030W64-NEXT:    s_sub_u32 s9, 0, s4
+; GFX1030W64-NEXT:    s_subb_u32 s10, 0, s5
 ; GFX1030W64-NEXT:    v_fmamk_f32 v0, v1, 0x4f800000, v0
 ; GFX1030W64-NEXT:    v_rcp_f32_e32 v0, v0
 ; GFX1030W64-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -2746,102 +2800,109 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX1030W64-NEXT:    v_fmamk_f32 v0, v1, 0xcf800000, v0
 ; GFX1030W64-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GFX1030W64-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GFX1030W64-NEXT:    v_readfirstlane_b32 s6, v1
-; GFX1030W64-NEXT:    v_readfirstlane_b32 s7, v0
-; GFX1030W64-NEXT:    s_mul_i32 s10, s8, s6
-; GFX1030W64-NEXT:    s_mul_hi_u32 s12, s8, s7
-; GFX1030W64-NEXT:    s_mul_i32 s11, s9, s7
-; GFX1030W64-NEXT:    s_add_i32 s10, s12, s10
-; GFX1030W64-NEXT:    s_mul_i32 s13, s8, s7
-; GFX1030W64-NEXT:    s_add_i32 s10, s10, s11
-; GFX1030W64-NEXT:    s_mul_hi_u32 s12, s7, s13
-; GFX1030W64-NEXT:    s_mul_i32 s15, s7, s10
-; GFX1030W64-NEXT:    s_mul_hi_u32 s14, s6, s13
-; GFX1030W64-NEXT:    s_mul_i32 s11, s6, s13
-; GFX1030W64-NEXT:    s_mul_hi_u32 s13, s7, s10
+; GFX1030W64-NEXT:    v_readfirstlane_b32 s8, v1
+; GFX1030W64-NEXT:    v_readfirstlane_b32 s6, v0
+; GFX1030W64-NEXT:    s_mul_i32 s7, s9, s8
+; GFX1030W64-NEXT:    s_mul_hi_u32 s12, s9, s6
+; GFX1030W64-NEXT:    s_mul_i32 s11, s10, s6
+; GFX1030W64-NEXT:    s_add_i32 s7, s12, s7
+; GFX1030W64-NEXT:    s_mul_i32 s13, s9, s6
+; GFX1030W64-NEXT:    s_add_i32 s7, s7, s11
+; GFX1030W64-NEXT:    s_mul_hi_u32 s12, s6, s13
+; GFX1030W64-NEXT:    s_mul_i32 s15, s6, s7
+; GFX1030W64-NEXT:    s_mul_hi_u32 s14, s8, s13
+; GFX1030W64-NEXT:    s_mul_i32 s11, s8, s13
+; GFX1030W64-NEXT:    s_mul_hi_u32 s13, s6, s7
 ; GFX1030W64-NEXT:    s_add_u32 s12, s12, s15
 ; GFX1030W64-NEXT:    s_addc_u32 s13, 0, s13
-; GFX1030W64-NEXT:    s_mul_hi_u32 s16, s6, s10
+; GFX1030W64-NEXT:    s_mul_hi_u32 s16, s8, s7
 ; GFX1030W64-NEXT:    s_add_u32 s11, s12, s11
-; GFX1030W64-NEXT:    s_mul_i32 s10, s6, s10
+; GFX1030W64-NEXT:    s_mul_i32 s7, s8, s7
 ; GFX1030W64-NEXT:    s_addc_u32 s11, s13, s14
 ; GFX1030W64-NEXT:    s_addc_u32 s12, s16, 0
-; GFX1030W64-NEXT:    s_add_u32 s10, s11, s10
+; GFX1030W64-NEXT:    s_add_u32 s7, s11, s7
 ; GFX1030W64-NEXT:    s_addc_u32 s11, 0, s12
-; GFX1030W64-NEXT:    s_add_u32 s7, s7, s10
-; GFX1030W64-NEXT:    s_addc_u32 s6, s6, s11
-; GFX1030W64-NEXT:    s_mul_hi_u32 s10, s8, s7
-; GFX1030W64-NEXT:    s_mul_i32 s11, s8, s7
-; GFX1030W64-NEXT:    s_mul_i32 s8, s8, s6
-; GFX1030W64-NEXT:    s_mul_i32 s9, s9, s7
-; GFX1030W64-NEXT:    s_add_i32 s8, s10, s8
-; GFX1030W64-NEXT:    s_mul_i32 s10, s6, s11
-; GFX1030W64-NEXT:    s_add_i32 s8, s8, s9
-; GFX1030W64-NEXT:    s_mul_hi_u32 s9, s7, s11
-; GFX1030W64-NEXT:    s_mul_i32 s14, s7, s8
-; GFX1030W64-NEXT:    s_mul_hi_u32 s13, s7, s8
-; GFX1030W64-NEXT:    s_add_u32 s9, s9, s14
-; GFX1030W64-NEXT:    s_mul_hi_u32 s12, s6, s11
+; GFX1030W64-NEXT:    s_add_u32 s12, s6, s7
+; GFX1030W64-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; GFX1030W64-NEXT:    s_mul_hi_u32 s13, s9, s12
+; GFX1030W64-NEXT:    s_cmp_lg_u64 s[6:7], 0
+; GFX1030W64-NEXT:    s_mul_i32 s6, s9, s12
+; GFX1030W64-NEXT:    s_addc_u32 s8, s8, s11
+; GFX1030W64-NEXT:    s_mul_i32 s10, s10, s12
+; GFX1030W64-NEXT:    s_mul_i32 s9, s9, s8
+; GFX1030W64-NEXT:    s_mul_hi_u32 s7, s12, s6
+; GFX1030W64-NEXT:    s_add_i32 s9, s13, s9
+; GFX1030W64-NEXT:    s_mul_hi_u32 s11, s8, s6
+; GFX1030W64-NEXT:    s_add_i32 s9, s9, s10
+; GFX1030W64-NEXT:    s_mul_i32 s6, s8, s6
+; GFX1030W64-NEXT:    s_mul_i32 s14, s12, s9
+; GFX1030W64-NEXT:    s_mul_hi_u32 s13, s12, s9
+; GFX1030W64-NEXT:    s_add_u32 s7, s7, s14
 ; GFX1030W64-NEXT:    s_addc_u32 s13, 0, s13
-; GFX1030W64-NEXT:    s_mul_hi_u32 s11, s6, s8
-; GFX1030W64-NEXT:    s_add_u32 s9, s9, s10
-; GFX1030W64-NEXT:    s_mul_i32 s8, s6, s8
-; GFX1030W64-NEXT:    s_addc_u32 s9, s13, s12
-; GFX1030W64-NEXT:    s_addc_u32 s10, s11, 0
-; GFX1030W64-NEXT:    s_add_u32 s8, s9, s8
-; GFX1030W64-NEXT:    s_addc_u32 s9, 0, s10
-; GFX1030W64-NEXT:    s_add_u32 s7, s7, s8
-; GFX1030W64-NEXT:    s_addc_u32 s6, s6, s9
-; GFX1030W64-NEXT:    s_mul_hi_u32 s8, s2, s7
-; GFX1030W64-NEXT:    s_mul_i32 s11, s2, s6
-; GFX1030W64-NEXT:    s_mul_hi_u32 s10, s2, s6
-; GFX1030W64-NEXT:    s_mul_hi_u32 s9, s3, s7
+; GFX1030W64-NEXT:    s_mul_hi_u32 s10, s8, s9
+; GFX1030W64-NEXT:    s_add_u32 s6, s7, s6
+; GFX1030W64-NEXT:    s_mul_i32 s9, s8, s9
+; GFX1030W64-NEXT:    s_addc_u32 s6, s13, s11
+; GFX1030W64-NEXT:    s_addc_u32 s7, s10, 0
+; GFX1030W64-NEXT:    s_add_u32 s6, s6, s9
+; GFX1030W64-NEXT:    s_addc_u32 s9, 0, s7
+; GFX1030W64-NEXT:    s_add_u32 s10, s12, s6
+; GFX1030W64-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; GFX1030W64-NEXT:    s_mul_hi_u32 s11, s2, s10
+; GFX1030W64-NEXT:    s_cmp_lg_u64 s[6:7], 0
+; GFX1030W64-NEXT:    s_mul_hi_u32 s6, s3, s10
+; GFX1030W64-NEXT:    s_addc_u32 s7, s8, s9
+; GFX1030W64-NEXT:    s_mul_i32 s8, s3, s10
+; GFX1030W64-NEXT:    s_mul_i32 s10, s2, s7
+; GFX1030W64-NEXT:    s_mul_hi_u32 s9, s2, s7
+; GFX1030W64-NEXT:    s_add_u32 s10, s11, s10
+; GFX1030W64-NEXT:    s_addc_u32 s9, 0, s9
+; GFX1030W64-NEXT:    s_mul_hi_u32 s12, s3, s7
+; GFX1030W64-NEXT:    s_add_u32 s8, s10, s8
 ; GFX1030W64-NEXT:    s_mul_i32 s7, s3, s7
-; GFX1030W64-NEXT:    s_add_u32 s8, s8, s11
-; GFX1030W64-NEXT:    s_addc_u32 s10, 0, s10
-; GFX1030W64-NEXT:    s_mul_hi_u32 s12, s3, s6
-; GFX1030W64-NEXT:    s_add_u32 s7, s8, s7
-; GFX1030W64-NEXT:    s_mul_i32 s6, s3, s6
-; GFX1030W64-NEXT:    s_addc_u32 s7, s10, s9
+; GFX1030W64-NEXT:    s_addc_u32 s6, s9, s6
 ; GFX1030W64-NEXT:    s_addc_u32 s8, s12, 0
-; GFX1030W64-NEXT:    s_add_u32 s10, s7, s6
+; GFX1030W64-NEXT:    s_add_u32 s10, s6, s7
 ; GFX1030W64-NEXT:    s_addc_u32 s11, 0, s8
 ; GFX1030W64-NEXT:    s_mul_hi_u32 s6, s4, s10
 ; GFX1030W64-NEXT:    s_mul_i32 s7, s4, s11
 ; GFX1030W64-NEXT:    s_mul_i32 s8, s5, s10
 ; GFX1030W64-NEXT:    s_add_i32 s6, s6, s7
-; GFX1030W64-NEXT:    s_add_i32 s8, s6, s8
+; GFX1030W64-NEXT:    s_add_i32 s12, s6, s8
 ; GFX1030W64-NEXT:    s_mul_i32 s6, s4, s10
-; GFX1030W64-NEXT:    s_sub_i32 s9, s3, s8
-; GFX1030W64-NEXT:    s_sub_u32 s12, s2, s6
+; GFX1030W64-NEXT:    s_sub_i32 s8, s3, s12
+; GFX1030W64-NEXT:    s_sub_u32 s13, s2, s6
 ; GFX1030W64-NEXT:    s_cselect_b64 s[6:7], -1, 0
-; GFX1030W64-NEXT:    s_subb_u32 s9, s9, s5
-; GFX1030W64-NEXT:    s_sub_u32 s13, s12, s4
-; GFX1030W64-NEXT:    s_subb_u32 s9, s9, 0
-; GFX1030W64-NEXT:    s_cmp_ge_u32 s9, s5
+; GFX1030W64-NEXT:    s_cmp_lg_u64 s[6:7], 0
+; GFX1030W64-NEXT:    s_subb_u32 s14, s8, s5
+; GFX1030W64-NEXT:    s_sub_u32 s15, s13, s4
+; GFX1030W64-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; GFX1030W64-NEXT:    s_cmp_lg_u64 s[8:9], 0
+; GFX1030W64-NEXT:    s_subb_u32 s8, s14, 0
+; GFX1030W64-NEXT:    s_cmp_ge_u32 s8, s5
+; GFX1030W64-NEXT:    s_cselect_b32 s9, -1, 0
+; GFX1030W64-NEXT:    s_cmp_ge_u32 s15, s4
 ; GFX1030W64-NEXT:    s_cselect_b32 s14, -1, 0
-; GFX1030W64-NEXT:    s_cmp_ge_u32 s13, s4
-; GFX1030W64-NEXT:    s_cselect_b32 s13, -1, 0
-; GFX1030W64-NEXT:    s_cmp_eq_u32 s9, s5
-; GFX1030W64-NEXT:    s_cselect_b32 s9, s13, s14
-; GFX1030W64-NEXT:    s_add_u32 s13, s10, 1
+; GFX1030W64-NEXT:    s_cmp_eq_u32 s8, s5
+; GFX1030W64-NEXT:    s_cselect_b32 s8, s14, s9
+; GFX1030W64-NEXT:    s_add_u32 s9, s10, 1
 ; GFX1030W64-NEXT:    s_addc_u32 s14, s11, 0
 ; GFX1030W64-NEXT:    s_add_u32 s15, s10, 2
 ; GFX1030W64-NEXT:    s_addc_u32 s16, s11, 0
-; GFX1030W64-NEXT:    s_cmp_lg_u32 s9, 0
-; GFX1030W64-NEXT:    s_cselect_b32 s13, s15, s13
+; GFX1030W64-NEXT:    s_cmp_lg_u32 s8, 0
+; GFX1030W64-NEXT:    s_cselect_b32 s15, s15, s9
 ; GFX1030W64-NEXT:    s_cselect_b32 s14, s16, s14
 ; GFX1030W64-NEXT:    s_cmp_lg_u64 s[6:7], 0
-; GFX1030W64-NEXT:    s_subb_u32 s3, s3, s8
+; GFX1030W64-NEXT:    s_subb_u32 s3, s3, s12
 ; GFX1030W64-NEXT:    s_cmp_ge_u32 s3, s5
 ; GFX1030W64-NEXT:    s_cselect_b32 s6, -1, 0
-; GFX1030W64-NEXT:    s_cmp_ge_u32 s12, s4
+; GFX1030W64-NEXT:    s_cmp_ge_u32 s13, s4
 ; GFX1030W64-NEXT:    s_cselect_b32 s7, -1, 0
 ; GFX1030W64-NEXT:    s_cmp_eq_u32 s3, s5
 ; GFX1030W64-NEXT:    s_cselect_b32 s3, s7, s6
 ; GFX1030W64-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX1030W64-NEXT:    s_cselect_b32 s7, s14, s11
-; GFX1030W64-NEXT:    s_cselect_b32 s6, s13, s10
+; GFX1030W64-NEXT:    s_cselect_b32 s6, s15, s10
 ; GFX1030W64-NEXT:    s_cbranch_execnz .LBB16_3
 ; GFX1030W64-NEXT:  .LBB16_2:
 ; GFX1030W64-NEXT:    v_cvt_f32_u32_e32 v0, s4
@@ -2927,40 +2988,44 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX11-NEXT:    s_add_u32 s11, s12, s11
 ; GFX11-NEXT:    s_addc_u32 s12, 0, s13
 ; GFX11-NEXT:    s_add_u32 s8, s8, s11
+; GFX11-NEXT:    s_cselect_b32 s11, -1, 0
+; GFX11-NEXT:    s_mul_hi_u32 s13, s9, s8
+; GFX11-NEXT:    s_cmp_lg_u32 s11, 0
+; GFX11-NEXT:    s_mul_i32 s11, s9, s8
 ; GFX11-NEXT:    s_addc_u32 s7, s7, s12
-; GFX11-NEXT:    s_mul_hi_u32 s11, s9, s8
-; GFX11-NEXT:    s_mul_i32 s12, s9, s8
-; GFX11-NEXT:    s_mul_i32 s9, s9, s7
 ; GFX11-NEXT:    s_mul_i32 s10, s10, s8
-; GFX11-NEXT:    s_add_i32 s9, s11, s9
-; GFX11-NEXT:    s_mul_i32 s11, s7, s12
+; GFX11-NEXT:    s_mul_i32 s9, s9, s7
+; GFX11-NEXT:    s_mul_hi_u32 s12, s8, s11
+; GFX11-NEXT:    s_add_i32 s9, s13, s9
+; GFX11-NEXT:    s_mul_hi_u32 s13, s7, s11
 ; GFX11-NEXT:    s_add_i32 s9, s9, s10
-; GFX11-NEXT:    s_mul_hi_u32 s10, s8, s12
+; GFX11-NEXT:    s_mul_i32 s10, s7, s11
 ; GFX11-NEXT:    s_mul_i32 s15, s8, s9
 ; GFX11-NEXT:    s_mul_hi_u32 s14, s8, s9
-; GFX11-NEXT:    s_add_u32 s10, s10, s15
-; GFX11-NEXT:    s_mul_hi_u32 s13, s7, s12
+; GFX11-NEXT:    s_add_u32 s12, s12, s15
 ; GFX11-NEXT:    s_addc_u32 s14, 0, s14
-; GFX11-NEXT:    s_mul_hi_u32 s12, s7, s9
-; GFX11-NEXT:    s_add_u32 s10, s10, s11
+; GFX11-NEXT:    s_mul_hi_u32 s11, s7, s9
+; GFX11-NEXT:    s_add_u32 s10, s12, s10
 ; GFX11-NEXT:    s_mul_i32 s9, s7, s9
 ; GFX11-NEXT:    s_addc_u32 s10, s14, s13
-; GFX11-NEXT:    s_addc_u32 s11, s12, 0
+; GFX11-NEXT:    s_addc_u32 s11, s11, 0
 ; GFX11-NEXT:    s_add_u32 s9, s10, s9
 ; GFX11-NEXT:    s_addc_u32 s10, 0, s11
 ; GFX11-NEXT:    s_add_u32 s8, s8, s9
+; GFX11-NEXT:    s_cselect_b32 s9, -1, 0
+; GFX11-NEXT:    s_mul_hi_u32 s11, s2, s8
+; GFX11-NEXT:    s_cmp_lg_u32 s9, 0
+; GFX11-NEXT:    s_mul_hi_u32 s9, s3, s8
 ; GFX11-NEXT:    s_addc_u32 s7, s7, s10
-; GFX11-NEXT:    s_mul_hi_u32 s9, s2, s8
-; GFX11-NEXT:    s_mul_i32 s12, s2, s7
-; GFX11-NEXT:    s_mul_hi_u32 s11, s2, s7
-; GFX11-NEXT:    s_mul_hi_u32 s10, s3, s8
 ; GFX11-NEXT:    s_mul_i32 s8, s3, s8
-; GFX11-NEXT:    s_add_u32 s9, s9, s12
-; GFX11-NEXT:    s_addc_u32 s11, 0, s11
+; GFX11-NEXT:    s_mul_i32 s12, s2, s7
+; GFX11-NEXT:    s_mul_hi_u32 s10, s2, s7
+; GFX11-NEXT:    s_add_u32 s11, s11, s12
+; GFX11-NEXT:    s_addc_u32 s10, 0, s10
 ; GFX11-NEXT:    s_mul_hi_u32 s13, s3, s7
-; GFX11-NEXT:    s_add_u32 s8, s9, s8
+; GFX11-NEXT:    s_add_u32 s8, s11, s8
 ; GFX11-NEXT:    s_mul_i32 s7, s3, s7
-; GFX11-NEXT:    s_addc_u32 s8, s11, s10
+; GFX11-NEXT:    s_addc_u32 s8, s10, s9
 ; GFX11-NEXT:    s_addc_u32 s9, s13, 0
 ; GFX11-NEXT:    s_add_u32 s7, s8, s7
 ; GFX11-NEXT:    s_addc_u32 s8, 0, s9
@@ -2970,14 +3035,17 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX11-NEXT:    s_add_i32 s9, s9, s10
 ; GFX11-NEXT:    s_mul_i32 s10, s4, s7
 ; GFX11-NEXT:    s_add_i32 s9, s9, s11
-; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
 ; GFX11-NEXT:    s_sub_i32 s11, s3, s9
 ; GFX11-NEXT:    s_sub_u32 s10, s2, s10
 ; GFX11-NEXT:    s_cselect_b32 s12, -1, 0
+; GFX11-NEXT:    s_cmp_lg_u32 s12, 0
 ; GFX11-NEXT:    s_subb_u32 s11, s11, s5
 ; GFX11-NEXT:    s_sub_u32 s13, s10, s4
+; GFX11-NEXT:    s_cselect_b32 s14, -1, 0
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX11-NEXT:    s_cmp_lg_u32 s14, 0
 ; GFX11-NEXT:    s_subb_u32 s11, s11, 0
-; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-NEXT:    s_cmp_ge_u32 s11, s5
 ; GFX11-NEXT:    s_cselect_b32 s14, -1, 0
 ; GFX11-NEXT:    s_cmp_ge_u32 s13, s4
@@ -3050,8 +3118,9 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX1250-NEXT:    s_load_b64 s[6:7], s[4:5], 0x34
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    s_or_b64 s[4:5], s[2:3], s[6:7]
-; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-NEXT:    s_and_b64 s[4:5], s[4:5], 0xffffffff00000000
+; GFX1250-NEXT:    s_cmp_lg_u64 s[4:5], 0
 ; GFX1250-NEXT:    s_cbranch_scc0 .LBB16_4
 ; GFX1250-NEXT:  ; %bb.1:
 ; GFX1250-NEXT:    s_cvt_f32_u32 s4, s6
@@ -3086,9 +3155,12 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-NEXT:    s_add_nc_u64 s[12:13], s[4:5], s[12:13]
 ; GFX1250-NEXT:    s_add_co_u32 s8, s8, s12
+; GFX1250-NEXT:    s_cselect_b32 s4, -1, 0
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX1250-NEXT:    s_add_co_ci_u32 s9, s9, s13
-; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-NEXT:    s_mul_u64 s[10:11], s[10:11], s[8:9]
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-NEXT:    s_mul_hi_u32 s13, s8, s11
 ; GFX1250-NEXT:    s_mul_i32 s12, s8, s11
 ; GFX1250-NEXT:    s_mul_hi_u32 s4, s8, s10
@@ -3103,17 +3175,19 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-NEXT:    s_add_nc_u64 s[10:11], s[4:5], s[10:11]
 ; GFX1250-NEXT:    s_add_co_u32 s8, s8, s10
-; GFX1250-NEXT:    s_add_co_ci_u32 s10, s9, s11
+; GFX1250-NEXT:    s_cselect_b32 s10, -1, 0
 ; GFX1250-NEXT:    s_mul_hi_u32 s4, s2, s8
-; GFX1250-NEXT:    s_mul_hi_u32 s11, s3, s8
-; GFX1250-NEXT:    s_mul_i32 s12, s3, s8
+; GFX1250-NEXT:    s_cmp_lg_u32 s10, 0
+; GFX1250-NEXT:    s_mul_hi_u32 s12, s3, s8
+; GFX1250-NEXT:    s_add_co_ci_u32 s10, s9, s11
+; GFX1250-NEXT:    s_mul_i32 s11, s3, s8
 ; GFX1250-NEXT:    s_mul_hi_u32 s9, s2, s10
 ; GFX1250-NEXT:    s_mul_i32 s8, s2, s10
 ; GFX1250-NEXT:    s_mul_hi_u32 s13, s3, s10
 ; GFX1250-NEXT:    s_add_nc_u64 s[8:9], s[4:5], s[8:9]
 ; GFX1250-NEXT:    s_mul_i32 s10, s3, s10
-; GFX1250-NEXT:    s_add_co_u32 s4, s8, s12
-; GFX1250-NEXT:    s_add_co_ci_u32 s4, s9, s11
+; GFX1250-NEXT:    s_add_co_u32 s4, s8, s11
+; GFX1250-NEXT:    s_add_co_ci_u32 s4, s9, s12
 ; GFX1250-NEXT:    s_add_co_ci_u32 s11, s13, 0
 ; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-NEXT:    s_add_nc_u64 s[8:9], s[4:5], s[10:11]
@@ -3128,8 +3202,10 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GFX1250-NEXT:    s_cmp_lg_u32 s8, 0
 ; GFX1250-NEXT:    s_sub_co_ci_u32 s12, s12, s7
 ; GFX1250-NEXT:    s_sub_co_u32 s13, s4, s6
+; GFX1250-NEXT:    s_cselect_b32 s14, -1, 0
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_cmp_lg_u32 s14, 0
 ; GFX1250-NEXT:    s_sub_co_ci_u32 s12, s12, 0
-; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-NEXT:    s_cmp_ge_u32 s12, s7
 ; GFX1250-NEXT:    s_cselect_b32 s14, -1, 0
 ; GFX1250-NEXT:    s_cmp_ge_u32 s13, s6
diff --git a/llvm/test/CodeGen/AMDGPU/ctlz_zero_undef.ll b/llvm/test/CodeGen/AMDGPU/ctlz_zero_undef.ll
index 07e6a76..4b151b9 100644
--- a/llvm/test/CodeGen/AMDGPU/ctlz_zero_undef.ll
+++ b/llvm/test/CodeGen/AMDGPU/ctlz_zero_undef.ll
@@ -714,8 +714,9 @@ define amdgpu_kernel void @v_ctlz_zero_undef_i16_with_select(ptr addrspace(1) no
 ; VI-NEXT:    s_lshl_b32 s2, s2, 8
 ; VI-NEXT:    s_or_b32 s2, s2, s3
 ; VI-NEXT:    s_lshl_b32 s3, s2, 16
-; VI-NEXT:    s_flbit_i32_b32 s3, s3
 ; VI-NEXT:    s_and_b32 s2, s2, 0xffff
+; VI-NEXT:    s_flbit_i32_b32 s3, s3
+; VI-NEXT:    s_cmp_lg_u32 s2, 0
 ; VI-NEXT:    s_cselect_b32 s2, s3, 32
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
diff --git a/llvm/test/CodeGen/AMDGPU/ctpop16.ll b/llvm/test/CodeGen/AMDGPU/ctpop16.ll
index fca57be..cefcbdd 100644
--- a/llvm/test/CodeGen/AMDGPU/ctpop16.ll
+++ b/llvm/test/CodeGen/AMDGPU/ctpop16.ll
@@ -1491,6 +1491,7 @@ define amdgpu_kernel void @ctpop_i16_in_br(ptr addrspace(1) %out, ptr addrspace(
 ; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    s_lshr_b32 s4, s6, 16
+; SI-NEXT:    s_cmp_lg_u32 s4, 0
 ; SI-NEXT:    s_cbranch_scc0 .LBB14_4
 ; SI-NEXT:  ; %bb.1: ; %else
 ; SI-NEXT:    s_mov_b32 s11, 0xf000
@@ -1520,6 +1521,7 @@ define amdgpu_kernel void @ctpop_i16_in_br(ptr addrspace(1) %out, ptr addrspace(
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
 ; VI-NEXT:    s_lshr_b32 s4, s6, 16
+; VI-NEXT:    s_cmp_lg_u32 s4, 0
 ; VI-NEXT:    s_cbranch_scc0 .LBB14_4
 ; VI-NEXT:  ; %bb.1: ; %else
 ; VI-NEXT:    s_mov_b32 s11, 0xf000
diff --git a/llvm/test/CodeGen/AMDGPU/expand-scalar-carry-out-select-user.ll b/llvm/test/CodeGen/AMDGPU/expand-scalar-carry-out-select-user.ll
index dbdea8e..d8a5e7fa 100644
--- a/llvm/test/CodeGen/AMDGPU/expand-scalar-carry-out-select-user.ll
+++ b/llvm/test/CodeGen/AMDGPU/expand-scalar-carry-out-select-user.ll
@@ -14,6 +14,7 @@ define i32 @s_add_co_select_user() {
 ; GFX7-NEXT:    s_add_u32 s7, s6, s6
 ; GFX7-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GFX7-NEXT:    s_or_b32 s4, s4, s5
+; GFX7-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX7-NEXT:    s_addc_u32 s8, s6, 0
 ; GFX7-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GFX7-NEXT:    s_and_b64 s[4:5], s[4:5], exec
@@ -30,6 +31,8 @@ define i32 @s_add_co_select_user() {
 ; GFX9-NEXT:    s_load_dword s6, s[4:5], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_add_u32 s7, s6, s6
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
 ; GFX9-NEXT:    s_addc_u32 s8, s6, 0
 ; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GFX9-NEXT:    s_and_b64 s[4:5], s[4:5], exec
@@ -46,6 +49,8 @@ define i32 @s_add_co_select_user() {
 ; GFX10-NEXT:    s_load_dword s4, s[4:5], 0x0
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    s_add_u32 s5, s4, s4
+; GFX10-NEXT:    s_cselect_b32 s6, -1, 0
+; GFX10-NEXT:    s_cmp_lg_u32 s6, 0
 ; GFX10-NEXT:    s_addc_u32 s6, s4, 0
 ; GFX10-NEXT:    s_cselect_b32 s7, -1, 0
 ; GFX10-NEXT:    s_and_b32 s7, s7, exec_lo
@@ -62,13 +67,16 @@ define i32 @s_add_co_select_user() {
 ; GFX11-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-NEXT:    s_add_u32 s1, s0, s0
+; GFX11-NEXT:    s_cselect_b32 s2, -1, 0
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
+; GFX11-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX11-NEXT:    s_addc_u32 s2, s0, 0
 ; GFX11-NEXT:    s_cselect_b32 s3, -1, 0
-; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_3) | instid1(SALU_CYCLE_1)
 ; GFX11-NEXT:    s_and_b32 s3, s3, exec_lo
 ; GFX11-NEXT:    s_cselect_b32 s2, s2, 0
 ; GFX11-NEXT:    s_cmp_gt_u32 s0, 31
 ; GFX11-NEXT:    s_cselect_b32 s0, s1, s2
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-NEXT:    s_setpc_b64 s[30:31]
 bb:
@@ -96,6 +104,7 @@ define amdgpu_kernel void @s_add_co_br_user(i32 %i) {
 ; GFX7-NEXT:    s_add_u32 s0, s2, s2
 ; GFX7-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GFX7-NEXT:    s_or_b32 s0, s0, s1
+; GFX7-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX7-NEXT:    s_addc_u32 s0, s2, 0
 ; GFX7-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GFX7-NEXT:    s_andn2_b64 vcc, exec, s[0:1]
@@ -116,10 +125,12 @@ define amdgpu_kernel void @s_add_co_br_user(i32 %i) {
 ;
 ; GFX9-LABEL: s_add_co_br_user:
 ; GFX9:       ; %bb.0: ; %bb
-; GFX9-NEXT:    s_load_dword s0, s[8:9], 0x0
+; GFX9-NEXT:    s_load_dword s2, s[8:9], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_add_u32 s1, s0, s0
-; GFX9-NEXT:    s_addc_u32 s0, s0, 0
+; GFX9-NEXT:    s_add_u32 s0, s2, s2
+; GFX9-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    s_addc_u32 s0, s2, 0
 ; GFX9-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GFX9-NEXT:    s_andn2_b64 vcc, exec, s[0:1]
 ; GFX9-NEXT:    s_cbranch_vccnz .LBB1_2
@@ -142,6 +153,8 @@ define amdgpu_kernel void @s_add_co_br_user(i32 %i) {
 ; GFX10-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    s_add_u32 s1, s0, s0
+; GFX10-NEXT:    s_cselect_b32 s1, -1, 0
+; GFX10-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX10-NEXT:    s_addc_u32 s0, s0, 0
 ; GFX10-NEXT:    s_cselect_b32 s0, -1, 0
 ; GFX10-NEXT:    s_andn2_b32 vcc_lo, exec_lo, s0
@@ -165,9 +178,11 @@ define amdgpu_kernel void @s_add_co_br_user(i32 %i) {
 ; GFX11-NEXT:    s_load_b32 s0, s[4:5], 0x0
 ; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-NEXT:    s_add_u32 s1, s0, s0
+; GFX11-NEXT:    s_cselect_b32 s1, -1, 0
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
+; GFX11-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX11-NEXT:    s_addc_u32 s0, s0, 0
 ; GFX11-NEXT:    s_cselect_b32 s0, -1, 0
-; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-NEXT:    s_and_not1_b32 vcc_lo, exec_lo, s0
 ; GFX11-NEXT:    s_cbranch_vccnz .LBB1_2
 ; GFX11-NEXT:  ; %bb.1: ; %bb0
diff --git a/llvm/test/CodeGen/AMDGPU/fcopysign.f16.ll b/llvm/test/CodeGen/AMDGPU/fcopysign.f16.ll
index 9a17538..62847b1 100644
--- a/llvm/test/CodeGen/AMDGPU/fcopysign.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/fcopysign.f16.ll
@@ -1117,6 +1117,7 @@ define amdgpu_ps i16 @s_copysign_out_f16_mag_f64_sign_f16(double inreg %mag, hal
 ; SI:       ; %bb.0:
 ; SI-NEXT:    s_and_b32 s3, s1, 0x1ff
 ; SI-NEXT:    s_or_b32 s0, s3, s0
+; SI-NEXT:    s_cmp_lg_u32 s0, 0
 ; SI-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; SI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[4:5]
 ; SI-NEXT:    s_lshr_b32 s0, s1, 8
@@ -1168,6 +1169,7 @@ define amdgpu_ps i16 @s_copysign_out_f16_mag_f64_sign_f16(double inreg %mag, hal
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_and_b32 s3, s1, 0x1ff
 ; VI-NEXT:    s_or_b32 s0, s3, s0
+; VI-NEXT:    s_cmp_lg_u32 s0, 0
 ; VI-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; VI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[4:5]
 ; VI-NEXT:    s_lshr_b32 s0, s1, 8
@@ -1215,6 +1217,7 @@ define amdgpu_ps i16 @s_copysign_out_f16_mag_f64_sign_f16(double inreg %mag, hal
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_and_b32 s3, s1, 0x1ff
 ; GFX9-NEXT:    s_or_b32 s0, s3, s0
+; GFX9-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GFX9-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[4:5]
 ; GFX9-NEXT:    s_lshr_b32 s0, s1, 8
@@ -1261,9 +1264,11 @@ define amdgpu_ps i16 @s_copysign_out_f16_mag_f64_sign_f16(double inreg %mag, hal
 ; GFX11-TRUE16-LABEL: s_copysign_out_f16_mag_f64_sign_f16:
 ; GFX11-TRUE16:       ; %bb.0:
 ; GFX11-TRUE16-NEXT:    s_and_b32 s3, s1, 0x1ff
-; GFX11-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX11-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX11-TRUE16-NEXT:    s_or_b32 s0, s3, s0
+; GFX11-TRUE16-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX11-TRUE16-NEXT:    s_cselect_b32 s0, -1, 0
+; GFX11-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-TRUE16-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s0
 ; GFX11-TRUE16-NEXT:    s_bfe_u32 s0, s1, 0xb0014
 ; GFX11-TRUE16-NEXT:    s_lshr_b32 s1, s1, 8
@@ -1315,9 +1320,11 @@ define amdgpu_ps i16 @s_copysign_out_f16_mag_f64_sign_f16(double inreg %mag, hal
 ; GFX11-FAKE16-LABEL: s_copysign_out_f16_mag_f64_sign_f16:
 ; GFX11-FAKE16:       ; %bb.0:
 ; GFX11-FAKE16-NEXT:    s_and_b32 s3, s1, 0x1ff
-; GFX11-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX11-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX11-FAKE16-NEXT:    s_or_b32 s0, s3, s0
+; GFX11-FAKE16-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX11-FAKE16-NEXT:    s_cselect_b32 s0, -1, 0
+; GFX11-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX11-FAKE16-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s0
 ; GFX11-FAKE16-NEXT:    s_bfe_u32 s0, s1, 0xb0014
 ; GFX11-FAKE16-NEXT:    s_lshr_b32 s1, s1, 8
@@ -4016,6 +4023,7 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
 ; SI-NEXT:    s_and_b32 s6, s4, 0xffe
 ; SI-NEXT:    s_and_b32 s4, s1, 0x1ff
 ; SI-NEXT:    s_or_b32 s0, s4, s0
+; SI-NEXT:    s_cmp_lg_u32 s0, 0
 ; SI-NEXT:    v_cvt_f16_f32_e32 v0, s5
 ; SI-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; SI-NEXT:    v_cndmask_b32_e64 v2, 0, 1, s[4:5]
@@ -4058,6 +4066,7 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
 ; SI-NEXT:    s_and_b32 s5, s0, 0xffe
 ; SI-NEXT:    s_and_b32 s0, s3, 0x1ff
 ; SI-NEXT:    s_or_b32 s0, s0, s2
+; SI-NEXT:    s_cmp_lg_u32 s0, 0
 ; SI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; SI-NEXT:    v_cndmask_b32_e64 v2, 0, 1, s[0:1]
 ; SI-NEXT:    v_readfirstlane_b32 s0, v2
@@ -4111,9 +4120,10 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
 ; VI-LABEL: s_copysign_out_v2f16_mag_v2f64_sign_v2f16:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_lshr_b32 s5, s3, 8
-; VI-NEXT:    s_and_b32 s5, s5, 0xffe
 ; VI-NEXT:    s_and_b32 s6, s3, 0x1ff
+; VI-NEXT:    s_and_b32 s5, s5, 0xffe
 ; VI-NEXT:    s_or_b32 s2, s6, s2
+; VI-NEXT:    s_cmp_lg_u32 s2, 0
 ; VI-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; VI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[6:7]
 ; VI-NEXT:    s_bfe_u32 s3, s3, 0xb0014
@@ -4153,6 +4163,7 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
 ; VI-NEXT:    s_and_b32 s7, s2, 0xffe
 ; VI-NEXT:    s_and_b32 s2, s1, 0x1ff
 ; VI-NEXT:    s_or_b32 s0, s2, s0
+; VI-NEXT:    s_cmp_lg_u32 s0, 0
 ; VI-NEXT:    s_cselect_b64 s[2:3], -1, 0
 ; VI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[2:3]
 ; VI-NEXT:    s_bfe_u32 s1, s1, 0xb0014
@@ -4198,9 +4209,10 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
 ; GFX9-LABEL: s_copysign_out_v2f16_mag_v2f64_sign_v2f16:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_lshr_b32 s5, s3, 8
-; GFX9-NEXT:    s_and_b32 s5, s5, 0xffe
 ; GFX9-NEXT:    s_and_b32 s6, s3, 0x1ff
+; GFX9-NEXT:    s_and_b32 s5, s5, 0xffe
 ; GFX9-NEXT:    s_or_b32 s2, s6, s2
+; GFX9-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX9-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GFX9-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[6:7]
 ; GFX9-NEXT:    s_bfe_u32 s6, s3, 0xb0014
@@ -4242,6 +4254,7 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
 ; GFX9-NEXT:    s_and_b32 s6, s2, 0xffe
 ; GFX9-NEXT:    s_and_b32 s2, s1, 0x1ff
 ; GFX9-NEXT:    s_or_b32 s0, s2, s0
+; GFX9-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX9-NEXT:    s_cselect_b64 s[2:3], -1, 0
 ; GFX9-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[2:3]
 ; GFX9-NEXT:    v_readfirstlane_b32 s0, v0
@@ -4288,10 +4301,11 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
 ;
 ; GFX11-LABEL: s_copysign_out_v2f16_mag_v2f64_sign_v2f16:
 ; GFX11:       ; %bb.0:
-; GFX11-NEXT:    s_lshr_b32 s5, s3, 8
-; GFX11-NEXT:    s_and_b32 s6, s3, 0x1ff
-; GFX11-NEXT:    s_and_b32 s5, s5, 0xffe
-; GFX11-NEXT:    s_or_b32 s2, s6, s2
+; GFX11-NEXT:    s_and_b32 s5, s3, 0x1ff
+; GFX11-NEXT:    s_lshr_b32 s6, s3, 8
+; GFX11-NEXT:    s_or_b32 s2, s5, s2
+; GFX11-NEXT:    s_and_b32 s5, s6, 0xffe
+; GFX11-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX11-NEXT:    s_cselect_b32 s2, -1, 0
 ; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
 ; GFX11-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s2
@@ -4334,12 +4348,13 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
 ; GFX11-NEXT:    s_cmpk_eq_i32 s2, 0x40f
 ; GFX11-NEXT:    s_cselect_b32 s2, s5, s6
 ; GFX11-NEXT:    s_lshr_b32 s3, s3, 16
+; GFX11-NEXT:    s_and_b32 s6, s1, 0x1ff
 ; GFX11-NEXT:    s_lshr_b32 s5, s1, 8
 ; GFX11-NEXT:    s_and_b32 s3, s3, 0x8000
-; GFX11-NEXT:    s_and_b32 s6, s1, 0x1ff
+; GFX11-NEXT:    s_or_b32 s0, s6, s0
 ; GFX11-NEXT:    s_and_b32 s5, s5, 0xffe
 ; GFX11-NEXT:    s_or_b32 s2, s3, s2
-; GFX11-NEXT:    s_or_b32 s0, s6, s0
+; GFX11-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX11-NEXT:    s_cselect_b32 s0, -1, 0
 ; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
 ; GFX11-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s0
diff --git a/llvm/test/CodeGen/AMDGPU/fptrunc.f16.ll b/llvm/test/CodeGen/AMDGPU/fptrunc.f16.ll
index c28b25c7..b0dd187 100644
--- a/llvm/test/CodeGen/AMDGPU/fptrunc.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/fptrunc.f16.ll
@@ -599,8 +599,10 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
 ; SI-GISEL-NEXT:    s_addk_i32 s3, 0xfc10
 ; SI-GISEL-NEXT:    s_and_b32 s6, s6, 0xffe
 ; SI-GISEL-NEXT:    s_or_b32 s4, s7, s4
+; SI-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; SI-GISEL-NEXT:    s_cselect_b32 s4, 1, 0
 ; SI-GISEL-NEXT:    s_or_b32 s4, s6, s4
+; SI-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; SI-GISEL-NEXT:    s_cselect_b32 s6, 1, 0
 ; SI-GISEL-NEXT:    s_lshl_b32 s6, s6, 9
 ; SI-GISEL-NEXT:    s_lshl_b32 s7, s3, 12
@@ -709,8 +711,10 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
 ; VI-GISEL-NEXT:    s_addk_i32 s4, 0xfc10
 ; VI-GISEL-NEXT:    s_and_b32 s5, s5, 0xffe
 ; VI-GISEL-NEXT:    s_or_b32 s2, s6, s2
+; VI-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; VI-GISEL-NEXT:    s_cselect_b32 s2, 1, 0
 ; VI-GISEL-NEXT:    s_or_b32 s2, s5, s2
+; VI-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; VI-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; VI-GISEL-NEXT:    s_sub_i32 s7, 1, s4
 ; VI-GISEL-NEXT:    s_lshl_b32 s6, s4, 12
@@ -820,8 +824,10 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
 ; GFX9-GISEL-NEXT:    s_addk_i32 s4, 0xfc10
 ; GFX9-GISEL-NEXT:    s_and_b32 s5, s5, 0xffe
 ; GFX9-GISEL-NEXT:    s_or_b32 s2, s6, s2
+; GFX9-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX9-GISEL-NEXT:    s_cselect_b32 s2, 1, 0
 ; GFX9-GISEL-NEXT:    s_or_b32 s2, s5, s2
+; GFX9-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX9-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX9-GISEL-NEXT:    s_sub_i32 s7, 1, s4
 ; GFX9-GISEL-NEXT:    s_lshl_b32 s6, s4, 12
@@ -931,8 +937,10 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
 ; GFX950-GISEL-NEXT:    s_addk_i32 s4, 0xfc10
 ; GFX950-GISEL-NEXT:    s_and_b32 s5, s5, 0xffe
 ; GFX950-GISEL-NEXT:    s_or_b32 s2, s6, s2
+; GFX950-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX950-GISEL-NEXT:    s_cselect_b32 s2, 1, 0
 ; GFX950-GISEL-NEXT:    s_or_b32 s2, s5, s2
+; GFX950-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX950-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX950-GISEL-NEXT:    s_sub_i32 s7, 1, s4
 ; GFX950-GISEL-NEXT:    s_lshl_b32 s6, s4, 12
@@ -1110,15 +1118,17 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
 ; GFX11-GISEL-TRUE16-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-GISEL-TRUE16-NEXT:    s_load_b64 s[2:3], s[2:3], 0x0
 ; GFX11-GISEL-TRUE16-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-GISEL-TRUE16-NEXT:    s_and_b32 s6, s3, 0x1ff
 ; GFX11-GISEL-TRUE16-NEXT:    s_bfe_u32 s4, s3, 0xb0014
 ; GFX11-GISEL-TRUE16-NEXT:    s_lshr_b32 s5, s3, 8
-; GFX11-GISEL-TRUE16-NEXT:    s_and_b32 s6, s3, 0x1ff
+; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s2, s6, s2
 ; GFX11-GISEL-TRUE16-NEXT:    s_addk_i32 s4, 0xfc10
 ; GFX11-GISEL-TRUE16-NEXT:    s_and_b32 s5, s5, 0xffe
-; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s2, s6, s2
+; GFX11-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX11-GISEL-TRUE16-NEXT:    s_cselect_b32 s2, 1, 0
-; GFX11-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s2, s5, s2
+; GFX11-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX11-GISEL-TRUE16-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX11-GISEL-TRUE16-NEXT:    s_sub_i32 s6, 1, s4
 ; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s8, s2, 0x1000
@@ -1165,15 +1175,17 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
 ; GFX11-GISEL-FAKE16-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-GISEL-FAKE16-NEXT:    s_load_b64 s[2:3], s[2:3], 0x0
 ; GFX11-GISEL-FAKE16-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-GISEL-FAKE16-NEXT:    s_and_b32 s6, s3, 0x1ff
 ; GFX11-GISEL-FAKE16-NEXT:    s_bfe_u32 s4, s3, 0xb0014
 ; GFX11-GISEL-FAKE16-NEXT:    s_lshr_b32 s5, s3, 8
-; GFX11-GISEL-FAKE16-NEXT:    s_and_b32 s6, s3, 0x1ff
+; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s2, s6, s2
 ; GFX11-GISEL-FAKE16-NEXT:    s_addk_i32 s4, 0xfc10
 ; GFX11-GISEL-FAKE16-NEXT:    s_and_b32 s5, s5, 0xffe
-; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s2, s6, s2
+; GFX11-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX11-GISEL-FAKE16-NEXT:    s_cselect_b32 s2, 1, 0
-; GFX11-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s2, s5, s2
+; GFX11-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX11-GISEL-FAKE16-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX11-GISEL-FAKE16-NEXT:    s_sub_i32 s6, 1, s4
 ; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s8, s2, 0x1000
@@ -1354,15 +1366,17 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
 ; GFX1250-GISEL-TRUE16-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_load_b64 s[2:3], s[2:3], 0x0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-GISEL-TRUE16-NEXT:    s_and_b32 s6, s3, 0x1ff
 ; GFX1250-GISEL-TRUE16-NEXT:    s_bfe_u32 s4, s3, 0xb0014
 ; GFX1250-GISEL-TRUE16-NEXT:    s_lshr_b32 s5, s3, 8
-; GFX1250-GISEL-TRUE16-NEXT:    s_and_b32 s6, s3, 0x1ff
+; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s2, s6, s2
 ; GFX1250-GISEL-TRUE16-NEXT:    s_addk_co_i32 s4, 0xfc10
 ; GFX1250-GISEL-TRUE16-NEXT:    s_and_b32 s5, s5, 0xffe
-; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s2, s6, s2
+; GFX1250-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_cselect_b32 s2, 1, 0
-; GFX1250-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s2, s5, s2
+; GFX1250-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_sub_co_i32 s6, 1, s4
 ; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s8, s2, 0x1000
@@ -1409,15 +1423,17 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
 ; GFX1250-GISEL-FAKE16-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_load_b64 s[2:3], s[2:3], 0x0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-GISEL-FAKE16-NEXT:    s_and_b32 s6, s3, 0x1ff
 ; GFX1250-GISEL-FAKE16-NEXT:    s_bfe_u32 s4, s3, 0xb0014
 ; GFX1250-GISEL-FAKE16-NEXT:    s_lshr_b32 s5, s3, 8
-; GFX1250-GISEL-FAKE16-NEXT:    s_and_b32 s6, s3, 0x1ff
+; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s2, s6, s2
 ; GFX1250-GISEL-FAKE16-NEXT:    s_addk_co_i32 s4, 0xfc10
 ; GFX1250-GISEL-FAKE16-NEXT:    s_and_b32 s5, s5, 0xffe
-; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s2, s6, s2
+; GFX1250-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_cselect_b32 s2, 1, 0
-; GFX1250-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s2, s5, s2
+; GFX1250-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_sub_co_i32 s6, 1, s4
 ; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s8, s2, 0x1000
@@ -2138,8 +2154,10 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; SI-GISEL-NEXT:    s_addk_i32 s3, 0xfc10
 ; SI-GISEL-NEXT:    s_and_b32 s8, s8, 0xffe
 ; SI-GISEL-NEXT:    s_or_b32 s4, s9, s4
+; SI-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; SI-GISEL-NEXT:    s_cselect_b32 s4, 1, 0
 ; SI-GISEL-NEXT:    s_or_b32 s4, s8, s4
+; SI-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; SI-GISEL-NEXT:    s_cselect_b32 s8, 1, 0
 ; SI-GISEL-NEXT:    s_lshl_b32 s8, s8, 9
 ; SI-GISEL-NEXT:    s_lshl_b32 s9, s3, 12
@@ -2175,10 +2193,12 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; SI-GISEL-NEXT:    s_and_b32 s4, s4, 0x8000
 ; SI-GISEL-NEXT:    s_addk_i32 s5, 0xfc10
 ; SI-GISEL-NEXT:    s_and_b32 s8, s8, 0xffe
+; SI-GISEL-NEXT:    s_or_b32 s6, s9, s6
 ; SI-GISEL-NEXT:    s_or_b32 s3, s4, s3
-; SI-GISEL-NEXT:    s_or_b32 s4, s9, s6
+; SI-GISEL-NEXT:    s_cmp_lg_u32 s6, 0
 ; SI-GISEL-NEXT:    s_cselect_b32 s4, 1, 0
 ; SI-GISEL-NEXT:    s_or_b32 s4, s8, s4
+; SI-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; SI-GISEL-NEXT:    s_cselect_b32 s6, 1, 0
 ; SI-GISEL-NEXT:    s_lshl_b32 s6, s6, 9
 ; SI-GISEL-NEXT:    s_lshl_b32 s8, s5, 12
@@ -2335,8 +2355,10 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; VI-GISEL-NEXT:    s_addk_i32 s2, 0xfc10
 ; VI-GISEL-NEXT:    s_and_b32 s3, s3, 0xffe
 ; VI-GISEL-NEXT:    s_or_b32 s4, s8, s4
+; VI-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; VI-GISEL-NEXT:    s_cselect_b32 s4, 1, 0
 ; VI-GISEL-NEXT:    s_or_b32 s3, s3, s4
+; VI-GISEL-NEXT:    s_cmp_lg_u32 s3, 0
 ; VI-GISEL-NEXT:    s_cselect_b32 s4, 1, 0
 ; VI-GISEL-NEXT:    s_sub_i32 s9, 1, s2
 ; VI-GISEL-NEXT:    s_lshl_b32 s8, s2, 12
@@ -2370,12 +2392,14 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; VI-GISEL-NEXT:    s_or_b32 s2, s3, s2
 ; VI-GISEL-NEXT:    s_bfe_u32 s3, s7, 0xb0014
 ; VI-GISEL-NEXT:    s_lshr_b32 s4, s7, 8
+; VI-GISEL-NEXT:    s_and_b32 s5, s7, 0x1ff
 ; VI-GISEL-NEXT:    s_addk_i32 s3, 0xfc10
 ; VI-GISEL-NEXT:    s_and_b32 s4, s4, 0xffe
-; VI-GISEL-NEXT:    s_and_b32 s5, s7, 0x1ff
 ; VI-GISEL-NEXT:    s_or_b32 s5, s5, s6
+; VI-GISEL-NEXT:    s_cmp_lg_u32 s5, 0
 ; VI-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; VI-GISEL-NEXT:    s_or_b32 s4, s4, s5
+; VI-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; VI-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; VI-GISEL-NEXT:    s_sub_i32 s8, 1, s3
 ; VI-GISEL-NEXT:    s_lshl_b32 s6, s3, 12
@@ -2531,8 +2555,10 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX9-GISEL-NEXT:    s_addk_i32 s2, 0xfc10
 ; GFX9-GISEL-NEXT:    s_and_b32 s3, s3, 0xffe
 ; GFX9-GISEL-NEXT:    s_or_b32 s4, s8, s4
+; GFX9-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX9-GISEL-NEXT:    s_cselect_b32 s4, 1, 0
 ; GFX9-GISEL-NEXT:    s_or_b32 s3, s3, s4
+; GFX9-GISEL-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX9-GISEL-NEXT:    s_cselect_b32 s4, 1, 0
 ; GFX9-GISEL-NEXT:    s_sub_i32 s9, 1, s2
 ; GFX9-GISEL-NEXT:    s_lshl_b32 s8, s2, 12
@@ -2566,12 +2592,14 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX9-GISEL-NEXT:    s_or_b32 s2, s3, s2
 ; GFX9-GISEL-NEXT:    s_bfe_u32 s3, s7, 0xb0014
 ; GFX9-GISEL-NEXT:    s_lshr_b32 s4, s7, 8
+; GFX9-GISEL-NEXT:    s_and_b32 s5, s7, 0x1ff
 ; GFX9-GISEL-NEXT:    s_addk_i32 s3, 0xfc10
 ; GFX9-GISEL-NEXT:    s_and_b32 s4, s4, 0xffe
-; GFX9-GISEL-NEXT:    s_and_b32 s5, s7, 0x1ff
 ; GFX9-GISEL-NEXT:    s_or_b32 s5, s5, s6
+; GFX9-GISEL-NEXT:    s_cmp_lg_u32 s5, 0
 ; GFX9-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX9-GISEL-NEXT:    s_or_b32 s4, s4, s5
+; GFX9-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX9-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX9-GISEL-NEXT:    s_sub_i32 s8, 1, s3
 ; GFX9-GISEL-NEXT:    s_lshl_b32 s6, s3, 12
@@ -2724,8 +2752,10 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX950-GISEL-NEXT:    s_addk_i32 s2, 0xfc10
 ; GFX950-GISEL-NEXT:    s_and_b32 s3, s3, 0xffe
 ; GFX950-GISEL-NEXT:    s_or_b32 s4, s8, s4
+; GFX950-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX950-GISEL-NEXT:    s_cselect_b32 s4, 1, 0
 ; GFX950-GISEL-NEXT:    s_or_b32 s3, s3, s4
+; GFX950-GISEL-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX950-GISEL-NEXT:    s_cselect_b32 s4, 1, 0
 ; GFX950-GISEL-NEXT:    s_sub_i32 s9, 1, s2
 ; GFX950-GISEL-NEXT:    s_lshl_b32 s8, s2, 12
@@ -2759,12 +2789,14 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX950-GISEL-NEXT:    s_or_b32 s2, s3, s2
 ; GFX950-GISEL-NEXT:    s_bfe_u32 s3, s7, 0xb0014
 ; GFX950-GISEL-NEXT:    s_lshr_b32 s4, s7, 8
+; GFX950-GISEL-NEXT:    s_and_b32 s5, s7, 0x1ff
 ; GFX950-GISEL-NEXT:    s_addk_i32 s3, 0xfc10
 ; GFX950-GISEL-NEXT:    s_and_b32 s4, s4, 0xffe
-; GFX950-GISEL-NEXT:    s_and_b32 s5, s7, 0x1ff
 ; GFX950-GISEL-NEXT:    s_or_b32 s5, s5, s6
+; GFX950-GISEL-NEXT:    s_cmp_lg_u32 s5, 0
 ; GFX950-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX950-GISEL-NEXT:    s_or_b32 s4, s4, s5
+; GFX950-GISEL-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX950-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX950-GISEL-NEXT:    s_sub_i32 s8, 1, s3
 ; GFX950-GISEL-NEXT:    s_lshl_b32 s6, s3, 12
@@ -3041,15 +3073,17 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX11-GISEL-TRUE16-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-GISEL-TRUE16-NEXT:    s_load_b128 s[4:7], s[2:3], 0x0
 ; GFX11-GISEL-TRUE16-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-GISEL-TRUE16-NEXT:    s_and_b32 s8, s5, 0x1ff
 ; GFX11-GISEL-TRUE16-NEXT:    s_bfe_u32 s2, s5, 0xb0014
 ; GFX11-GISEL-TRUE16-NEXT:    s_lshr_b32 s3, s5, 8
-; GFX11-GISEL-TRUE16-NEXT:    s_and_b32 s8, s5, 0x1ff
+; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s4, s8, s4
 ; GFX11-GISEL-TRUE16-NEXT:    s_addk_i32 s2, 0xfc10
 ; GFX11-GISEL-TRUE16-NEXT:    s_and_b32 s3, s3, 0xffe
-; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s4, s8, s4
+; GFX11-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX11-GISEL-TRUE16-NEXT:    s_cselect_b32 s4, 1, 0
-; GFX11-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s3, s3, s4
+; GFX11-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX11-GISEL-TRUE16-NEXT:    s_cselect_b32 s4, 1, 0
 ; GFX11-GISEL-TRUE16-NEXT:    s_sub_i32 s8, 1, s2
 ; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s10, s3, 0x1000
@@ -3081,17 +3115,19 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX11-GISEL-TRUE16-NEXT:    s_cmpk_eq_i32 s2, 0x40f
 ; GFX11-GISEL-TRUE16-NEXT:    s_cselect_b32 s2, s4, s3
 ; GFX11-GISEL-TRUE16-NEXT:    s_lshr_b32 s3, s5, 16
+; GFX11-GISEL-TRUE16-NEXT:    s_and_b32 s8, s7, 0x1ff
 ; GFX11-GISEL-TRUE16-NEXT:    s_bfe_u32 s4, s7, 0xb0014
 ; GFX11-GISEL-TRUE16-NEXT:    s_lshr_b32 s5, s7, 8
 ; GFX11-GISEL-TRUE16-NEXT:    s_and_b32 s3, s3, 0x8000
-; GFX11-GISEL-TRUE16-NEXT:    s_and_b32 s8, s7, 0x1ff
+; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s6, s8, s6
 ; GFX11-GISEL-TRUE16-NEXT:    s_addk_i32 s4, 0xfc10
 ; GFX11-GISEL-TRUE16-NEXT:    s_and_b32 s5, s5, 0xffe
 ; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s2, s3, s2
-; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s3, s8, s6
+; GFX11-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s6, 0
 ; GFX11-GISEL-TRUE16-NEXT:    s_cselect_b32 s3, 1, 0
-; GFX11-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s3, s5, s3
+; GFX11-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX11-GISEL-TRUE16-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX11-GISEL-TRUE16-NEXT:    s_sub_i32 s6, 1, s4
 ; GFX11-GISEL-TRUE16-NEXT:    s_or_b32 s9, s3, 0x1000
@@ -3140,15 +3176,17 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX11-GISEL-FAKE16-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-GISEL-FAKE16-NEXT:    s_load_b128 s[4:7], s[2:3], 0x0
 ; GFX11-GISEL-FAKE16-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-GISEL-FAKE16-NEXT:    s_and_b32 s8, s5, 0x1ff
 ; GFX11-GISEL-FAKE16-NEXT:    s_bfe_u32 s2, s5, 0xb0014
 ; GFX11-GISEL-FAKE16-NEXT:    s_lshr_b32 s3, s5, 8
-; GFX11-GISEL-FAKE16-NEXT:    s_and_b32 s8, s5, 0x1ff
+; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s4, s8, s4
 ; GFX11-GISEL-FAKE16-NEXT:    s_addk_i32 s2, 0xfc10
 ; GFX11-GISEL-FAKE16-NEXT:    s_and_b32 s3, s3, 0xffe
-; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s4, s8, s4
+; GFX11-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX11-GISEL-FAKE16-NEXT:    s_cselect_b32 s4, 1, 0
-; GFX11-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s3, s3, s4
+; GFX11-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX11-GISEL-FAKE16-NEXT:    s_cselect_b32 s4, 1, 0
 ; GFX11-GISEL-FAKE16-NEXT:    s_sub_i32 s8, 1, s2
 ; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s10, s3, 0x1000
@@ -3180,17 +3218,19 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX11-GISEL-FAKE16-NEXT:    s_cmpk_eq_i32 s2, 0x40f
 ; GFX11-GISEL-FAKE16-NEXT:    s_cselect_b32 s2, s4, s3
 ; GFX11-GISEL-FAKE16-NEXT:    s_lshr_b32 s3, s5, 16
+; GFX11-GISEL-FAKE16-NEXT:    s_and_b32 s8, s7, 0x1ff
 ; GFX11-GISEL-FAKE16-NEXT:    s_bfe_u32 s4, s7, 0xb0014
 ; GFX11-GISEL-FAKE16-NEXT:    s_lshr_b32 s5, s7, 8
 ; GFX11-GISEL-FAKE16-NEXT:    s_and_b32 s3, s3, 0x8000
-; GFX11-GISEL-FAKE16-NEXT:    s_and_b32 s8, s7, 0x1ff
+; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s6, s8, s6
 ; GFX11-GISEL-FAKE16-NEXT:    s_addk_i32 s4, 0xfc10
 ; GFX11-GISEL-FAKE16-NEXT:    s_and_b32 s5, s5, 0xffe
 ; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s2, s3, s2
-; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s3, s8, s6
+; GFX11-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s6, 0
 ; GFX11-GISEL-FAKE16-NEXT:    s_cselect_b32 s3, 1, 0
-; GFX11-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s3, s5, s3
+; GFX11-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX11-GISEL-FAKE16-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX11-GISEL-FAKE16-NEXT:    s_sub_i32 s6, 1, s4
 ; GFX11-GISEL-FAKE16-NEXT:    s_or_b32 s9, s3, 0x1000
@@ -3471,15 +3511,17 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX1250-GISEL-TRUE16-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_load_b128 s[4:7], s[2:3], 0x0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-GISEL-TRUE16-NEXT:    s_and_b32 s8, s5, 0x1ff
 ; GFX1250-GISEL-TRUE16-NEXT:    s_bfe_u32 s2, s5, 0xb0014
 ; GFX1250-GISEL-TRUE16-NEXT:    s_lshr_b32 s3, s5, 8
-; GFX1250-GISEL-TRUE16-NEXT:    s_and_b32 s8, s5, 0x1ff
+; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s4, s8, s4
 ; GFX1250-GISEL-TRUE16-NEXT:    s_addk_co_i32 s2, 0xfc10
 ; GFX1250-GISEL-TRUE16-NEXT:    s_and_b32 s3, s3, 0xffe
-; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s4, s8, s4
+; GFX1250-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_cselect_b32 s4, 1, 0
-; GFX1250-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s3, s3, s4
+; GFX1250-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_cselect_b32 s4, 1, 0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_sub_co_i32 s8, 1, s2
 ; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s10, s3, 0x1000
@@ -3511,17 +3553,19 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX1250-GISEL-TRUE16-NEXT:    s_cmp_eq_u32 s2, 0x40f
 ; GFX1250-GISEL-TRUE16-NEXT:    s_cselect_b32 s2, s4, s3
 ; GFX1250-GISEL-TRUE16-NEXT:    s_lshr_b32 s3, s5, 16
+; GFX1250-GISEL-TRUE16-NEXT:    s_and_b32 s8, s7, 0x1ff
 ; GFX1250-GISEL-TRUE16-NEXT:    s_bfe_u32 s4, s7, 0xb0014
 ; GFX1250-GISEL-TRUE16-NEXT:    s_lshr_b32 s5, s7, 8
 ; GFX1250-GISEL-TRUE16-NEXT:    s_and_b32 s3, s3, 0x8000
-; GFX1250-GISEL-TRUE16-NEXT:    s_and_b32 s8, s7, 0x1ff
+; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s6, s8, s6
 ; GFX1250-GISEL-TRUE16-NEXT:    s_addk_co_i32 s4, 0xfc10
 ; GFX1250-GISEL-TRUE16-NEXT:    s_and_b32 s5, s5, 0xffe
 ; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s2, s3, s2
-; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s3, s8, s6
+; GFX1250-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s6, 0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_cselect_b32 s3, 1, 0
-; GFX1250-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-GISEL-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s3, s5, s3
+; GFX1250-GISEL-TRUE16-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX1250-GISEL-TRUE16-NEXT:    s_sub_co_i32 s6, 1, s4
 ; GFX1250-GISEL-TRUE16-NEXT:    s_or_b32 s9, s3, 0x1000
@@ -3570,15 +3614,17 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX1250-GISEL-FAKE16-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_load_b128 s[4:7], s[2:3], 0x0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-GISEL-FAKE16-NEXT:    s_and_b32 s8, s5, 0x1ff
 ; GFX1250-GISEL-FAKE16-NEXT:    s_bfe_u32 s2, s5, 0xb0014
 ; GFX1250-GISEL-FAKE16-NEXT:    s_lshr_b32 s3, s5, 8
-; GFX1250-GISEL-FAKE16-NEXT:    s_and_b32 s8, s5, 0x1ff
+; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s4, s8, s4
 ; GFX1250-GISEL-FAKE16-NEXT:    s_addk_co_i32 s2, 0xfc10
 ; GFX1250-GISEL-FAKE16-NEXT:    s_and_b32 s3, s3, 0xffe
-; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s4, s8, s4
+; GFX1250-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_cselect_b32 s4, 1, 0
-; GFX1250-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s3, s3, s4
+; GFX1250-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_cselect_b32 s4, 1, 0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_sub_co_i32 s8, 1, s2
 ; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s10, s3, 0x1000
@@ -3610,17 +3656,19 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
 ; GFX1250-GISEL-FAKE16-NEXT:    s_cmp_eq_u32 s2, 0x40f
 ; GFX1250-GISEL-FAKE16-NEXT:    s_cselect_b32 s2, s4, s3
 ; GFX1250-GISEL-FAKE16-NEXT:    s_lshr_b32 s3, s5, 16
+; GFX1250-GISEL-FAKE16-NEXT:    s_and_b32 s8, s7, 0x1ff
 ; GFX1250-GISEL-FAKE16-NEXT:    s_bfe_u32 s4, s7, 0xb0014
 ; GFX1250-GISEL-FAKE16-NEXT:    s_lshr_b32 s5, s7, 8
 ; GFX1250-GISEL-FAKE16-NEXT:    s_and_b32 s3, s3, 0x8000
-; GFX1250-GISEL-FAKE16-NEXT:    s_and_b32 s8, s7, 0x1ff
+; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s6, s8, s6
 ; GFX1250-GISEL-FAKE16-NEXT:    s_addk_co_i32 s4, 0xfc10
 ; GFX1250-GISEL-FAKE16-NEXT:    s_and_b32 s5, s5, 0xffe
 ; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s2, s3, s2
-; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s3, s8, s6
+; GFX1250-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s6, 0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_cselect_b32 s3, 1, 0
-; GFX1250-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-GISEL-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s3, s5, s3
+; GFX1250-GISEL-FAKE16-NEXT:    s_cmp_lg_u32 s3, 0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX1250-GISEL-FAKE16-NEXT:    s_sub_co_i32 s6, 1, s4
 ; GFX1250-GISEL-FAKE16-NEXT:    s_or_b32 s9, s3, 0x1000
diff --git a/llvm/test/CodeGen/AMDGPU/fptrunc.ll b/llvm/test/CodeGen/AMDGPU/fptrunc.ll
index 0deef8b..5d31177 100644
--- a/llvm/test/CodeGen/AMDGPU/fptrunc.ll
+++ b/llvm/test/CodeGen/AMDGPU/fptrunc.ll
@@ -182,6 +182,7 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
 ; SI-NEXT:    s_and_b32 s1, s7, 0x1ff
 ; SI-NEXT:    s_and_b32 s8, s0, 0xffe
 ; SI-NEXT:    s_or_b32 s0, s1, s6
+; SI-NEXT:    s_cmp_lg_u32 s0, 0
 ; SI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; SI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; SI-NEXT:    s_bfe_u32 s0, s7, 0xb0014
@@ -236,6 +237,7 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
 ; VI-SDAG-NEXT:    s_and_b32 s8, s4, 0xffe
 ; VI-SDAG-NEXT:    s_and_b32 s4, s7, 0x1ff
 ; VI-SDAG-NEXT:    s_or_b32 s4, s4, s6
+; VI-SDAG-NEXT:    s_cmp_lg_u32 s4, 0
 ; VI-SDAG-NEXT:    s_mov_b32 s1, s5
 ; VI-SDAG-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; VI-SDAG-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[4:5]
@@ -288,8 +290,10 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
 ; VI-GISEL-NEXT:    s_addk_i32 s4, 0xfc10
 ; VI-GISEL-NEXT:    s_and_b32 s5, s5, 0xffe
 ; VI-GISEL-NEXT:    s_or_b32 s2, s6, s2
+; VI-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; VI-GISEL-NEXT:    s_cselect_b32 s2, 1, 0
 ; VI-GISEL-NEXT:    s_or_b32 s2, s5, s2
+; VI-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; VI-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; VI-GISEL-NEXT:    s_sub_i32 s7, 1, s4
 ; VI-GISEL-NEXT:    s_lshl_b32 s6, s4, 12
@@ -331,10 +335,11 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
 ; GFX10-SDAG:       ; %bb.0:
 ; GFX10-SDAG-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
 ; GFX10-SDAG-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-SDAG-NEXT:    s_lshr_b32 s4, s3, 8
-; GFX10-SDAG-NEXT:    s_and_b32 s5, s3, 0x1ff
-; GFX10-SDAG-NEXT:    s_and_b32 s4, s4, 0xffe
-; GFX10-SDAG-NEXT:    s_or_b32 s2, s5, s2
+; GFX10-SDAG-NEXT:    s_and_b32 s4, s3, 0x1ff
+; GFX10-SDAG-NEXT:    s_lshr_b32 s5, s3, 8
+; GFX10-SDAG-NEXT:    s_or_b32 s2, s4, s2
+; GFX10-SDAG-NEXT:    s_and_b32 s4, s5, 0xffe
+; GFX10-SDAG-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX10-SDAG-NEXT:    s_cselect_b32 s2, -1, 0
 ; GFX10-SDAG-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s2
 ; GFX10-SDAG-NEXT:    s_bfe_u32 s2, s3, 0xb0014
@@ -382,14 +387,16 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
 ; GFX10-GISEL:       ; %bb.0:
 ; GFX10-GISEL-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
 ; GFX10-GISEL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-GISEL-NEXT:    s_and_b32 s6, s3, 0x1ff
 ; GFX10-GISEL-NEXT:    s_bfe_u32 s4, s3, 0xb0014
 ; GFX10-GISEL-NEXT:    s_lshr_b32 s5, s3, 8
-; GFX10-GISEL-NEXT:    s_and_b32 s6, s3, 0x1ff
+; GFX10-GISEL-NEXT:    s_or_b32 s2, s6, s2
 ; GFX10-GISEL-NEXT:    s_addk_i32 s4, 0xfc10
 ; GFX10-GISEL-NEXT:    s_and_b32 s5, s5, 0xffe
-; GFX10-GISEL-NEXT:    s_or_b32 s2, s6, s2
+; GFX10-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX10-GISEL-NEXT:    s_cselect_b32 s2, 1, 0
 ; GFX10-GISEL-NEXT:    s_or_b32 s2, s5, s2
+; GFX10-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX10-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX10-GISEL-NEXT:    s_sub_i32 s6, 1, s4
 ; GFX10-GISEL-NEXT:    s_or_b32 s8, s2, 0x1000
@@ -431,10 +438,11 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
 ; GFX11-SDAG:       ; %bb.0:
 ; GFX11-SDAG-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
 ; GFX11-SDAG-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-SDAG-NEXT:    s_lshr_b32 s4, s3, 8
-; GFX11-SDAG-NEXT:    s_and_b32 s5, s3, 0x1ff
-; GFX11-SDAG-NEXT:    s_and_b32 s4, s4, 0xffe
-; GFX11-SDAG-NEXT:    s_or_b32 s2, s5, s2
+; GFX11-SDAG-NEXT:    s_and_b32 s4, s3, 0x1ff
+; GFX11-SDAG-NEXT:    s_lshr_b32 s5, s3, 8
+; GFX11-SDAG-NEXT:    s_or_b32 s2, s4, s2
+; GFX11-SDAG-NEXT:    s_and_b32 s4, s5, 0xffe
+; GFX11-SDAG-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX11-SDAG-NEXT:    s_cselect_b32 s2, -1, 0
 ; GFX11-SDAG-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
 ; GFX11-SDAG-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s2
@@ -490,15 +498,17 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
 ; GFX11-GISEL:       ; %bb.0:
 ; GFX11-GISEL-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
 ; GFX11-GISEL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-GISEL-NEXT:    s_and_b32 s6, s3, 0x1ff
 ; GFX11-GISEL-NEXT:    s_bfe_u32 s4, s3, 0xb0014
 ; GFX11-GISEL-NEXT:    s_lshr_b32 s5, s3, 8
-; GFX11-GISEL-NEXT:    s_and_b32 s6, s3, 0x1ff
+; GFX11-GISEL-NEXT:    s_or_b32 s2, s6, s2
 ; GFX11-GISEL-NEXT:    s_addk_i32 s4, 0xfc10
 ; GFX11-GISEL-NEXT:    s_and_b32 s5, s5, 0xffe
-; GFX11-GISEL-NEXT:    s_or_b32 s2, s6, s2
+; GFX11-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX11-GISEL-NEXT:    s_cselect_b32 s2, 1, 0
-; GFX11-GISEL-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-GISEL-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX11-GISEL-NEXT:    s_or_b32 s2, s5, s2
+; GFX11-GISEL-NEXT:    s_cmp_lg_u32 s2, 0
 ; GFX11-GISEL-NEXT:    s_cselect_b32 s5, 1, 0
 ; GFX11-GISEL-NEXT:    s_sub_i32 s6, 1, s4
 ; GFX11-GISEL-NEXT:    s_or_b32 s8, s2, 0x1000
diff --git a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fadd.ll b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fadd.ll
index 31f277f..37756d1 100644
--- a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fadd.ll
+++ b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fadd.ll
@@ -472,6 +472,7 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -535,10 +536,11 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1064-NEXT:  .LBB1_1: ; %ComputeLoop
 ; GFX1064-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1064-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1064-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1064-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -604,6 +606,7 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -657,11 +660,12 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1164-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1164-NEXT:    v_add_f32_e32 v1, s3, v1
+; GFX1164-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT:    v_add_f32_e32 v1, s4, v1
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -706,8 +710,9 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1132-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_add_f32_e32 v1, s2, v1
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1685,6 +1690,7 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_one_as_scope
 ; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1748,10 +1754,11 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_one_as_scope
 ; GFX1064-NEXT:  .LBB3_1: ; %ComputeLoop
 ; GFX1064-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1064-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1064-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1064-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1817,6 +1824,7 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_one_as_scope
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1870,11 +1878,12 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_one_as_scope
 ; GFX1164-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1164-NEXT:    v_add_f32_e32 v1, s3, v1
+; GFX1164-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT:    v_add_f32_e32 v1, s4, v1
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1919,8 +1928,9 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_one_as_scope
 ; GFX1132-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_add_f32_e32 v1, s2, v1
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -2958,6 +2968,7 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3021,10 +3032,11 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1064-NEXT:  .LBB5_1: ; %ComputeLoop
 ; GFX1064-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1064-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1064-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1064-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3090,6 +3102,7 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3143,11 +3156,12 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1164-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1164-NEXT:    v_add_f32_e32 v1, s3, v1
+; GFX1164-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT:    v_add_f32_e32 v1, s4, v1
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3192,8 +3206,9 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1132-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_add_f32_e32 v1, s2, v1
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3727,6 +3742,7 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3790,10 +3806,11 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1064-NEXT:  .LBB6_1: ; %ComputeLoop
 ; GFX1064-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1064-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1064-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1064-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3859,6 +3876,7 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3912,11 +3930,12 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1164-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1164-NEXT:    v_add_f32_e32 v1, s3, v1
+; GFX1164-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT:    v_add_f32_e32 v1, s4, v1
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3961,8 +3980,9 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
 ; GFX1132-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_add_f32_e32 v1, s2, v1
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -4999,6 +5019,7 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_default_scop
 ; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -5062,10 +5083,11 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_default_scop
 ; GFX1064-NEXT:  .LBB8_1: ; %ComputeLoop
 ; GFX1064-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1064-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1064-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1064-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5131,6 +5153,7 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_default_scop
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -5184,11 +5207,12 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_default_scop
 ; GFX1164-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1164-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1164-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5246,8 +5270,9 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_default_scop
 ; GFX1132-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6259,6 +6284,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX9-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6328,6 +6354,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1064-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6397,6 +6424,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6457,6 +6485,8 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1164-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6520,6 +6550,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7686,6 +7717,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_one_a
 ; GFX9-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB12_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7755,6 +7787,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_one_a
 ; GFX1064-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB12_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7824,6 +7857,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_one_a
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB12_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7884,6 +7918,8 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_one_a
 ; GFX1164-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB12_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7947,6 +7983,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_one_a
 ; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB12_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9113,6 +9150,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX9-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9182,6 +9220,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1064-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9251,6 +9290,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9311,6 +9351,8 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1164-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9374,6 +9416,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10022,6 +10065,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX9-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10091,6 +10135,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1064-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10160,6 +10205,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10220,6 +10266,8 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1164-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10283,6 +10331,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
 ; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11449,6 +11498,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_defau
 ; GFX9-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11518,6 +11568,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_defau
 ; GFX1064-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11587,6 +11638,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_defau
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11647,6 +11699,8 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_defau
 ; GFX1164-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11710,6 +11764,7 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_defau
 ; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll
index 4581efc..6351bb3 100644
--- a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll
+++ b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll
@@ -381,12 +381,13 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_agent_scope_
 ; GFX9-NEXT:  .LBB1_1: ; %ComputeLoop
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX9-NEXT:    v_max_f32_e64 v1, s3, s3
-; GFX9-NEXT:    v_max_f32_e32 v2, v2, v2
+; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX9-NEXT:    v_max_f32_e32 v1, v2, v2
+; GFX9-NEXT:    v_max_f32_e64 v2, s4, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT:    v_max_f32_e32 v2, v2, v1
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    v_max_f32_e32 v2, v1, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -456,6 +457,7 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_agent_scope_
 ; GFX1064-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -511,6 +513,7 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_agent_scope_
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_max_f32_e64 v2, s2, s2
 ; GFX1032-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB1_1
@@ -559,7 +562,8 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_agent_scope_
 ; GFX1164-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -606,9 +610,11 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_agent_scope_
 ; GFX1132-NEXT:    v_max_f32_e32 v1, v1, v1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_max_f32_e64 v2, s2, s2
+; GFX1132-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1132-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1414,12 +1420,13 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_one_as_scope
 ; GFX9-NEXT:  .LBB3_1: ; %ComputeLoop
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX9-NEXT:    v_max_f32_e64 v1, s3, s3
-; GFX9-NEXT:    v_max_f32_e32 v2, v2, v2
+; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX9-NEXT:    v_max_f32_e32 v1, v2, v2
+; GFX9-NEXT:    v_max_f32_e64 v2, s4, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT:    v_max_f32_e32 v2, v2, v1
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    v_max_f32_e32 v2, v1, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1489,6 +1496,7 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_one_as_scope
 ; GFX1064-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1544,6 +1552,7 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_one_as_scope
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_max_f32_e64 v2, s2, s2
 ; GFX1032-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB3_1
@@ -1592,7 +1601,8 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_one_as_scope
 ; GFX1164-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1639,9 +1649,11 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_one_as_scope
 ; GFX1132-NEXT:    v_max_f32_e32 v1, v1, v1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_max_f32_e64 v2, s2, s2
+; GFX1132-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1132-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -2447,12 +2459,13 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_default_scop
 ; GFX9-NEXT:  .LBB5_1: ; %ComputeLoop
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX9-NEXT:    v_max_f32_e64 v1, s3, s3
-; GFX9-NEXT:    v_max_f32_e32 v2, v2, v2
+; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX9-NEXT:    v_max_f32_e32 v1, v2, v2
+; GFX9-NEXT:    v_max_f32_e64 v2, s4, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT:    v_max_f32_e32 v2, v2, v1
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    v_max_f32_e32 v2, v1, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2522,6 +2535,7 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_default_scop
 ; GFX1064-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -2577,6 +2591,7 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_default_scop
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_max_f32_e64 v2, s2, s2
 ; GFX1032-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB5_1
@@ -2625,7 +2640,8 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_default_scop
 ; GFX1164-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -2672,9 +2688,11 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_default_scop
 ; GFX1132-NEXT:    v_max_f32_e32 v1, v1, v1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_max_f32_e64 v2, s2, s2
+; GFX1132-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1132-NEXT:    v_max_f32_e32 v1, v1, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3573,6 +3591,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_agent
 ; GFX9-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_max_f64 v[4:5], v[2:3], v[4:5]
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3646,6 +3665,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_agent
 ; GFX1064-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_max_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3704,6 +3724,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_agent
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_max_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3753,7 +3774,8 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_agent
 ; GFX1164-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_max_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3819,9 +3841,10 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_agent
 ; GFX1132-NEXT:    v_readlane_b32 s3, v1, s1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_max_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -4836,6 +4859,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_one_a
 ; GFX9-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_max_f64 v[4:5], v[2:3], v[4:5]
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB9_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -4909,6 +4933,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_one_a
 ; GFX1064-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_max_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB9_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -4967,6 +4992,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_one_a
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_max_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB9_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -5016,7 +5042,8 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_one_a
 ; GFX1164-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_max_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB9_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -5082,9 +5109,10 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_one_a
 ; GFX1132-NEXT:    v_readlane_b32 s3, v1, s1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_max_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB9_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6099,6 +6127,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_defau
 ; GFX9-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_max_f64 v[4:5], v[2:3], v[4:5]
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6172,6 +6201,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_defau
 ; GFX1064-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_max_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6230,6 +6260,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_defau
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_max_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6279,7 +6310,8 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_defau
 ; GFX1164-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_max_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6345,9 +6377,10 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_defau
 ; GFX1132-NEXT:    v_readlane_b32 s3, v1, s1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_max_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
diff --git a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll
index bd570d9..a9ac008 100644
--- a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll
+++ b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll
@@ -381,12 +381,13 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_agent_scope_
 ; GFX9-NEXT:  .LBB1_1: ; %ComputeLoop
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX9-NEXT:    v_max_f32_e64 v1, s3, s3
-; GFX9-NEXT:    v_max_f32_e32 v2, v2, v2
+; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX9-NEXT:    v_max_f32_e32 v1, v2, v2
+; GFX9-NEXT:    v_max_f32_e64 v2, s4, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT:    v_min_f32_e32 v2, v2, v1
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    v_min_f32_e32 v2, v1, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -456,6 +457,7 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_agent_scope_
 ; GFX1064-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -511,6 +513,7 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_agent_scope_
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_max_f32_e64 v2, s2, s2
 ; GFX1032-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB1_1
@@ -559,7 +562,8 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_agent_scope_
 ; GFX1164-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -606,9 +610,11 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_agent_scope_
 ; GFX1132-NEXT:    v_max_f32_e32 v1, v1, v1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_max_f32_e64 v2, s2, s2
+; GFX1132-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1132-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1414,12 +1420,13 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_one_as_scope
 ; GFX9-NEXT:  .LBB3_1: ; %ComputeLoop
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX9-NEXT:    v_max_f32_e64 v1, s3, s3
-; GFX9-NEXT:    v_max_f32_e32 v2, v2, v2
+; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX9-NEXT:    v_max_f32_e32 v1, v2, v2
+; GFX9-NEXT:    v_max_f32_e64 v2, s4, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT:    v_min_f32_e32 v2, v2, v1
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    v_min_f32_e32 v2, v1, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1489,6 +1496,7 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_one_as_scope
 ; GFX1064-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1544,6 +1552,7 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_one_as_scope
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_max_f32_e64 v2, s2, s2
 ; GFX1032-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB3_1
@@ -1592,7 +1601,8 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_one_as_scope
 ; GFX1164-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1639,9 +1649,11 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_one_as_scope
 ; GFX1132-NEXT:    v_max_f32_e32 v1, v1, v1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_max_f32_e64 v2, s2, s2
+; GFX1132-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1132-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -2447,12 +2459,13 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_default_scop
 ; GFX9-NEXT:  .LBB5_1: ; %ComputeLoop
 ; GFX9-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX9-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX9-NEXT:    v_max_f32_e64 v1, s3, s3
-; GFX9-NEXT:    v_max_f32_e32 v2, v2, v2
+; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX9-NEXT:    v_max_f32_e32 v1, v2, v2
+; GFX9-NEXT:    v_max_f32_e64 v2, s4, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT:    v_min_f32_e32 v2, v2, v1
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    v_min_f32_e32 v2, v1, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2522,6 +2535,7 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_default_scop
 ; GFX1064-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -2577,6 +2591,7 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_default_scop
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_max_f32_e64 v2, s2, s2
 ; GFX1032-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB5_1
@@ -2625,7 +2640,8 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_default_scop
 ; GFX1164-NEXT:    v_max_f32_e64 v2, s3, s3
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -2672,9 +2688,11 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_default_scop
 ; GFX1132-NEXT:    v_max_f32_e32 v1, v1, v1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_max_f32_e64 v2, s2, s2
+; GFX1132-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1132-NEXT:    v_min_f32_e32 v1, v1, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3573,6 +3591,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_agent
 ; GFX9-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_min_f64 v[4:5], v[2:3], v[4:5]
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3646,6 +3665,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_agent
 ; GFX1064-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_min_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3704,6 +3724,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_agent
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_min_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3753,7 +3774,8 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_agent
 ; GFX1164-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_min_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3819,9 +3841,10 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_agent
 ; GFX1132-NEXT:    v_readlane_b32 s3, v1, s1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_min_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB7_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -4836,6 +4859,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_one_a
 ; GFX9-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_min_f64 v[4:5], v[2:3], v[4:5]
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB9_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -4909,6 +4933,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_one_a
 ; GFX1064-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_min_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB9_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -4967,6 +4992,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_one_a
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_min_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB9_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -5016,7 +5042,8 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_one_a
 ; GFX1164-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_min_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB9_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -5082,9 +5109,10 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_one_a
 ; GFX1132-NEXT:    v_readlane_b32 s3, v1, s1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_min_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB9_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6099,6 +6127,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_defau
 ; GFX9-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_min_f64 v[4:5], v[2:3], v[4:5]
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6172,6 +6201,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_defau
 ; GFX1064-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    v_min_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6230,6 +6260,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_defau
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_min_f64 v[2:3], v[2:3], v[4:5]
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6279,7 +6310,8 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_defau
 ; GFX1164-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    v_min_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6345,9 +6377,10 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_defau
 ; GFX1132-NEXT:    v_readlane_b32 s3, v1, s1
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_max_f64 v[4:5], s[2:3], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_min_f64 v[4:5], v[2:3], v[4:5]
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB11_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
diff --git a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fsub.ll b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fsub.ll
index 1f2d70c..6311143 100644
--- a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fsub.ll
+++ b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fsub.ll
@@ -532,6 +532,7 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -595,10 +596,11 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1064-NEXT:  .LBB1_1: ; %ComputeLoop
 ; GFX1064-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1064-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1064-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1064-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -664,6 +666,7 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -717,11 +720,12 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1164-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1164-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1164-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -779,8 +783,9 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1132-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB1_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1857,6 +1862,7 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_one_as_scope
 ; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -1920,10 +1926,11 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_one_as_scope
 ; GFX1064-NEXT:  .LBB3_1: ; %ComputeLoop
 ; GFX1064-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1064-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1064-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1064-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1989,6 +1996,7 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_one_as_scope
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -2042,11 +2050,12 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_one_as_scope
 ; GFX1164-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1164-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1164-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2104,8 +2113,9 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_one_as_scope
 ; GFX1132-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB3_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3182,6 +3192,7 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3245,10 +3256,11 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1064-NEXT:  .LBB5_1: ; %ComputeLoop
 ; GFX1064-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1064-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1064-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1064-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3314,6 +3326,7 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -3367,11 +3380,12 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1164-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1164-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1164-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3429,8 +3443,9 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1132-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB5_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -4003,6 +4018,7 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -4066,10 +4082,11 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1064-NEXT:  .LBB6_1: ; %ComputeLoop
 ; GFX1064-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1064-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1064-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1064-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4135,6 +4152,7 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -4188,11 +4206,12 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1164-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1164-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1164-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4250,8 +4269,9 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
 ; GFX1132-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB6_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -5327,6 +5347,7 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_default_scop
 ; GFX9-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -5390,10 +5411,11 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_default_scop
 ; GFX1064-NEXT:  .LBB8_1: ; %ComputeLoop
 ; GFX1064-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1064-NEXT:    s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1064-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1064-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s2
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1064-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5459,6 +5481,7 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_default_scop
 ; GFX1032-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -5512,11 +5535,12 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_default_scop
 ; GFX1164-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT:    v_readlane_b32 s3, v0, s2
-; GFX1164-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1164-NEXT:    v_add_f32_e32 v2, s3, v2
+; GFX1164-NEXT:    v_readlane_b32 s4, v0, s2
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s2
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT:    v_add_f32_e32 v2, s4, v2
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5574,8 +5598,9 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_default_scop
 ; GFX1132-NEXT:    s_ctz_i32_b32 s1, s0
 ; GFX1132-NEXT:    v_readlane_b32 s2, v0, s1
 ; GFX1132-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    v_add_f32_e32 v2, s2, v2
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB8_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
@@ -6587,6 +6612,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX9-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6656,6 +6682,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1064-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6725,6 +6752,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6785,6 +6813,8 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1164-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6848,6 +6878,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB10_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8013,6 +8044,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_one_a
 ; GFX9-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB12_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8082,6 +8114,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_one_a
 ; GFX1064-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB12_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8151,6 +8184,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_one_a
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB12_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8211,6 +8245,8 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_one_a
 ; GFX1164-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB12_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8274,6 +8310,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_one_a
 ; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB12_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9440,6 +9477,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX9-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9509,6 +9547,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1064-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9578,6 +9617,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9638,6 +9678,8 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1164-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9701,6 +9743,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB14_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10349,6 +10392,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX9-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10418,6 +10462,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1064-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10487,6 +10532,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10547,6 +10593,8 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1164-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10610,6 +10658,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
 ; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB15_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11775,6 +11824,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_defau
 ; GFX9-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX9-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX9-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX9-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX9-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11844,6 +11894,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_defau
 ; GFX1064-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1064-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1064-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; GFX1064-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1064-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1064-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11913,6 +11964,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_defau
 ; GFX1032-NEXT:    s_lshl_b32 s1, 1, s1
 ; GFX1032-NEXT:    s_andn2_b32 s0, s0, s1
 ; GFX1032-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1032-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1032-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1032-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11973,6 +12025,8 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_defau
 ; GFX1164-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
 ; GFX1164-NEXT:    s_lshl_b64 s[2:3], 1, s4
 ; GFX1164-NEXT:    s_and_not1_b64 s[0:1], s[0:1], s[2:3]
+; GFX1164-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; GFX1164-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1164-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1164-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12036,6 +12090,7 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_defau
 ; GFX1132-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1132-NEXT:    s_and_not1_b32 s0, s0, s1
 ; GFX1132-NEXT:    v_add_f64 v[4:5], v[4:5], s[2:3]
+; GFX1132-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX1132-NEXT:    s_cbranch_scc1 .LBB17_1
 ; GFX1132-NEXT:  ; %bb.2: ; %ComputeEnd
 ; GFX1132-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/insert-delay-alu-bug.ll b/llvm/test/CodeGen/AMDGPU/insert-delay-alu-bug.ll
index c3f3917..eee232a 100644
--- a/llvm/test/CodeGen/AMDGPU/insert-delay-alu-bug.ll
+++ b/llvm/test/CodeGen/AMDGPU/insert-delay-alu-bug.ll
@@ -136,17 +136,19 @@ define amdgpu_kernel void @f2(i32 %arg, i32 %arg1, i32 %arg2, i1 %arg3, i32 %arg
 ; GFX11-NEXT:  .LBB2_6: ; %bb18
 ; GFX11-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX11-NEXT:    s_and_b32 s1, 0xffff, s1
-; GFX11-NEXT:    s_cselect_b32 s13, -1, 0
-; GFX11-NEXT:    v_readfirstlane_b32 s1, v0
-; GFX11-NEXT:    v_cndmask_b32_e64 v2, 0, 1, s13
-; GFX11-NEXT:    s_and_b32 s13, s8, s13
-; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX11-NEXT:    s_and_b32 s13, s13, exec_lo
+; GFX11-NEXT:    v_readfirstlane_b32 s13, v0
+; GFX11-NEXT:    s_cmp_lg_u32 s1, 0
+; GFX11-NEXT:    s_cselect_b32 s1, -1, 0
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX11-NEXT:    v_cndmask_b32_e64 v2, 0, 1, s1
+; GFX11-NEXT:    s_and_b32 s1, s8, s1
+; GFX11-NEXT:    s_and_b32 s1, s1, exec_lo
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX11-NEXT:    v_readfirstlane_b32 s19, v2
-; GFX11-NEXT:    s_cselect_b32 s1, s19, s1
-; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
-; GFX11-NEXT:    s_and_b32 s1, s1, 1
+; GFX11-NEXT:    s_cselect_b32 s1, s19, s13
 ; GFX11-NEXT:    s_and_b32 s13, 0xffff, s0
+; GFX11-NEXT:    s_and_b32 s1, s1, 1
+; GFX11-NEXT:    s_cmp_lg_u32 s13, 0
 ; GFX11-NEXT:    s_cselect_b32 s13, -1, 0
 ; GFX11-NEXT:    s_and_b32 s20, s9, exec_lo
 ; GFX11-NEXT:    v_cndmask_b32_e64 v2, 0, 1, s13
diff --git a/llvm/test/CodeGen/AMDGPU/local-atomicrmw-fadd.ll b/llvm/test/CodeGen/AMDGPU/local-atomicrmw-fadd.ll
index 6dc9199..8748aff 100644
--- a/llvm/test/CodeGen/AMDGPU/local-atomicrmw-fadd.ll
+++ b/llvm/test/CodeGen/AMDGPU/local-atomicrmw-fadd.ll
@@ -8265,10 +8265,12 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
 ; GFX12-NEXT:    s_wait_alu 0xfffe
 ; GFX12-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX12-NEXT:    v_readlane_b32 s6, v1, s3
+; GFX12-NEXT:    s_lshl_b32 s7, 1, s3
 ; GFX12-NEXT:    v_writelane_b32 v0, s0, s3
-; GFX12-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX12-NEXT:    s_wait_alu 0xfffe
-; GFX12-NEXT:    s_and_not1_b32 s1, s1, s3
+; GFX12-NEXT:    s_and_not1_b32 s1, s1, s7
+; GFX12-NEXT:    s_wait_alu 0xfffe
+; GFX12-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX12-NEXT:    s_add_f32 s0, s0, s6
 ; GFX12-NEXT:    s_cbranch_scc1 .LBB28_5
 ; GFX12-NEXT:  ; %bb.6: ; %ComputeEnd
@@ -8349,13 +8351,14 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
 ; GFX942-NEXT:  .LBB28_5: ; %ComputeLoop
 ; GFX942-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX942-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX942-NEXT:    v_readfirstlane_b32 s6, v1
-; GFX942-NEXT:    s_mov_b32 m0, s3
-; GFX942-NEXT:    v_readlane_b32 s8, v2, s3
-; GFX942-NEXT:    v_writelane_b32 v0, s6, m0
 ; GFX942-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX942-NEXT:    v_readfirstlane_b32 s8, v1
+; GFX942-NEXT:    v_readlane_b32 s9, v2, s3
+; GFX942-NEXT:    s_mov_b32 m0, s3
 ; GFX942-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX942-NEXT:    v_add_f32_e32 v1, s8, v1
+; GFX942-NEXT:    v_writelane_b32 v0, s8, m0
+; GFX942-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX942-NEXT:    v_add_f32_e32 v1, s9, v1
 ; GFX942-NEXT:    s_cbranch_scc1 .LBB28_5
 ; GFX942-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX942-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8437,14 +8440,15 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
 ; GFX11-NEXT:  .LBB28_5: ; %ComputeLoop
 ; GFX11-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX11-NEXT:    s_ctz_i32_b32 s1, s0
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
 ; GFX11-NEXT:    v_readfirstlane_b32 s3, v1
-; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
 ; GFX11-NEXT:    v_readlane_b32 s6, v2, s1
+; GFX11-NEXT:    s_lshl_b32 s7, 1, s1
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-NEXT:    s_and_not1_b32 s0, s0, s7
 ; GFX11-NEXT:    v_writelane_b32 v0, s3, s1
-; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
 ; GFX11-NEXT:    v_add_f32_e32 v1, s6, v1
-; GFX11-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX11-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX11-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX11-NEXT:    s_cbranch_scc1 .LBB28_5
 ; GFX11-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX11-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8524,10 +8528,11 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
 ; GFX10-NEXT:    s_ff1_i32_b32 s1, s0
 ; GFX10-NEXT:    v_readfirstlane_b32 s3, v1
 ; GFX10-NEXT:    v_readlane_b32 s6, v2, s1
+; GFX10-NEXT:    s_lshl_b32 s7, 1, s1
+; GFX10-NEXT:    s_andn2_b32 s0, s0, s7
 ; GFX10-NEXT:    v_writelane_b32 v0, s3, s1
 ; GFX10-NEXT:    v_add_f32_e32 v1, s6, v1
-; GFX10-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX10-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX10-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX10-NEXT:    s_cbranch_scc1 .LBB28_5
 ; GFX10-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX10-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8604,13 +8609,14 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
 ; GFX90A-NEXT:  .LBB28_5: ; %ComputeLoop
 ; GFX90A-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX90A-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX90A-NEXT:    v_readfirstlane_b32 s6, v1
-; GFX90A-NEXT:    s_mov_b32 m0, s3
-; GFX90A-NEXT:    v_readlane_b32 s8, v2, s3
-; GFX90A-NEXT:    v_writelane_b32 v0, s6, m0
 ; GFX90A-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX90A-NEXT:    v_readfirstlane_b32 s8, v1
+; GFX90A-NEXT:    v_readlane_b32 s9, v2, s3
+; GFX90A-NEXT:    s_mov_b32 m0, s3
 ; GFX90A-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX90A-NEXT:    v_add_f32_e32 v1, s8, v1
+; GFX90A-NEXT:    v_writelane_b32 v0, s8, m0
+; GFX90A-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX90A-NEXT:    v_add_f32_e32 v1, s9, v1
 ; GFX90A-NEXT:    s_cbranch_scc1 .LBB28_5
 ; GFX90A-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX90A-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8686,13 +8692,14 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
 ; GFX908-NEXT:  .LBB28_5: ; %ComputeLoop
 ; GFX908-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX908-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX908-NEXT:    v_readfirstlane_b32 s6, v1
-; GFX908-NEXT:    s_mov_b32 m0, s3
-; GFX908-NEXT:    v_readlane_b32 s8, v2, s3
-; GFX908-NEXT:    v_writelane_b32 v0, s6, m0
 ; GFX908-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX908-NEXT:    v_readfirstlane_b32 s8, v1
+; GFX908-NEXT:    v_readlane_b32 s9, v2, s3
+; GFX908-NEXT:    s_mov_b32 m0, s3
 ; GFX908-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX908-NEXT:    v_add_f32_e32 v1, s8, v1
+; GFX908-NEXT:    v_writelane_b32 v0, s8, m0
+; GFX908-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX908-NEXT:    v_add_f32_e32 v1, s9, v1
 ; GFX908-NEXT:    s_cbranch_scc1 .LBB28_5
 ; GFX908-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX908-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8769,13 +8776,14 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
 ; GFX8-NEXT:  .LBB28_5: ; %ComputeLoop
 ; GFX8-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX8-NEXT:    v_readfirstlane_b32 s6, v1
-; GFX8-NEXT:    s_mov_b32 m0, s3
-; GFX8-NEXT:    v_readlane_b32 s8, v2, s3
-; GFX8-NEXT:    v_writelane_b32 v0, s6, m0
 ; GFX8-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT:    v_readfirstlane_b32 s8, v1
+; GFX8-NEXT:    v_readlane_b32 s9, v2, s3
+; GFX8-NEXT:    s_mov_b32 m0, s3
 ; GFX8-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT:    v_add_f32_e32 v1, s8, v1
+; GFX8-NEXT:    v_writelane_b32 v0, s8, m0
+; GFX8-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX8-NEXT:    v_add_f32_e32 v1, s9, v1
 ; GFX8-NEXT:    s_cbranch_scc1 .LBB28_5
 ; GFX8-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9122,10 +9130,12 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
 ; GFX12-NEXT:    s_wait_alu 0xfffe
 ; GFX12-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX12-NEXT:    v_readlane_b32 s6, v1, s3
+; GFX12-NEXT:    s_lshl_b32 s7, 1, s3
 ; GFX12-NEXT:    v_writelane_b32 v0, s0, s3
-; GFX12-NEXT:    s_lshl_b32 s3, 1, s3
 ; GFX12-NEXT:    s_wait_alu 0xfffe
-; GFX12-NEXT:    s_and_not1_b32 s1, s1, s3
+; GFX12-NEXT:    s_and_not1_b32 s1, s1, s7
+; GFX12-NEXT:    s_wait_alu 0xfffe
+; GFX12-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX12-NEXT:    s_add_f32 s0, s0, s6
 ; GFX12-NEXT:    s_cbranch_scc1 .LBB29_5
 ; GFX12-NEXT:  ; %bb.6: ; %ComputeEnd
@@ -9202,13 +9212,14 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
 ; GFX942-NEXT:  .LBB29_5: ; %ComputeLoop
 ; GFX942-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX942-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX942-NEXT:    v_readfirstlane_b32 s6, v1
-; GFX942-NEXT:    s_mov_b32 m0, s3
-; GFX942-NEXT:    v_readlane_b32 s8, v2, s3
-; GFX942-NEXT:    v_writelane_b32 v0, s6, m0
 ; GFX942-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX942-NEXT:    v_readfirstlane_b32 s8, v1
+; GFX942-NEXT:    v_readlane_b32 s9, v2, s3
+; GFX942-NEXT:    s_mov_b32 m0, s3
 ; GFX942-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX942-NEXT:    v_add_f32_e32 v1, s8, v1
+; GFX942-NEXT:    v_writelane_b32 v0, s8, m0
+; GFX942-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX942-NEXT:    v_add_f32_e32 v1, s9, v1
 ; GFX942-NEXT:    s_cbranch_scc1 .LBB29_5
 ; GFX942-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX942-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9285,14 +9296,15 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
 ; GFX11-NEXT:  .LBB29_5: ; %ComputeLoop
 ; GFX11-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX11-NEXT:    s_ctz_i32_b32 s1, s0
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
 ; GFX11-NEXT:    v_readfirstlane_b32 s3, v1
-; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
 ; GFX11-NEXT:    v_readlane_b32 s6, v2, s1
+; GFX11-NEXT:    s_lshl_b32 s7, 1, s1
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX11-NEXT:    s_and_not1_b32 s0, s0, s7
 ; GFX11-NEXT:    v_writelane_b32 v0, s3, s1
-; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
 ; GFX11-NEXT:    v_add_f32_e32 v1, s6, v1
-; GFX11-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX11-NEXT:    s_and_not1_b32 s0, s0, s1
+; GFX11-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX11-NEXT:    s_cbranch_scc1 .LBB29_5
 ; GFX11-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX11-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9365,10 +9377,11 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
 ; GFX10-NEXT:    s_ff1_i32_b32 s1, s0
 ; GFX10-NEXT:    v_readfirstlane_b32 s3, v1
 ; GFX10-NEXT:    v_readlane_b32 s6, v2, s1
+; GFX10-NEXT:    s_lshl_b32 s7, 1, s1
+; GFX10-NEXT:    s_andn2_b32 s0, s0, s7
 ; GFX10-NEXT:    v_writelane_b32 v0, s3, s1
 ; GFX10-NEXT:    v_add_f32_e32 v1, s6, v1
-; GFX10-NEXT:    s_lshl_b32 s1, 1, s1
-; GFX10-NEXT:    s_andn2_b32 s0, s0, s1
+; GFX10-NEXT:    s_cmp_lg_u32 s0, 0
 ; GFX10-NEXT:    s_cbranch_scc1 .LBB29_5
 ; GFX10-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX10-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9440,13 +9453,14 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
 ; GFX90A-NEXT:  .LBB29_5: ; %ComputeLoop
 ; GFX90A-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX90A-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX90A-NEXT:    v_readfirstlane_b32 s6, v1
-; GFX90A-NEXT:    s_mov_b32 m0, s3
-; GFX90A-NEXT:    v_readlane_b32 s8, v2, s3
-; GFX90A-NEXT:    v_writelane_b32 v0, s6, m0
 ; GFX90A-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX90A-NEXT:    v_readfirstlane_b32 s8, v1
+; GFX90A-NEXT:    v_readlane_b32 s9, v2, s3
+; GFX90A-NEXT:    s_mov_b32 m0, s3
 ; GFX90A-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX90A-NEXT:    v_add_f32_e32 v1, s8, v1
+; GFX90A-NEXT:    v_writelane_b32 v0, s8, m0
+; GFX90A-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX90A-NEXT:    v_add_f32_e32 v1, s9, v1
 ; GFX90A-NEXT:    s_cbranch_scc1 .LBB29_5
 ; GFX90A-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX90A-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9519,13 +9533,14 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
 ; GFX908-NEXT:  .LBB29_5: ; %ComputeLoop
 ; GFX908-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX908-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX908-NEXT:    v_readfirstlane_b32 s6, v1
-; GFX908-NEXT:    s_mov_b32 m0, s3
-; GFX908-NEXT:    v_readlane_b32 s8, v2, s3
-; GFX908-NEXT:    v_writelane_b32 v0, s6, m0
 ; GFX908-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX908-NEXT:    v_readfirstlane_b32 s8, v1
+; GFX908-NEXT:    v_readlane_b32 s9, v2, s3
+; GFX908-NEXT:    s_mov_b32 m0, s3
 ; GFX908-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX908-NEXT:    v_add_f32_e32 v1, s8, v1
+; GFX908-NEXT:    v_writelane_b32 v0, s8, m0
+; GFX908-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX908-NEXT:    v_add_f32_e32 v1, s9, v1
 ; GFX908-NEXT:    s_cbranch_scc1 .LBB29_5
 ; GFX908-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX908-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9599,13 +9614,14 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
 ; GFX8-NEXT:  .LBB29_5: ; %ComputeLoop
 ; GFX8-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX8-NEXT:    s_ff1_i32_b64 s3, s[0:1]
-; GFX8-NEXT:    v_readfirstlane_b32 s6, v1
-; GFX8-NEXT:    s_mov_b32 m0, s3
-; GFX8-NEXT:    v_readlane_b32 s8, v2, s3
-; GFX8-NEXT:    v_writelane_b32 v0, s6, m0
 ; GFX8-NEXT:    s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT:    v_readfirstlane_b32 s8, v1
+; GFX8-NEXT:    v_readlane_b32 s9, v2, s3
+; GFX8-NEXT:    s_mov_b32 m0, s3
 ; GFX8-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT:    v_add_f32_e32 v1, s8, v1
+; GFX8-NEXT:    v_writelane_b32 v0, s8, m0
+; GFX8-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX8-NEXT:    v_add_f32_e32 v1, s9, v1
 ; GFX8-NEXT:    s_cbranch_scc1 .LBB29_5
 ; GFX8-NEXT:  ; %bb.6: ; %ComputeEnd
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v2, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/optimize-compare.mir b/llvm/test/CodeGen/AMDGPU/optimize-compare.mir
index fba42c4..c1cf06e 100644
--- a/llvm/test/CodeGen/AMDGPU/optimize-compare.mir
+++ b/llvm/test/CodeGen/AMDGPU/optimize-compare.mir
@@ -388,8 +388,9 @@ body:             |
   ; GCN-NEXT:   liveins: $sgpr0, $vgpr0_vgpr1
   ; GCN-NEXT: {{  $}}
   ; GCN-NEXT:   [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
-  ; GCN-NEXT:   S_BITCMP1_B32 killed [[COPY]], 0, implicit-def $scc
-  ; GCN-NEXT:   S_NOP 0, implicit $scc
+  ; GCN-NEXT:   [[S_AND_B32_:%[0-9]+]]:sreg_32 = S_AND_B32 1, killed [[COPY]], implicit-def $scc
+  ; GCN-NEXT:   S_NOP 0, implicit killed $scc
+  ; GCN-NEXT:   S_CMP_EQ_U32 killed [[S_AND_B32_]], 1, implicit-def $scc
   ; GCN-NEXT:   S_CBRANCH_SCC0 %bb.2, implicit $scc
   ; GCN-NEXT:   S_BRANCH %bb.1
   ; GCN-NEXT: {{  $}}
@@ -416,80 +417,6 @@ body:             |
     S_ENDPGM 0
 
 ...
----
-name:            xor_1_cmp_lg_0_killed_scc
-body:             |
-  ; GCN-LABEL: name: xor_1_cmp_lg_0_killed_scc
-  ; GCN: bb.0:
-  ; GCN-NEXT:   successors: %bb.1(0x40000000), %bb.2(0x40000000)
-  ; GCN-NEXT:   liveins: $sgpr0, $vgpr0_vgpr1
-  ; GCN-NEXT: {{  $}}
-  ; GCN-NEXT:   [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
-  ; GCN-NEXT:   [[S_XOR_B32_:%[0-9]+]]:sreg_32 = S_XOR_B32 1, killed [[COPY]], implicit-def $scc
-  ; GCN-NEXT:   S_NOP 0, implicit $scc
-  ; GCN-NEXT:   S_CBRANCH_SCC0 %bb.2, implicit $scc
-  ; GCN-NEXT:   S_BRANCH %bb.1
-  ; GCN-NEXT: {{  $}}
-  ; GCN-NEXT: bb.1:
-  ; GCN-NEXT:   successors: %bb.2(0x80000000)
-  ; GCN-NEXT: {{  $}}
-  ; GCN-NEXT: bb.2:
-  ; GCN-NEXT:   S_ENDPGM 0
-  bb.0:
-    successors: %bb.1(0x40000000), %bb.2(0x40000000)
-    liveins: $sgpr0, $vgpr0_vgpr1
-
-    %0:sreg_32 = COPY $sgpr0
-    %1:sreg_32 = S_XOR_B32 1, killed %0, implicit-def $scc
-    S_NOP 0, implicit killed $scc
-    S_CMP_LG_U32 killed %1:sreg_32, 0, implicit-def $scc
-    S_CBRANCH_SCC0 %bb.2, implicit $scc
-    S_BRANCH %bb.1
-
-  bb.1:
-    successors: %bb.2(0x80000000)
-
-  bb.2:
-    S_ENDPGM 0
-
-...
----
-name:            absdiff_1_cmp_lg_0_killed_scc
-body:             |
-  ; GCN-LABEL: name: absdiff_1_cmp_lg_0_killed_scc
-  ; GCN: bb.0:
-  ; GCN-NEXT:   successors: %bb.1(0x40000000), %bb.2(0x40000000)
-  ; GCN-NEXT:   liveins: $sgpr0, $vgpr0_vgpr1
-  ; GCN-NEXT: {{  $}}
-  ; GCN-NEXT:   [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
-  ; GCN-NEXT:   [[S_ABSDIFF_I32_:%[0-9]+]]:sreg_32 = S_ABSDIFF_I32 1, killed [[COPY]], implicit-def $scc
-  ; GCN-NEXT:   S_NOP 0, implicit $scc
-  ; GCN-NEXT:   S_CBRANCH_SCC0 %bb.2, implicit $scc
-  ; GCN-NEXT:   S_BRANCH %bb.1
-  ; GCN-NEXT: {{  $}}
-  ; GCN-NEXT: bb.1:
-  ; GCN-NEXT:   successors: %bb.2(0x80000000)
-  ; GCN-NEXT: {{  $}}
-  ; GCN-NEXT: bb.2:
-  ; GCN-NEXT:   S_ENDPGM 0
-  bb.0:
-    successors: %bb.1(0x40000000), %bb.2(0x40000000)
-    liveins: $sgpr0, $vgpr0_vgpr1
-
-    %0:sreg_32 = COPY $sgpr0
-    %1:sreg_32 = S_ABSDIFF_I32 1, killed %0, implicit-def $scc
-    S_NOP 0, implicit killed $scc
-    S_CMP_LG_U32 killed %1:sreg_32, 0, implicit-def $scc
-    S_CBRANCH_SCC0 %bb.2, implicit $scc
-    S_BRANCH %bb.1
-
-  bb.1:
-    successors: %bb.2(0x80000000)
-
-  bb.2:
-    S_ENDPGM 0
-
-...
 
 ---
 name:            and_1_cmp_eq_1_clobbered_scc
@@ -2143,7 +2070,8 @@ body:             |
   ; GCN-NEXT:   liveins: $sgpr0, $vgpr0_vgpr1
   ; GCN-NEXT: {{  $}}
   ; GCN-NEXT:   [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
-  ; GCN-NEXT:   [[S_AND_B32_:%[0-9]+]]:sreg_32 = S_AND_B32 3, killed [[COPY]], implicit-def $scc
+  ; GCN-NEXT:   [[S_AND_B32_:%[0-9]+]]:sreg_32 = S_AND_B32 3, killed [[COPY]], implicit-def dead $scc
+  ; GCN-NEXT:   S_CMP_LG_U32 killed [[S_AND_B32_]], 0, implicit-def $scc
   ; GCN-NEXT:   S_CBRANCH_SCC0 %bb.2, implicit $scc
   ; GCN-NEXT:   S_BRANCH %bb.1
   ; GCN-NEXT: {{  $}}
diff --git a/llvm/test/CodeGen/AMDGPU/s_cmp_0.ll b/llvm/test/CodeGen/AMDGPU/s_cmp_0.ll
index 19cc7f7..f53aaaa 100644
--- a/llvm/test/CodeGen/AMDGPU/s_cmp_0.ll
+++ b/llvm/test/CodeGen/AMDGPU/s_cmp_0.ll
@@ -10,6 +10,7 @@ define amdgpu_ps i32 @shl32(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: shl32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_lshl_b32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -24,6 +25,7 @@ define amdgpu_ps i32 @shl64(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: shl64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_lshl_b64 s[0:1], s[0:1], s2
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -38,6 +40,7 @@ define amdgpu_ps i32 @lshr32(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: lshr32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_lshr_b32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -52,6 +55,7 @@ define amdgpu_ps i32 @lshr64(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: lshr64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_lshr_b64 s[0:1], s[0:1], s2
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -66,6 +70,7 @@ define amdgpu_ps i32 @ashr32(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: ashr32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_ashr_i32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -80,6 +85,7 @@ define amdgpu_ps i32 @ashr64(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: ashr64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_ashr_i64 s[0:1], s[0:1], s2
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -94,6 +100,7 @@ define amdgpu_ps i32 @abs32(i32 inreg %val0) {
 ; CHECK-LABEL: abs32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_abs_i32 s0, s0
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s0
 ; CHECK-NEXT:    ;;#ASMEND
@@ -114,6 +121,7 @@ define amdgpu_ps i32 @and32(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: and32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_and_b32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -128,6 +136,7 @@ define amdgpu_ps i32 @and64(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: and64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_and_b64 s[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -142,6 +151,7 @@ define amdgpu_ps i32 @or32(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: or32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_or_b32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -156,6 +166,7 @@ define amdgpu_ps i32 @or64(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: or64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_or_b64 s[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -170,6 +181,7 @@ define amdgpu_ps i32 @xor32(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: xor32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_xor_b32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -184,6 +196,7 @@ define amdgpu_ps i32 @xor64(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: xor64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_xor_b64 s[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -198,6 +211,7 @@ define amdgpu_ps i32 @nand32(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: nand32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_nand_b32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s0
 ; CHECK-NEXT:    ;;#ASMEND
@@ -217,6 +231,7 @@ define amdgpu_ps i32 @nand64(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: nand64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_nand_b64 s[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s[0:1]
 ; CHECK-NEXT:    ;;#ASMEND
@@ -236,6 +251,7 @@ define amdgpu_ps i32 @nor32(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: nor32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_nor_b32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s0
 ; CHECK-NEXT:    ;;#ASMEND
@@ -255,6 +271,7 @@ define amdgpu_ps i32 @nor64(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: nor64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_nor_b64 s[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s[0:1]
 ; CHECK-NEXT:    ;;#ASMEND
@@ -274,6 +291,7 @@ define amdgpu_ps i32 @xnor32(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: xnor32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_xnor_b32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s0
 ; CHECK-NEXT:    ;;#ASMEND
@@ -293,6 +311,7 @@ define amdgpu_ps i32 @xnor64(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: xnor64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_xnor_b64 s[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s[0:1]
 ; CHECK-NEXT:    ;;#ASMEND
@@ -312,6 +331,7 @@ define amdgpu_ps i32 @andn232(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: andn232:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_andn2_b32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -327,6 +347,7 @@ define amdgpu_ps i32 @nandn264(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: nandn264:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_andn2_b64 s[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -342,6 +363,7 @@ define amdgpu_ps i32 @orn232(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: orn232:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_orn2_b32 s0, s0, s1
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -357,6 +379,7 @@ define amdgpu_ps i32 @orn264(i64 inreg %val0, i64 inreg %val1) {
 ; CHECK-LABEL: orn264:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_orn2_b64 s[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -372,6 +395,7 @@ define amdgpu_ps i32 @bfe_i32(i32 inreg %val0) {
 ; CHECK-LABEL: bfe_i32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_bfe_i32 s0, s0, 0x80010
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -409,6 +433,7 @@ define amdgpu_ps i32 @bfe_u32(i32 inreg %val0) {
 ; CHECK-LABEL: bfe_u32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_bfe_u32 s0, s0, 0x80010
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; CHECK-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
@@ -488,6 +513,7 @@ define amdgpu_ps i32 @bcnt132(i32 inreg %val0) {
 ; CHECK-LABEL: bcnt132:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_bcnt1_i32_b32 s0, s0
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s0
 ; CHECK-NEXT:    ;;#ASMEND
@@ -526,6 +552,7 @@ define amdgpu_ps i32 @quadmask32(i32 inreg %val0) {
 ; CHECK-LABEL: quadmask32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_quadmask_b32 s0, s0
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s0
 ; CHECK-NEXT:    ;;#ASMEND
@@ -544,6 +571,7 @@ define amdgpu_ps i32 @quadmask64(i64 inreg %val0) {
 ; CHECK-LABEL: quadmask64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_quadmask_b64 s[0:1], s[0:1]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s[0:1]
 ; CHECK-NEXT:    ;;#ASMEND
@@ -562,6 +590,7 @@ define amdgpu_ps i32 @not32(i32 inreg %val0) {
 ; CHECK-LABEL: not32:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_not_b32 s0, s0
+; CHECK-NEXT:    s_cmp_lg_u32 s0, 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s0
 ; CHECK-NEXT:    ;;#ASMEND
@@ -580,6 +609,7 @@ define amdgpu_ps i32 @not64(i64 inreg %val0) {
 ; CHECK-LABEL: not64:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_not_b64 s[0:1], s[0:1]
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ; use s[0:1]
 ; CHECK-NEXT:    ;;#ASMEND
diff --git a/llvm/test/CodeGen/AMDGPU/s_uaddo_usubo_pseudo.ll b/llvm/test/CodeGen/AMDGPU/s_uaddo_usubo_pseudo.ll
index 7552f6b..a828ee0 100644
--- a/llvm/test/CodeGen/AMDGPU/s_uaddo_usubo_pseudo.ll
+++ b/llvm/test/CodeGen/AMDGPU/s_uaddo_usubo_pseudo.ll
@@ -12,6 +12,8 @@ define amdgpu_ps i32 @s_uaddo_pseudo(i32 inreg %val0) {
 ; CHECK-LABEL: s_uaddo_pseudo:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_add_u32 s0, s0, 1
+; CHECK-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[0:1], 0
 ; CHECK-NEXT:    s_addc_u32 s0, 1, 0
 ; CHECK-NEXT:    ; return to shader part epilog
   %pair = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 %val0, i32 1)
@@ -30,6 +32,8 @@ define amdgpu_ps i32 @s_usubo_pseudo(i32 inreg %val0, i32 inreg %val1) {
 ; CHECK-LABEL: s_usubo_pseudo:
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    s_sub_u32 s0, s0, 1
+; CHECK-NEXT:    s_cselect_b64 s[2:3], -1, 0
+; CHECK-NEXT:    s_cmp_lg_u64 s[2:3], 0
 ; CHECK-NEXT:    s_subb_u32 s0, s1, 0
 ; CHECK-NEXT:    ; return to shader part epilog
   %pair = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 %val0, i32 1)
diff --git a/llvm/test/CodeGen/AMDGPU/sdiv64.ll b/llvm/test/CodeGen/AMDGPU/sdiv64.ll
index 71f5a94..5f6d622 100644
--- a/llvm/test/CodeGen/AMDGPU/sdiv64.ll
+++ b/llvm/test/CodeGen/AMDGPU/sdiv64.ll
@@ -56,9 +56,10 @@ define amdgpu_kernel void @s_test_sdiv(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GCN-NEXT:    s_addc_u32 s15, 0, s16
 ; GCN-NEXT:    s_add_u32 s16, s0, s1
 ; GCN-NEXT:    v_mov_b32_e32 v0, s16
-; GCN-NEXT:    v_mul_hi_u32 v0, s12, v0
 ; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GCN-NEXT:    v_mul_hi_u32 v0, s12, v0
 ; GCN-NEXT:    s_or_b32 s0, s0, s1
+; GCN-NEXT:    s_cmp_lg_u32 s0, 0
 ; GCN-NEXT:    s_addc_u32 s14, s14, s15
 ; GCN-NEXT:    s_mul_i32 s0, s12, s14
 ; GCN-NEXT:    v_readfirstlane_b32 s1, v0
@@ -89,6 +90,7 @@ define amdgpu_kernel void @s_test_sdiv(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GCN-NEXT:    s_add_u32 s15, s16, s0
 ; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GCN-NEXT:    s_or_b32 s0, s0, s1
+; GCN-NEXT:    s_cmp_lg_u32 s0, 0
 ; GCN-NEXT:    s_addc_u32 s14, s14, s12
 ; GCN-NEXT:    s_ashr_i32 s12, s7, 31
 ; GCN-NEXT:    s_add_u32 s0, s6, s12
@@ -114,50 +116,52 @@ define amdgpu_kernel void @s_test_sdiv(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GCN-NEXT:    v_readfirstlane_b32 s4, v0
 ; GCN-NEXT:    s_addc_u32 s4, s4, 0
 ; GCN-NEXT:    s_mul_i32 s14, s7, s14
-; GCN-NEXT:    s_add_u32 s16, s1, s14
-; GCN-NEXT:    v_mov_b32_e32 v0, s16
+; GCN-NEXT:    s_add_u32 s14, s1, s14
+; GCN-NEXT:    v_mov_b32_e32 v0, s14
 ; GCN-NEXT:    v_mul_hi_u32 v0, s10, v0
-; GCN-NEXT:    s_addc_u32 s17, 0, s4
+; GCN-NEXT:    s_addc_u32 s15, 0, s4
 ; GCN-NEXT:    s_mov_b32 s1, s5
-; GCN-NEXT:    s_mul_i32 s4, s10, s17
+; GCN-NEXT:    s_mul_i32 s4, s10, s15
 ; GCN-NEXT:    v_readfirstlane_b32 s5, v0
 ; GCN-NEXT:    s_add_i32 s4, s5, s4
-; GCN-NEXT:    s_mul_i32 s5, s11, s16
-; GCN-NEXT:    s_add_i32 s18, s4, s5
-; GCN-NEXT:    s_sub_i32 s14, s7, s18
-; GCN-NEXT:    s_mul_i32 s4, s10, s16
+; GCN-NEXT:    s_mul_i32 s5, s11, s14
+; GCN-NEXT:    s_add_i32 s16, s4, s5
+; GCN-NEXT:    s_sub_i32 s17, s7, s16
+; GCN-NEXT:    s_mul_i32 s4, s10, s14
 ; GCN-NEXT:    s_sub_u32 s6, s6, s4
 ; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
-; GCN-NEXT:    s_or_b32 s15, s4, s5
-; GCN-NEXT:    s_subb_u32 s19, s14, s11
-; GCN-NEXT:    s_sub_u32 s20, s6, s10
-; GCN-NEXT:    s_cselect_b64 s[14:15], -1, 0
-; GCN-NEXT:    s_or_b32 s14, s14, s15
-; GCN-NEXT:    s_subb_u32 s14, s19, 0
-; GCN-NEXT:    s_cmp_ge_u32 s14, s11
-; GCN-NEXT:    s_cselect_b32 s15, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s20, s10
-; GCN-NEXT:    s_cselect_b32 s19, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s14, s11
-; GCN-NEXT:    s_cselect_b32 s14, s19, s15
-; GCN-NEXT:    s_add_u32 s15, s16, 1
-; GCN-NEXT:    s_addc_u32 s19, s17, 0
-; GCN-NEXT:    s_add_u32 s20, s16, 2
-; GCN-NEXT:    s_addc_u32 s21, s17, 0
-; GCN-NEXT:    s_cmp_lg_u32 s14, 0
-; GCN-NEXT:    s_cselect_b32 s14, s20, s15
-; GCN-NEXT:    s_cselect_b32 s15, s21, s19
+; GCN-NEXT:    s_or_b32 s18, s4, s5
+; GCN-NEXT:    s_cmp_lg_u32 s18, 0
+; GCN-NEXT:    s_subb_u32 s17, s17, s11
+; GCN-NEXT:    s_sub_u32 s19, s6, s10
+; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GCN-NEXT:    s_or_b32 s4, s4, s5
-; GCN-NEXT:    s_subb_u32 s4, s7, s18
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
+; GCN-NEXT:    s_subb_u32 s4, s17, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s4, s11
 ; GCN-NEXT:    s_cselect_b32 s5, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s6, s10
-; GCN-NEXT:    s_cselect_b32 s6, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s19, s10
+; GCN-NEXT:    s_cselect_b32 s17, -1, 0
 ; GCN-NEXT:    s_cmp_eq_u32 s4, s11
-; GCN-NEXT:    s_cselect_b32 s4, s6, s5
+; GCN-NEXT:    s_cselect_b32 s4, s17, s5
+; GCN-NEXT:    s_add_u32 s5, s14, 1
+; GCN-NEXT:    s_addc_u32 s17, s15, 0
+; GCN-NEXT:    s_add_u32 s19, s14, 2
+; GCN-NEXT:    s_addc_u32 s20, s15, 0
 ; GCN-NEXT:    s_cmp_lg_u32 s4, 0
-; GCN-NEXT:    s_cselect_b32 s5, s15, s17
-; GCN-NEXT:    s_cselect_b32 s4, s14, s16
+; GCN-NEXT:    s_cselect_b32 s4, s19, s5
+; GCN-NEXT:    s_cselect_b32 s5, s20, s17
+; GCN-NEXT:    s_cmp_lg_u32 s18, 0
+; GCN-NEXT:    s_subb_u32 s7, s7, s16
+; GCN-NEXT:    s_cmp_ge_u32 s7, s11
+; GCN-NEXT:    s_cselect_b32 s16, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s6, s10
+; GCN-NEXT:    s_cselect_b32 s6, -1, 0
+; GCN-NEXT:    s_cmp_eq_u32 s7, s11
+; GCN-NEXT:    s_cselect_b32 s6, s6, s16
+; GCN-NEXT:    s_cmp_lg_u32 s6, 0
+; GCN-NEXT:    s_cselect_b32 s5, s5, s15
+; GCN-NEXT:    s_cselect_b32 s4, s4, s14
 ; GCN-NEXT:    s_xor_b64 s[6:7], s[12:13], s[8:9]
 ; GCN-NEXT:    s_xor_b64 s[4:5], s[4:5], s[6:7]
 ; GCN-NEXT:    s_sub_u32 s4, s4, s6
@@ -204,6 +208,7 @@ define amdgpu_kernel void @s_test_sdiv(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GCN-IR-NEXT:    s_add_u32 s18, s16, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[10:11], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s10, s10, s11
+; GCN-IR-NEXT:    s_cmp_lg_u32 s10, 0
 ; GCN-IR-NEXT:    s_addc_u32 s10, s17, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[10:11], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s16, 63, s16
@@ -237,6 +242,7 @@ define amdgpu_kernel void @s_test_sdiv(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GCN-IR-NEXT:    s_add_u32 s14, s14, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[20:21], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s20, s20, s21
+; GCN-IR-NEXT:    s_cmp_lg_u32 s20, 0
 ; GCN-IR-NEXT:    s_addc_u32 s15, s15, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[20:21], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[12:13], s[8:9]
@@ -1189,9 +1195,10 @@ define amdgpu_kernel void @s_test_sdiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    s_addc_u32 s12, 0, s13
 ; GCN-NEXT:    s_add_u32 s13, s8, s9
 ; GCN-NEXT:    v_mov_b32_e32 v0, s13
-; GCN-NEXT:    v_mul_hi_u32 v0, s2, v0
 ; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; GCN-NEXT:    v_mul_hi_u32 v0, s2, v0
 ; GCN-NEXT:    s_or_b32 s8, s8, s9
+; GCN-NEXT:    s_cmp_lg_u32 s8, 0
 ; GCN-NEXT:    s_addc_u32 s11, s11, s12
 ; GCN-NEXT:    s_mul_i32 s8, s2, s11
 ; GCN-NEXT:    v_readfirstlane_b32 s9, v0
@@ -1222,6 +1229,7 @@ define amdgpu_kernel void @s_test_sdiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    s_add_u32 s2, s13, s2
 ; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GCN-NEXT:    s_or_b32 s8, s8, s9
+; GCN-NEXT:    s_cmp_lg_u32 s8, 0
 ; GCN-NEXT:    s_addc_u32 s8, s11, s10
 ; GCN-NEXT:    v_mul_hi_u32 v1, s2, 24
 ; GCN-NEXT:    v_mul_hi_u32 v0, s8, 24
@@ -1230,46 +1238,48 @@ define amdgpu_kernel void @s_test_sdiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    v_readfirstlane_b32 s10, v1
 ; GCN-NEXT:    v_readfirstlane_b32 s9, v0
 ; GCN-NEXT:    s_add_u32 s8, s10, s8
-; GCN-NEXT:    s_addc_u32 s12, 0, s9
-; GCN-NEXT:    v_mov_b32_e32 v0, s12
+; GCN-NEXT:    s_addc_u32 s10, 0, s9
+; GCN-NEXT:    v_mov_b32_e32 v0, s10
 ; GCN-NEXT:    v_mul_hi_u32 v0, s6, v0
-; GCN-NEXT:    s_mul_i32 s8, s7, s12
+; GCN-NEXT:    s_mul_i32 s8, s7, s10
 ; GCN-NEXT:    v_readfirstlane_b32 s9, v0
-; GCN-NEXT:    s_add_i32 s13, s9, s8
-; GCN-NEXT:    s_sub_i32 s10, 0, s13
-; GCN-NEXT:    s_mul_i32 s8, s6, s12
-; GCN-NEXT:    s_sub_u32 s14, 24, s8
+; GCN-NEXT:    s_add_i32 s11, s9, s8
+; GCN-NEXT:    s_sub_i32 s12, 0, s11
+; GCN-NEXT:    s_mul_i32 s8, s6, s10
+; GCN-NEXT:    s_sub_u32 s13, 24, s8
+; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; GCN-NEXT:    s_or_b32 s14, s8, s9
+; GCN-NEXT:    s_cmp_lg_u32 s14, 0
+; GCN-NEXT:    s_subb_u32 s12, s12, s7
+; GCN-NEXT:    s_sub_u32 s15, s13, s6
 ; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
-; GCN-NEXT:    s_or_b32 s11, s8, s9
-; GCN-NEXT:    s_subb_u32 s15, s10, s7
-; GCN-NEXT:    s_sub_u32 s16, s14, s6
-; GCN-NEXT:    s_cselect_b64 s[10:11], -1, 0
-; GCN-NEXT:    s_or_b32 s10, s10, s11
-; GCN-NEXT:    s_subb_u32 s10, s15, 0
-; GCN-NEXT:    s_cmp_ge_u32 s10, s7
-; GCN-NEXT:    s_cselect_b32 s11, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s16, s6
-; GCN-NEXT:    s_cselect_b32 s15, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s10, s7
-; GCN-NEXT:    s_cselect_b32 s10, s15, s11
-; GCN-NEXT:    s_add_u32 s11, s12, 1
-; GCN-NEXT:    s_addc_u32 s15, 0, 0
-; GCN-NEXT:    s_add_u32 s16, s12, 2
-; GCN-NEXT:    s_addc_u32 s17, 0, 0
-; GCN-NEXT:    s_cmp_lg_u32 s10, 0
-; GCN-NEXT:    s_cselect_b32 s10, s16, s11
-; GCN-NEXT:    s_cselect_b32 s11, s17, s15
 ; GCN-NEXT:    s_or_b32 s8, s8, s9
-; GCN-NEXT:    s_subb_u32 s8, 0, s13
+; GCN-NEXT:    s_cmp_lg_u32 s8, 0
+; GCN-NEXT:    s_subb_u32 s8, s12, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s8, s7
 ; GCN-NEXT:    s_cselect_b32 s9, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s14, s6
-; GCN-NEXT:    s_cselect_b32 s6, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s15, s6
+; GCN-NEXT:    s_cselect_b32 s12, -1, 0
 ; GCN-NEXT:    s_cmp_eq_u32 s8, s7
-; GCN-NEXT:    s_cselect_b32 s6, s6, s9
+; GCN-NEXT:    s_cselect_b32 s8, s12, s9
+; GCN-NEXT:    s_add_u32 s9, s10, 1
+; GCN-NEXT:    s_addc_u32 s12, 0, 0
+; GCN-NEXT:    s_add_u32 s15, s10, 2
+; GCN-NEXT:    s_addc_u32 s16, 0, 0
+; GCN-NEXT:    s_cmp_lg_u32 s8, 0
+; GCN-NEXT:    s_cselect_b32 s8, s15, s9
+; GCN-NEXT:    s_cselect_b32 s9, s16, s12
+; GCN-NEXT:    s_cmp_lg_u32 s14, 0
+; GCN-NEXT:    s_subb_u32 s11, 0, s11
+; GCN-NEXT:    s_cmp_ge_u32 s11, s7
+; GCN-NEXT:    s_cselect_b32 s12, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s13, s6
+; GCN-NEXT:    s_cselect_b32 s6, -1, 0
+; GCN-NEXT:    s_cmp_eq_u32 s11, s7
+; GCN-NEXT:    s_cselect_b32 s6, s6, s12
 ; GCN-NEXT:    s_cmp_lg_u32 s6, 0
-; GCN-NEXT:    s_cselect_b32 s7, s11, 0
-; GCN-NEXT:    s_cselect_b32 s6, s10, s12
+; GCN-NEXT:    s_cselect_b32 s7, s9, 0
+; GCN-NEXT:    s_cselect_b32 s6, s8, s10
 ; GCN-NEXT:    s_xor_b64 s[6:7], s[6:7], s[4:5]
 ; GCN-NEXT:    s_sub_u32 s6, s6, s4
 ; GCN-NEXT:    s_subb_u32 s7, s7, s4
@@ -1305,6 +1315,7 @@ define amdgpu_kernel void @s_test_sdiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s12, s10, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s8, s8, s9
+; GCN-IR-NEXT:    s_cmp_lg_u32 s8, 0
 ; GCN-IR-NEXT:    s_addc_u32 s8, s11, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s10, 63, s10
@@ -1337,6 +1348,7 @@ define amdgpu_kernel void @s_test_sdiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s16, s16, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[18:19], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s18, s18, s19
+; GCN-IR-NEXT:    s_cmp_lg_u32 s18, 0
 ; GCN-IR-NEXT:    s_addc_u32 s17, s17, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[18:19], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[10:11], s[6:7]
diff --git a/llvm/test/CodeGen/AMDGPU/srem.ll b/llvm/test/CodeGen/AMDGPU/srem.ll
index e12e31b..bbd1793 100644
--- a/llvm/test/CodeGen/AMDGPU/srem.ll
+++ b/llvm/test/CodeGen/AMDGPU/srem.ll
@@ -1513,7 +1513,7 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v0, s8
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v1, s9
 ; GCN-NEXT:    s_sub_u32 s3, 0, s8
-; GCN-NEXT:    s_subb_u32 s10, 0, s9
+; GCN-NEXT:    s_subb_u32 s12, 0, s9
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GCN-NEXT:    v_rcp_f32_e32 v0, v0
 ; GCN-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -1522,52 +1522,56 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0xcf800000, v0
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT:    v_readfirstlane_b32 s11, v1
-; GCN-NEXT:    v_readfirstlane_b32 s12, v0
-; GCN-NEXT:    s_mul_i32 s13, s3, s11
-; GCN-NEXT:    s_mul_hi_u32 s15, s3, s12
-; GCN-NEXT:    s_mul_i32 s14, s10, s12
-; GCN-NEXT:    s_add_i32 s13, s15, s13
-; GCN-NEXT:    s_add_i32 s13, s13, s14
-; GCN-NEXT:    s_mul_i32 s16, s3, s12
-; GCN-NEXT:    s_mul_i32 s15, s12, s13
-; GCN-NEXT:    s_mul_hi_u32 s17, s12, s16
-; GCN-NEXT:    s_mul_hi_u32 s14, s12, s13
+; GCN-NEXT:    v_readfirstlane_b32 s13, v1
+; GCN-NEXT:    v_readfirstlane_b32 s10, v0
+; GCN-NEXT:    s_mul_i32 s11, s3, s13
+; GCN-NEXT:    s_mul_hi_u32 s15, s3, s10
+; GCN-NEXT:    s_mul_i32 s14, s12, s10
+; GCN-NEXT:    s_add_i32 s11, s15, s11
+; GCN-NEXT:    s_add_i32 s11, s11, s14
+; GCN-NEXT:    s_mul_i32 s16, s3, s10
+; GCN-NEXT:    s_mul_i32 s15, s10, s11
+; GCN-NEXT:    s_mul_hi_u32 s17, s10, s16
+; GCN-NEXT:    s_mul_hi_u32 s14, s10, s11
 ; GCN-NEXT:    s_add_u32 s15, s17, s15
 ; GCN-NEXT:    s_addc_u32 s14, 0, s14
-; GCN-NEXT:    s_mul_hi_u32 s18, s11, s16
-; GCN-NEXT:    s_mul_i32 s16, s11, s16
+; GCN-NEXT:    s_mul_hi_u32 s18, s13, s16
+; GCN-NEXT:    s_mul_i32 s16, s13, s16
 ; GCN-NEXT:    s_add_u32 s15, s15, s16
-; GCN-NEXT:    s_mul_hi_u32 s17, s11, s13
+; GCN-NEXT:    s_mul_hi_u32 s17, s13, s11
 ; GCN-NEXT:    s_addc_u32 s14, s14, s18
 ; GCN-NEXT:    s_addc_u32 s15, s17, 0
-; GCN-NEXT:    s_mul_i32 s13, s11, s13
-; GCN-NEXT:    s_add_u32 s13, s14, s13
+; GCN-NEXT:    s_mul_i32 s11, s13, s11
+; GCN-NEXT:    s_add_u32 s11, s14, s11
 ; GCN-NEXT:    s_addc_u32 s14, 0, s15
-; GCN-NEXT:    s_add_u32 s12, s12, s13
-; GCN-NEXT:    s_addc_u32 s11, s11, s14
-; GCN-NEXT:    s_mul_i32 s13, s3, s11
-; GCN-NEXT:    s_mul_hi_u32 s14, s3, s12
-; GCN-NEXT:    s_add_i32 s13, s14, s13
-; GCN-NEXT:    s_mul_i32 s10, s10, s12
-; GCN-NEXT:    s_add_i32 s13, s13, s10
-; GCN-NEXT:    s_mul_i32 s3, s3, s12
-; GCN-NEXT:    s_mul_hi_u32 s14, s11, s3
-; GCN-NEXT:    s_mul_i32 s15, s11, s3
-; GCN-NEXT:    s_mul_i32 s17, s12, s13
-; GCN-NEXT:    s_mul_hi_u32 s3, s12, s3
-; GCN-NEXT:    s_mul_hi_u32 s16, s12, s13
+; GCN-NEXT:    s_add_u32 s15, s10, s11
+; GCN-NEXT:    s_cselect_b64 s[10:11], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[10:11], 0
+; GCN-NEXT:    s_addc_u32 s13, s13, s14
+; GCN-NEXT:    s_mul_i32 s10, s3, s13
+; GCN-NEXT:    s_mul_hi_u32 s11, s3, s15
+; GCN-NEXT:    s_add_i32 s10, s11, s10
+; GCN-NEXT:    s_mul_i32 s12, s12, s15
+; GCN-NEXT:    s_add_i32 s10, s10, s12
+; GCN-NEXT:    s_mul_i32 s3, s3, s15
+; GCN-NEXT:    s_mul_hi_u32 s12, s13, s3
+; GCN-NEXT:    s_mul_i32 s14, s13, s3
+; GCN-NEXT:    s_mul_i32 s17, s15, s10
+; GCN-NEXT:    s_mul_hi_u32 s3, s15, s3
+; GCN-NEXT:    s_mul_hi_u32 s16, s15, s10
 ; GCN-NEXT:    s_add_u32 s3, s3, s17
 ; GCN-NEXT:    s_addc_u32 s16, 0, s16
-; GCN-NEXT:    s_add_u32 s3, s3, s15
-; GCN-NEXT:    s_mul_hi_u32 s10, s11, s13
-; GCN-NEXT:    s_addc_u32 s3, s16, s14
-; GCN-NEXT:    s_addc_u32 s10, s10, 0
-; GCN-NEXT:    s_mul_i32 s13, s11, s13
-; GCN-NEXT:    s_add_u32 s3, s3, s13
-; GCN-NEXT:    s_addc_u32 s10, 0, s10
-; GCN-NEXT:    s_add_u32 s3, s12, s3
-; GCN-NEXT:    s_addc_u32 s14, s11, s10
+; GCN-NEXT:    s_add_u32 s3, s3, s14
+; GCN-NEXT:    s_mul_hi_u32 s11, s13, s10
+; GCN-NEXT:    s_addc_u32 s3, s16, s12
+; GCN-NEXT:    s_addc_u32 s11, s11, 0
+; GCN-NEXT:    s_mul_i32 s10, s13, s10
+; GCN-NEXT:    s_add_u32 s3, s3, s10
+; GCN-NEXT:    s_addc_u32 s12, 0, s11
+; GCN-NEXT:    s_add_u32 s3, s15, s3
+; GCN-NEXT:    s_cselect_b64 s[10:11], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[10:11], 0
+; GCN-NEXT:    s_addc_u32 s14, s13, s12
 ; GCN-NEXT:    s_ashr_i32 s10, s5, 31
 ; GCN-NEXT:    s_add_u32 s12, s4, s10
 ; GCN-NEXT:    s_mov_b32 s11, s10
@@ -1596,9 +1600,11 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
 ; GCN-NEXT:    s_mul_i32 s3, s8, s3
 ; GCN-NEXT:    s_sub_u32 s3, s12, s3
 ; GCN-NEXT:    s_cselect_b64 s[14:15], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[14:15], 0
 ; GCN-NEXT:    s_subb_u32 s12, s16, s9
 ; GCN-NEXT:    s_sub_u32 s18, s3, s8
 ; GCN-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; GCN-NEXT:    s_subb_u32 s19, s12, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s19, s9
 ; GCN-NEXT:    s_cselect_b32 s20, -1, 0
@@ -1608,10 +1614,12 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
 ; GCN-NEXT:    s_cselect_b32 s20, s21, s20
 ; GCN-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; GCN-NEXT:    s_subb_u32 s12, s12, s9
-; GCN-NEXT:    s_sub_u32 s16, s18, s8
+; GCN-NEXT:    s_sub_u32 s21, s18, s8
+; GCN-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; GCN-NEXT:    s_subb_u32 s12, s12, 0
 ; GCN-NEXT:    s_cmp_lg_u32 s20, 0
-; GCN-NEXT:    s_cselect_b32 s16, s16, s18
+; GCN-NEXT:    s_cselect_b32 s16, s21, s18
 ; GCN-NEXT:    s_cselect_b32 s12, s12, s19
 ; GCN-NEXT:    s_cmp_lg_u64 s[14:15], 0
 ; GCN-NEXT:    s_subb_u32 s5, s13, s5
@@ -1923,9 +1931,11 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
 ; TONGA-NEXT:    v_readfirstlane_b32 s14, v0
 ; TONGA-NEXT:    s_sub_u32 s12, s12, s14
 ; TONGA-NEXT:    s_cselect_b64 s[14:15], -1, 0
+; TONGA-NEXT:    s_cmp_lg_u64 s[14:15], 0
 ; TONGA-NEXT:    s_subb_u32 s3, s3, s7
 ; TONGA-NEXT:    s_sub_u32 s18, s12, s6
 ; TONGA-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; TONGA-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; TONGA-NEXT:    s_subb_u32 s19, s3, 0
 ; TONGA-NEXT:    s_cmp_ge_u32 s19, s7
 ; TONGA-NEXT:    s_cselect_b32 s20, -1, 0
@@ -1935,10 +1945,12 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
 ; TONGA-NEXT:    s_cselect_b32 s20, s21, s20
 ; TONGA-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; TONGA-NEXT:    s_subb_u32 s3, s3, s7
-; TONGA-NEXT:    s_sub_u32 s16, s18, s6
+; TONGA-NEXT:    s_sub_u32 s21, s18, s6
+; TONGA-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; TONGA-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; TONGA-NEXT:    s_subb_u32 s3, s3, 0
 ; TONGA-NEXT:    s_cmp_lg_u32 s20, 0
-; TONGA-NEXT:    s_cselect_b32 s16, s16, s18
+; TONGA-NEXT:    s_cselect_b32 s16, s21, s18
 ; TONGA-NEXT:    s_cselect_b32 s3, s3, s19
 ; TONGA-NEXT:    s_cmp_lg_u64 s[14:15], 0
 ; TONGA-NEXT:    s_subb_u32 s5, s13, s5
@@ -2718,7 +2730,7 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v0, s6
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v1, s7
 ; GCN-NEXT:    s_sub_u32 s9, 0, s6
-; GCN-NEXT:    s_subb_u32 s14, 0, s7
+; GCN-NEXT:    s_subb_u32 s16, 0, s7
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GCN-NEXT:    v_rcp_f32_e32 v0, v0
 ; GCN-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -2727,52 +2739,56 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0xcf800000, v0
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT:    v_readfirstlane_b32 s15, v1
-; GCN-NEXT:    v_readfirstlane_b32 s16, v0
-; GCN-NEXT:    s_mul_i32 s17, s9, s15
-; GCN-NEXT:    s_mul_hi_u32 s19, s9, s16
-; GCN-NEXT:    s_mul_i32 s18, s14, s16
-; GCN-NEXT:    s_add_i32 s17, s19, s17
-; GCN-NEXT:    s_add_i32 s17, s17, s18
-; GCN-NEXT:    s_mul_i32 s20, s9, s16
-; GCN-NEXT:    s_mul_i32 s19, s16, s17
-; GCN-NEXT:    s_mul_hi_u32 s21, s16, s20
-; GCN-NEXT:    s_mul_hi_u32 s18, s16, s17
+; GCN-NEXT:    v_readfirstlane_b32 s17, v1
+; GCN-NEXT:    v_readfirstlane_b32 s14, v0
+; GCN-NEXT:    s_mul_i32 s15, s9, s17
+; GCN-NEXT:    s_mul_hi_u32 s19, s9, s14
+; GCN-NEXT:    s_mul_i32 s18, s16, s14
+; GCN-NEXT:    s_add_i32 s15, s19, s15
+; GCN-NEXT:    s_add_i32 s15, s15, s18
+; GCN-NEXT:    s_mul_i32 s20, s9, s14
+; GCN-NEXT:    s_mul_i32 s19, s14, s15
+; GCN-NEXT:    s_mul_hi_u32 s21, s14, s20
+; GCN-NEXT:    s_mul_hi_u32 s18, s14, s15
 ; GCN-NEXT:    s_add_u32 s19, s21, s19
 ; GCN-NEXT:    s_addc_u32 s18, 0, s18
-; GCN-NEXT:    s_mul_hi_u32 s22, s15, s20
-; GCN-NEXT:    s_mul_i32 s20, s15, s20
+; GCN-NEXT:    s_mul_hi_u32 s22, s17, s20
+; GCN-NEXT:    s_mul_i32 s20, s17, s20
 ; GCN-NEXT:    s_add_u32 s19, s19, s20
-; GCN-NEXT:    s_mul_hi_u32 s21, s15, s17
+; GCN-NEXT:    s_mul_hi_u32 s21, s17, s15
 ; GCN-NEXT:    s_addc_u32 s18, s18, s22
 ; GCN-NEXT:    s_addc_u32 s19, s21, 0
-; GCN-NEXT:    s_mul_i32 s17, s15, s17
-; GCN-NEXT:    s_add_u32 s17, s18, s17
+; GCN-NEXT:    s_mul_i32 s15, s17, s15
+; GCN-NEXT:    s_add_u32 s15, s18, s15
 ; GCN-NEXT:    s_addc_u32 s18, 0, s19
-; GCN-NEXT:    s_add_u32 s16, s16, s17
-; GCN-NEXT:    s_addc_u32 s15, s15, s18
-; GCN-NEXT:    s_mul_i32 s17, s9, s15
-; GCN-NEXT:    s_mul_hi_u32 s18, s9, s16
-; GCN-NEXT:    s_add_i32 s17, s18, s17
-; GCN-NEXT:    s_mul_i32 s14, s14, s16
-; GCN-NEXT:    s_add_i32 s17, s17, s14
-; GCN-NEXT:    s_mul_i32 s9, s9, s16
-; GCN-NEXT:    s_mul_hi_u32 s18, s15, s9
-; GCN-NEXT:    s_mul_i32 s19, s15, s9
-; GCN-NEXT:    s_mul_i32 s21, s16, s17
-; GCN-NEXT:    s_mul_hi_u32 s9, s16, s9
-; GCN-NEXT:    s_mul_hi_u32 s20, s16, s17
+; GCN-NEXT:    s_add_u32 s19, s14, s15
+; GCN-NEXT:    s_cselect_b64 s[14:15], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[14:15], 0
+; GCN-NEXT:    s_addc_u32 s17, s17, s18
+; GCN-NEXT:    s_mul_i32 s14, s9, s17
+; GCN-NEXT:    s_mul_hi_u32 s15, s9, s19
+; GCN-NEXT:    s_add_i32 s14, s15, s14
+; GCN-NEXT:    s_mul_i32 s16, s16, s19
+; GCN-NEXT:    s_add_i32 s14, s14, s16
+; GCN-NEXT:    s_mul_i32 s9, s9, s19
+; GCN-NEXT:    s_mul_hi_u32 s16, s17, s9
+; GCN-NEXT:    s_mul_i32 s18, s17, s9
+; GCN-NEXT:    s_mul_i32 s21, s19, s14
+; GCN-NEXT:    s_mul_hi_u32 s9, s19, s9
+; GCN-NEXT:    s_mul_hi_u32 s20, s19, s14
 ; GCN-NEXT:    s_add_u32 s9, s9, s21
 ; GCN-NEXT:    s_addc_u32 s20, 0, s20
-; GCN-NEXT:    s_add_u32 s9, s9, s19
-; GCN-NEXT:    s_mul_hi_u32 s14, s15, s17
-; GCN-NEXT:    s_addc_u32 s9, s20, s18
-; GCN-NEXT:    s_addc_u32 s14, s14, 0
-; GCN-NEXT:    s_mul_i32 s17, s15, s17
-; GCN-NEXT:    s_add_u32 s9, s9, s17
-; GCN-NEXT:    s_addc_u32 s14, 0, s14
-; GCN-NEXT:    s_add_u32 s9, s16, s9
-; GCN-NEXT:    s_addc_u32 s18, s15, s14
+; GCN-NEXT:    s_add_u32 s9, s9, s18
+; GCN-NEXT:    s_mul_hi_u32 s15, s17, s14
+; GCN-NEXT:    s_addc_u32 s9, s20, s16
+; GCN-NEXT:    s_addc_u32 s15, s15, 0
+; GCN-NEXT:    s_mul_i32 s14, s17, s14
+; GCN-NEXT:    s_add_u32 s9, s9, s14
+; GCN-NEXT:    s_addc_u32 s16, 0, s15
+; GCN-NEXT:    s_add_u32 s9, s19, s9
+; GCN-NEXT:    s_cselect_b64 s[14:15], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[14:15], 0
+; GCN-NEXT:    s_addc_u32 s18, s17, s16
 ; GCN-NEXT:    s_ashr_i32 s14, s11, 31
 ; GCN-NEXT:    s_add_u32 s16, s10, s14
 ; GCN-NEXT:    s_mov_b32 s15, s14
@@ -2801,9 +2817,11 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_mul_i32 s9, s6, s9
 ; GCN-NEXT:    s_sub_u32 s9, s16, s9
 ; GCN-NEXT:    s_cselect_b64 s[18:19], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[18:19], 0
 ; GCN-NEXT:    s_subb_u32 s16, s20, s7
 ; GCN-NEXT:    s_sub_u32 s22, s9, s6
 ; GCN-NEXT:    s_cselect_b64 s[20:21], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[20:21], 0
 ; GCN-NEXT:    s_subb_u32 s23, s16, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s23, s7
 ; GCN-NEXT:    s_cselect_b32 s24, -1, 0
@@ -2813,10 +2831,12 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_cselect_b32 s24, s25, s24
 ; GCN-NEXT:    s_cmp_lg_u64 s[20:21], 0
 ; GCN-NEXT:    s_subb_u32 s16, s16, s7
-; GCN-NEXT:    s_sub_u32 s20, s22, s6
+; GCN-NEXT:    s_sub_u32 s25, s22, s6
+; GCN-NEXT:    s_cselect_b64 s[20:21], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[20:21], 0
 ; GCN-NEXT:    s_subb_u32 s16, s16, 0
 ; GCN-NEXT:    s_cmp_lg_u32 s24, 0
-; GCN-NEXT:    s_cselect_b32 s20, s20, s22
+; GCN-NEXT:    s_cselect_b32 s20, s25, s22
 ; GCN-NEXT:    s_cselect_b32 s16, s16, s23
 ; GCN-NEXT:    s_cmp_lg_u64 s[18:19], 0
 ; GCN-NEXT:    s_subb_u32 s11, s17, s11
@@ -2867,7 +2887,7 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v0, s10
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v1, s11
 ; GCN-NEXT:    s_sub_u32 s3, 0, s10
-; GCN-NEXT:    s_subb_u32 s12, 0, s11
+; GCN-NEXT:    s_subb_u32 s14, 0, s11
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GCN-NEXT:    v_rcp_f32_e32 v0, v0
 ; GCN-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -2876,52 +2896,56 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0xcf800000, v0
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT:    v_readfirstlane_b32 s13, v1
-; GCN-NEXT:    v_readfirstlane_b32 s14, v0
-; GCN-NEXT:    s_mul_i32 s15, s3, s13
-; GCN-NEXT:    s_mul_hi_u32 s17, s3, s14
-; GCN-NEXT:    s_mul_i32 s16, s12, s14
-; GCN-NEXT:    s_add_i32 s15, s17, s15
-; GCN-NEXT:    s_add_i32 s15, s15, s16
-; GCN-NEXT:    s_mul_i32 s18, s3, s14
-; GCN-NEXT:    s_mul_i32 s17, s14, s15
-; GCN-NEXT:    s_mul_hi_u32 s19, s14, s18
-; GCN-NEXT:    s_mul_hi_u32 s16, s14, s15
+; GCN-NEXT:    v_readfirstlane_b32 s15, v1
+; GCN-NEXT:    v_readfirstlane_b32 s12, v0
+; GCN-NEXT:    s_mul_i32 s13, s3, s15
+; GCN-NEXT:    s_mul_hi_u32 s17, s3, s12
+; GCN-NEXT:    s_mul_i32 s16, s14, s12
+; GCN-NEXT:    s_add_i32 s13, s17, s13
+; GCN-NEXT:    s_add_i32 s13, s13, s16
+; GCN-NEXT:    s_mul_i32 s18, s3, s12
+; GCN-NEXT:    s_mul_i32 s17, s12, s13
+; GCN-NEXT:    s_mul_hi_u32 s19, s12, s18
+; GCN-NEXT:    s_mul_hi_u32 s16, s12, s13
 ; GCN-NEXT:    s_add_u32 s17, s19, s17
 ; GCN-NEXT:    s_addc_u32 s16, 0, s16
-; GCN-NEXT:    s_mul_hi_u32 s20, s13, s18
-; GCN-NEXT:    s_mul_i32 s18, s13, s18
+; GCN-NEXT:    s_mul_hi_u32 s20, s15, s18
+; GCN-NEXT:    s_mul_i32 s18, s15, s18
 ; GCN-NEXT:    s_add_u32 s17, s17, s18
-; GCN-NEXT:    s_mul_hi_u32 s19, s13, s15
+; GCN-NEXT:    s_mul_hi_u32 s19, s15, s13
 ; GCN-NEXT:    s_addc_u32 s16, s16, s20
 ; GCN-NEXT:    s_addc_u32 s17, s19, 0
-; GCN-NEXT:    s_mul_i32 s15, s13, s15
-; GCN-NEXT:    s_add_u32 s15, s16, s15
+; GCN-NEXT:    s_mul_i32 s13, s15, s13
+; GCN-NEXT:    s_add_u32 s13, s16, s13
 ; GCN-NEXT:    s_addc_u32 s16, 0, s17
-; GCN-NEXT:    s_add_u32 s14, s14, s15
-; GCN-NEXT:    s_addc_u32 s13, s13, s16
-; GCN-NEXT:    s_mul_i32 s15, s3, s13
-; GCN-NEXT:    s_mul_hi_u32 s16, s3, s14
-; GCN-NEXT:    s_add_i32 s15, s16, s15
-; GCN-NEXT:    s_mul_i32 s12, s12, s14
-; GCN-NEXT:    s_add_i32 s15, s15, s12
-; GCN-NEXT:    s_mul_i32 s3, s3, s14
-; GCN-NEXT:    s_mul_hi_u32 s16, s13, s3
-; GCN-NEXT:    s_mul_i32 s17, s13, s3
-; GCN-NEXT:    s_mul_i32 s19, s14, s15
-; GCN-NEXT:    s_mul_hi_u32 s3, s14, s3
-; GCN-NEXT:    s_mul_hi_u32 s18, s14, s15
+; GCN-NEXT:    s_add_u32 s17, s12, s13
+; GCN-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[12:13], 0
+; GCN-NEXT:    s_addc_u32 s15, s15, s16
+; GCN-NEXT:    s_mul_i32 s12, s3, s15
+; GCN-NEXT:    s_mul_hi_u32 s13, s3, s17
+; GCN-NEXT:    s_add_i32 s12, s13, s12
+; GCN-NEXT:    s_mul_i32 s14, s14, s17
+; GCN-NEXT:    s_add_i32 s12, s12, s14
+; GCN-NEXT:    s_mul_i32 s3, s3, s17
+; GCN-NEXT:    s_mul_hi_u32 s14, s15, s3
+; GCN-NEXT:    s_mul_i32 s16, s15, s3
+; GCN-NEXT:    s_mul_i32 s19, s17, s12
+; GCN-NEXT:    s_mul_hi_u32 s3, s17, s3
+; GCN-NEXT:    s_mul_hi_u32 s18, s17, s12
 ; GCN-NEXT:    s_add_u32 s3, s3, s19
 ; GCN-NEXT:    s_addc_u32 s18, 0, s18
-; GCN-NEXT:    s_add_u32 s3, s3, s17
-; GCN-NEXT:    s_mul_hi_u32 s12, s13, s15
-; GCN-NEXT:    s_addc_u32 s3, s18, s16
-; GCN-NEXT:    s_addc_u32 s12, s12, 0
-; GCN-NEXT:    s_mul_i32 s15, s13, s15
-; GCN-NEXT:    s_add_u32 s3, s3, s15
-; GCN-NEXT:    s_addc_u32 s12, 0, s12
-; GCN-NEXT:    s_add_u32 s3, s14, s3
-; GCN-NEXT:    s_addc_u32 s16, s13, s12
+; GCN-NEXT:    s_add_u32 s3, s3, s16
+; GCN-NEXT:    s_mul_hi_u32 s13, s15, s12
+; GCN-NEXT:    s_addc_u32 s3, s18, s14
+; GCN-NEXT:    s_addc_u32 s13, s13, 0
+; GCN-NEXT:    s_mul_i32 s12, s15, s12
+; GCN-NEXT:    s_add_u32 s3, s3, s12
+; GCN-NEXT:    s_addc_u32 s14, 0, s13
+; GCN-NEXT:    s_add_u32 s3, s17, s3
+; GCN-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[12:13], 0
+; GCN-NEXT:    s_addc_u32 s16, s15, s14
 ; GCN-NEXT:    s_ashr_i32 s12, s5, 31
 ; GCN-NEXT:    s_add_u32 s14, s4, s12
 ; GCN-NEXT:    s_mov_b32 s13, s12
@@ -2950,9 +2974,11 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_mul_i32 s3, s10, s3
 ; GCN-NEXT:    s_sub_u32 s3, s14, s3
 ; GCN-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; GCN-NEXT:    s_subb_u32 s14, s18, s11
 ; GCN-NEXT:    s_sub_u32 s20, s3, s10
 ; GCN-NEXT:    s_cselect_b64 s[18:19], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[18:19], 0
 ; GCN-NEXT:    s_subb_u32 s21, s14, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s21, s11
 ; GCN-NEXT:    s_cselect_b32 s22, -1, 0
@@ -2962,10 +2988,12 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_cselect_b32 s22, s23, s22
 ; GCN-NEXT:    s_cmp_lg_u64 s[18:19], 0
 ; GCN-NEXT:    s_subb_u32 s14, s14, s11
-; GCN-NEXT:    s_sub_u32 s18, s20, s10
+; GCN-NEXT:    s_sub_u32 s23, s20, s10
+; GCN-NEXT:    s_cselect_b64 s[18:19], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[18:19], 0
 ; GCN-NEXT:    s_subb_u32 s14, s14, 0
 ; GCN-NEXT:    s_cmp_lg_u32 s22, 0
-; GCN-NEXT:    s_cselect_b32 s18, s18, s20
+; GCN-NEXT:    s_cselect_b32 s18, s23, s20
 ; GCN-NEXT:    s_cselect_b32 s14, s14, s21
 ; GCN-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; GCN-NEXT:    s_subb_u32 s5, s15, s5
@@ -3435,9 +3463,11 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; TONGA-NEXT:    v_readfirstlane_b32 s14, v0
 ; TONGA-NEXT:    s_sub_u32 s12, s12, s14
 ; TONGA-NEXT:    s_cselect_b64 s[14:15], -1, 0
+; TONGA-NEXT:    s_cmp_lg_u64 s[14:15], 0
 ; TONGA-NEXT:    s_subb_u32 s1, s1, s7
 ; TONGA-NEXT:    s_sub_u32 s18, s12, s6
 ; TONGA-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; TONGA-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; TONGA-NEXT:    s_subb_u32 s19, s1, 0
 ; TONGA-NEXT:    s_cmp_ge_u32 s19, s7
 ; TONGA-NEXT:    s_cselect_b32 s20, -1, 0
@@ -3447,10 +3477,12 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; TONGA-NEXT:    s_cselect_b32 s20, s21, s20
 ; TONGA-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; TONGA-NEXT:    s_subb_u32 s1, s1, s7
-; TONGA-NEXT:    s_sub_u32 s16, s18, s6
+; TONGA-NEXT:    s_sub_u32 s21, s18, s6
+; TONGA-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; TONGA-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; TONGA-NEXT:    s_subb_u32 s1, s1, 0
 ; TONGA-NEXT:    s_cmp_lg_u32 s20, 0
-; TONGA-NEXT:    s_cselect_b32 s16, s16, s18
+; TONGA-NEXT:    s_cselect_b32 s16, s21, s18
 ; TONGA-NEXT:    s_cselect_b32 s1, s1, s19
 ; TONGA-NEXT:    s_cmp_lg_u64 s[14:15], 0
 ; TONGA-NEXT:    s_subb_u32 s3, s13, s3
@@ -4902,7 +4934,7 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v0, s6
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v1, s7
 ; GCN-NEXT:    s_sub_u32 s17, 0, s6
-; GCN-NEXT:    s_subb_u32 s22, 0, s7
+; GCN-NEXT:    s_subb_u32 s24, 0, s7
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GCN-NEXT:    v_rcp_f32_e32 v0, v0
 ; GCN-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -4911,52 +4943,56 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0xcf800000, v0
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT:    v_readfirstlane_b32 s23, v1
-; GCN-NEXT:    v_readfirstlane_b32 s24, v0
-; GCN-NEXT:    s_mul_i32 s25, s17, s23
-; GCN-NEXT:    s_mul_hi_u32 s27, s17, s24
-; GCN-NEXT:    s_mul_i32 s26, s22, s24
-; GCN-NEXT:    s_add_i32 s25, s27, s25
-; GCN-NEXT:    s_add_i32 s25, s25, s26
-; GCN-NEXT:    s_mul_i32 s28, s17, s24
-; GCN-NEXT:    s_mul_i32 s27, s24, s25
-; GCN-NEXT:    s_mul_hi_u32 s29, s24, s28
-; GCN-NEXT:    s_mul_hi_u32 s26, s24, s25
+; GCN-NEXT:    v_readfirstlane_b32 s25, v1
+; GCN-NEXT:    v_readfirstlane_b32 s22, v0
+; GCN-NEXT:    s_mul_i32 s23, s17, s25
+; GCN-NEXT:    s_mul_hi_u32 s27, s17, s22
+; GCN-NEXT:    s_mul_i32 s26, s24, s22
+; GCN-NEXT:    s_add_i32 s23, s27, s23
+; GCN-NEXT:    s_add_i32 s23, s23, s26
+; GCN-NEXT:    s_mul_i32 s28, s17, s22
+; GCN-NEXT:    s_mul_i32 s27, s22, s23
+; GCN-NEXT:    s_mul_hi_u32 s29, s22, s28
+; GCN-NEXT:    s_mul_hi_u32 s26, s22, s23
 ; GCN-NEXT:    s_add_u32 s27, s29, s27
 ; GCN-NEXT:    s_addc_u32 s26, 0, s26
-; GCN-NEXT:    s_mul_hi_u32 s30, s23, s28
-; GCN-NEXT:    s_mul_i32 s28, s23, s28
+; GCN-NEXT:    s_mul_hi_u32 s30, s25, s28
+; GCN-NEXT:    s_mul_i32 s28, s25, s28
 ; GCN-NEXT:    s_add_u32 s27, s27, s28
-; GCN-NEXT:    s_mul_hi_u32 s29, s23, s25
+; GCN-NEXT:    s_mul_hi_u32 s29, s25, s23
 ; GCN-NEXT:    s_addc_u32 s26, s26, s30
 ; GCN-NEXT:    s_addc_u32 s27, s29, 0
-; GCN-NEXT:    s_mul_i32 s25, s23, s25
-; GCN-NEXT:    s_add_u32 s25, s26, s25
+; GCN-NEXT:    s_mul_i32 s23, s25, s23
+; GCN-NEXT:    s_add_u32 s23, s26, s23
 ; GCN-NEXT:    s_addc_u32 s26, 0, s27
-; GCN-NEXT:    s_add_u32 s24, s24, s25
-; GCN-NEXT:    s_addc_u32 s23, s23, s26
-; GCN-NEXT:    s_mul_i32 s25, s17, s23
-; GCN-NEXT:    s_mul_hi_u32 s26, s17, s24
-; GCN-NEXT:    s_add_i32 s25, s26, s25
-; GCN-NEXT:    s_mul_i32 s22, s22, s24
-; GCN-NEXT:    s_add_i32 s25, s25, s22
-; GCN-NEXT:    s_mul_i32 s17, s17, s24
-; GCN-NEXT:    s_mul_hi_u32 s26, s23, s17
-; GCN-NEXT:    s_mul_i32 s27, s23, s17
-; GCN-NEXT:    s_mul_i32 s29, s24, s25
-; GCN-NEXT:    s_mul_hi_u32 s17, s24, s17
-; GCN-NEXT:    s_mul_hi_u32 s28, s24, s25
+; GCN-NEXT:    s_add_u32 s27, s22, s23
+; GCN-NEXT:    s_cselect_b64 s[22:23], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[22:23], 0
+; GCN-NEXT:    s_addc_u32 s25, s25, s26
+; GCN-NEXT:    s_mul_i32 s22, s17, s25
+; GCN-NEXT:    s_mul_hi_u32 s23, s17, s27
+; GCN-NEXT:    s_add_i32 s22, s23, s22
+; GCN-NEXT:    s_mul_i32 s24, s24, s27
+; GCN-NEXT:    s_add_i32 s22, s22, s24
+; GCN-NEXT:    s_mul_i32 s17, s17, s27
+; GCN-NEXT:    s_mul_hi_u32 s24, s25, s17
+; GCN-NEXT:    s_mul_i32 s26, s25, s17
+; GCN-NEXT:    s_mul_i32 s29, s27, s22
+; GCN-NEXT:    s_mul_hi_u32 s17, s27, s17
+; GCN-NEXT:    s_mul_hi_u32 s28, s27, s22
 ; GCN-NEXT:    s_add_u32 s17, s17, s29
 ; GCN-NEXT:    s_addc_u32 s28, 0, s28
-; GCN-NEXT:    s_add_u32 s17, s17, s27
-; GCN-NEXT:    s_mul_hi_u32 s22, s23, s25
-; GCN-NEXT:    s_addc_u32 s17, s28, s26
-; GCN-NEXT:    s_addc_u32 s22, s22, 0
-; GCN-NEXT:    s_mul_i32 s25, s23, s25
-; GCN-NEXT:    s_add_u32 s17, s17, s25
-; GCN-NEXT:    s_addc_u32 s22, 0, s22
-; GCN-NEXT:    s_add_u32 s17, s24, s17
-; GCN-NEXT:    s_addc_u32 s26, s23, s22
+; GCN-NEXT:    s_add_u32 s17, s17, s26
+; GCN-NEXT:    s_mul_hi_u32 s23, s25, s22
+; GCN-NEXT:    s_addc_u32 s17, s28, s24
+; GCN-NEXT:    s_addc_u32 s23, s23, 0
+; GCN-NEXT:    s_mul_i32 s22, s25, s22
+; GCN-NEXT:    s_add_u32 s17, s17, s22
+; GCN-NEXT:    s_addc_u32 s24, 0, s23
+; GCN-NEXT:    s_add_u32 s17, s27, s17
+; GCN-NEXT:    s_cselect_b64 s[22:23], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[22:23], 0
+; GCN-NEXT:    s_addc_u32 s26, s25, s24
 ; GCN-NEXT:    s_ashr_i32 s22, s19, 31
 ; GCN-NEXT:    s_add_u32 s24, s18, s22
 ; GCN-NEXT:    s_mov_b32 s23, s22
@@ -4985,9 +5021,11 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_mul_i32 s17, s6, s17
 ; GCN-NEXT:    s_sub_u32 s17, s24, s17
 ; GCN-NEXT:    s_cselect_b64 s[26:27], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[26:27], 0
 ; GCN-NEXT:    s_subb_u32 s24, s28, s7
 ; GCN-NEXT:    s_sub_u32 s30, s17, s6
 ; GCN-NEXT:    s_cselect_b64 s[28:29], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[28:29], 0
 ; GCN-NEXT:    s_subb_u32 s31, s24, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s31, s7
 ; GCN-NEXT:    s_cselect_b32 s33, -1, 0
@@ -4997,10 +5035,12 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_cselect_b32 s33, s34, s33
 ; GCN-NEXT:    s_cmp_lg_u64 s[28:29], 0
 ; GCN-NEXT:    s_subb_u32 s24, s24, s7
-; GCN-NEXT:    s_sub_u32 s28, s30, s6
+; GCN-NEXT:    s_sub_u32 s34, s30, s6
+; GCN-NEXT:    s_cselect_b64 s[28:29], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[28:29], 0
 ; GCN-NEXT:    s_subb_u32 s24, s24, 0
 ; GCN-NEXT:    s_cmp_lg_u32 s33, 0
-; GCN-NEXT:    s_cselect_b32 s28, s28, s30
+; GCN-NEXT:    s_cselect_b32 s28, s34, s30
 ; GCN-NEXT:    s_cselect_b32 s24, s24, s31
 ; GCN-NEXT:    s_cmp_lg_u64 s[26:27], 0
 ; GCN-NEXT:    s_subb_u32 s19, s25, s19
@@ -5051,7 +5091,7 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v0, s18
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v1, s19
 ; GCN-NEXT:    s_sub_u32 s13, 0, s18
-; GCN-NEXT:    s_subb_u32 s20, 0, s19
+; GCN-NEXT:    s_subb_u32 s22, 0, s19
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GCN-NEXT:    v_rcp_f32_e32 v0, v0
 ; GCN-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -5060,52 +5100,56 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0xcf800000, v0
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT:    v_readfirstlane_b32 s21, v1
-; GCN-NEXT:    v_readfirstlane_b32 s22, v0
-; GCN-NEXT:    s_mul_i32 s23, s13, s21
-; GCN-NEXT:    s_mul_hi_u32 s25, s13, s22
-; GCN-NEXT:    s_mul_i32 s24, s20, s22
-; GCN-NEXT:    s_add_i32 s23, s25, s23
-; GCN-NEXT:    s_add_i32 s23, s23, s24
-; GCN-NEXT:    s_mul_i32 s26, s13, s22
-; GCN-NEXT:    s_mul_i32 s25, s22, s23
-; GCN-NEXT:    s_mul_hi_u32 s27, s22, s26
-; GCN-NEXT:    s_mul_hi_u32 s24, s22, s23
+; GCN-NEXT:    v_readfirstlane_b32 s23, v1
+; GCN-NEXT:    v_readfirstlane_b32 s20, v0
+; GCN-NEXT:    s_mul_i32 s21, s13, s23
+; GCN-NEXT:    s_mul_hi_u32 s25, s13, s20
+; GCN-NEXT:    s_mul_i32 s24, s22, s20
+; GCN-NEXT:    s_add_i32 s21, s25, s21
+; GCN-NEXT:    s_add_i32 s21, s21, s24
+; GCN-NEXT:    s_mul_i32 s26, s13, s20
+; GCN-NEXT:    s_mul_i32 s25, s20, s21
+; GCN-NEXT:    s_mul_hi_u32 s27, s20, s26
+; GCN-NEXT:    s_mul_hi_u32 s24, s20, s21
 ; GCN-NEXT:    s_add_u32 s25, s27, s25
 ; GCN-NEXT:    s_addc_u32 s24, 0, s24
-; GCN-NEXT:    s_mul_hi_u32 s28, s21, s26
-; GCN-NEXT:    s_mul_i32 s26, s21, s26
+; GCN-NEXT:    s_mul_hi_u32 s28, s23, s26
+; GCN-NEXT:    s_mul_i32 s26, s23, s26
 ; GCN-NEXT:    s_add_u32 s25, s25, s26
-; GCN-NEXT:    s_mul_hi_u32 s27, s21, s23
+; GCN-NEXT:    s_mul_hi_u32 s27, s23, s21
 ; GCN-NEXT:    s_addc_u32 s24, s24, s28
 ; GCN-NEXT:    s_addc_u32 s25, s27, 0
-; GCN-NEXT:    s_mul_i32 s23, s21, s23
-; GCN-NEXT:    s_add_u32 s23, s24, s23
+; GCN-NEXT:    s_mul_i32 s21, s23, s21
+; GCN-NEXT:    s_add_u32 s21, s24, s21
 ; GCN-NEXT:    s_addc_u32 s24, 0, s25
-; GCN-NEXT:    s_add_u32 s22, s22, s23
-; GCN-NEXT:    s_addc_u32 s21, s21, s24
-; GCN-NEXT:    s_mul_i32 s23, s13, s21
-; GCN-NEXT:    s_mul_hi_u32 s24, s13, s22
-; GCN-NEXT:    s_add_i32 s23, s24, s23
-; GCN-NEXT:    s_mul_i32 s20, s20, s22
-; GCN-NEXT:    s_add_i32 s23, s23, s20
-; GCN-NEXT:    s_mul_i32 s13, s13, s22
-; GCN-NEXT:    s_mul_hi_u32 s24, s21, s13
-; GCN-NEXT:    s_mul_i32 s25, s21, s13
-; GCN-NEXT:    s_mul_i32 s27, s22, s23
-; GCN-NEXT:    s_mul_hi_u32 s13, s22, s13
-; GCN-NEXT:    s_mul_hi_u32 s26, s22, s23
+; GCN-NEXT:    s_add_u32 s25, s20, s21
+; GCN-NEXT:    s_cselect_b64 s[20:21], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[20:21], 0
+; GCN-NEXT:    s_addc_u32 s23, s23, s24
+; GCN-NEXT:    s_mul_i32 s20, s13, s23
+; GCN-NEXT:    s_mul_hi_u32 s21, s13, s25
+; GCN-NEXT:    s_add_i32 s20, s21, s20
+; GCN-NEXT:    s_mul_i32 s22, s22, s25
+; GCN-NEXT:    s_add_i32 s20, s20, s22
+; GCN-NEXT:    s_mul_i32 s13, s13, s25
+; GCN-NEXT:    s_mul_hi_u32 s22, s23, s13
+; GCN-NEXT:    s_mul_i32 s24, s23, s13
+; GCN-NEXT:    s_mul_i32 s27, s25, s20
+; GCN-NEXT:    s_mul_hi_u32 s13, s25, s13
+; GCN-NEXT:    s_mul_hi_u32 s26, s25, s20
 ; GCN-NEXT:    s_add_u32 s13, s13, s27
 ; GCN-NEXT:    s_addc_u32 s26, 0, s26
-; GCN-NEXT:    s_add_u32 s13, s13, s25
-; GCN-NEXT:    s_mul_hi_u32 s20, s21, s23
-; GCN-NEXT:    s_addc_u32 s13, s26, s24
-; GCN-NEXT:    s_addc_u32 s20, s20, 0
-; GCN-NEXT:    s_mul_i32 s23, s21, s23
-; GCN-NEXT:    s_add_u32 s13, s13, s23
-; GCN-NEXT:    s_addc_u32 s20, 0, s20
-; GCN-NEXT:    s_add_u32 s13, s22, s13
-; GCN-NEXT:    s_addc_u32 s24, s21, s20
+; GCN-NEXT:    s_add_u32 s13, s13, s24
+; GCN-NEXT:    s_mul_hi_u32 s21, s23, s20
+; GCN-NEXT:    s_addc_u32 s13, s26, s22
+; GCN-NEXT:    s_addc_u32 s21, s21, 0
+; GCN-NEXT:    s_mul_i32 s20, s23, s20
+; GCN-NEXT:    s_add_u32 s13, s13, s20
+; GCN-NEXT:    s_addc_u32 s22, 0, s21
+; GCN-NEXT:    s_add_u32 s13, s25, s13
+; GCN-NEXT:    s_cselect_b64 s[20:21], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[20:21], 0
+; GCN-NEXT:    s_addc_u32 s24, s23, s22
 ; GCN-NEXT:    s_ashr_i32 s20, s15, 31
 ; GCN-NEXT:    s_add_u32 s22, s14, s20
 ; GCN-NEXT:    s_mov_b32 s21, s20
@@ -5134,9 +5178,11 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_mul_i32 s13, s18, s13
 ; GCN-NEXT:    s_sub_u32 s13, s22, s13
 ; GCN-NEXT:    s_cselect_b64 s[24:25], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[24:25], 0
 ; GCN-NEXT:    s_subb_u32 s22, s26, s19
 ; GCN-NEXT:    s_sub_u32 s28, s13, s18
 ; GCN-NEXT:    s_cselect_b64 s[26:27], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[26:27], 0
 ; GCN-NEXT:    s_subb_u32 s29, s22, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s29, s19
 ; GCN-NEXT:    s_cselect_b32 s30, -1, 0
@@ -5146,10 +5192,12 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_cselect_b32 s30, s31, s30
 ; GCN-NEXT:    s_cmp_lg_u64 s[26:27], 0
 ; GCN-NEXT:    s_subb_u32 s22, s22, s19
-; GCN-NEXT:    s_sub_u32 s26, s28, s18
+; GCN-NEXT:    s_sub_u32 s31, s28, s18
+; GCN-NEXT:    s_cselect_b64 s[26:27], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[26:27], 0
 ; GCN-NEXT:    s_subb_u32 s22, s22, 0
 ; GCN-NEXT:    s_cmp_lg_u32 s30, 0
-; GCN-NEXT:    s_cselect_b32 s26, s26, s28
+; GCN-NEXT:    s_cselect_b32 s26, s31, s28
 ; GCN-NEXT:    s_cselect_b32 s22, s22, s29
 ; GCN-NEXT:    s_cmp_lg_u64 s[24:25], 0
 ; GCN-NEXT:    s_subb_u32 s15, s23, s15
@@ -5209,7 +5257,7 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v0, s14
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v1, s15
 ; GCN-NEXT:    s_sub_u32 s9, 0, s14
-; GCN-NEXT:    s_subb_u32 s16, 0, s15
+; GCN-NEXT:    s_subb_u32 s18, 0, s15
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GCN-NEXT:    v_rcp_f32_e32 v0, v0
 ; GCN-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -5218,52 +5266,56 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0xcf800000, v0
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT:    v_readfirstlane_b32 s17, v1
-; GCN-NEXT:    v_readfirstlane_b32 s18, v0
-; GCN-NEXT:    s_mul_i32 s19, s9, s17
-; GCN-NEXT:    s_mul_hi_u32 s21, s9, s18
-; GCN-NEXT:    s_mul_i32 s20, s16, s18
-; GCN-NEXT:    s_add_i32 s19, s21, s19
-; GCN-NEXT:    s_add_i32 s19, s19, s20
-; GCN-NEXT:    s_mul_i32 s22, s9, s18
-; GCN-NEXT:    s_mul_i32 s21, s18, s19
-; GCN-NEXT:    s_mul_hi_u32 s23, s18, s22
-; GCN-NEXT:    s_mul_hi_u32 s20, s18, s19
+; GCN-NEXT:    v_readfirstlane_b32 s19, v1
+; GCN-NEXT:    v_readfirstlane_b32 s16, v0
+; GCN-NEXT:    s_mul_i32 s17, s9, s19
+; GCN-NEXT:    s_mul_hi_u32 s21, s9, s16
+; GCN-NEXT:    s_mul_i32 s20, s18, s16
+; GCN-NEXT:    s_add_i32 s17, s21, s17
+; GCN-NEXT:    s_add_i32 s17, s17, s20
+; GCN-NEXT:    s_mul_i32 s22, s9, s16
+; GCN-NEXT:    s_mul_i32 s21, s16, s17
+; GCN-NEXT:    s_mul_hi_u32 s23, s16, s22
+; GCN-NEXT:    s_mul_hi_u32 s20, s16, s17
 ; GCN-NEXT:    s_add_u32 s21, s23, s21
 ; GCN-NEXT:    s_addc_u32 s20, 0, s20
-; GCN-NEXT:    s_mul_hi_u32 s24, s17, s22
-; GCN-NEXT:    s_mul_i32 s22, s17, s22
+; GCN-NEXT:    s_mul_hi_u32 s24, s19, s22
+; GCN-NEXT:    s_mul_i32 s22, s19, s22
 ; GCN-NEXT:    s_add_u32 s21, s21, s22
-; GCN-NEXT:    s_mul_hi_u32 s23, s17, s19
+; GCN-NEXT:    s_mul_hi_u32 s23, s19, s17
 ; GCN-NEXT:    s_addc_u32 s20, s20, s24
 ; GCN-NEXT:    s_addc_u32 s21, s23, 0
-; GCN-NEXT:    s_mul_i32 s19, s17, s19
-; GCN-NEXT:    s_add_u32 s19, s20, s19
+; GCN-NEXT:    s_mul_i32 s17, s19, s17
+; GCN-NEXT:    s_add_u32 s17, s20, s17
 ; GCN-NEXT:    s_addc_u32 s20, 0, s21
-; GCN-NEXT:    s_add_u32 s18, s18, s19
-; GCN-NEXT:    s_addc_u32 s17, s17, s20
-; GCN-NEXT:    s_mul_i32 s19, s9, s17
-; GCN-NEXT:    s_mul_hi_u32 s20, s9, s18
-; GCN-NEXT:    s_add_i32 s19, s20, s19
-; GCN-NEXT:    s_mul_i32 s16, s16, s18
-; GCN-NEXT:    s_add_i32 s19, s19, s16
-; GCN-NEXT:    s_mul_i32 s9, s9, s18
-; GCN-NEXT:    s_mul_hi_u32 s20, s17, s9
-; GCN-NEXT:    s_mul_i32 s21, s17, s9
-; GCN-NEXT:    s_mul_i32 s23, s18, s19
-; GCN-NEXT:    s_mul_hi_u32 s9, s18, s9
-; GCN-NEXT:    s_mul_hi_u32 s22, s18, s19
+; GCN-NEXT:    s_add_u32 s21, s16, s17
+; GCN-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[16:17], 0
+; GCN-NEXT:    s_addc_u32 s19, s19, s20
+; GCN-NEXT:    s_mul_i32 s16, s9, s19
+; GCN-NEXT:    s_mul_hi_u32 s17, s9, s21
+; GCN-NEXT:    s_add_i32 s16, s17, s16
+; GCN-NEXT:    s_mul_i32 s18, s18, s21
+; GCN-NEXT:    s_add_i32 s16, s16, s18
+; GCN-NEXT:    s_mul_i32 s9, s9, s21
+; GCN-NEXT:    s_mul_hi_u32 s18, s19, s9
+; GCN-NEXT:    s_mul_i32 s20, s19, s9
+; GCN-NEXT:    s_mul_i32 s23, s21, s16
+; GCN-NEXT:    s_mul_hi_u32 s9, s21, s9
+; GCN-NEXT:    s_mul_hi_u32 s22, s21, s16
 ; GCN-NEXT:    s_add_u32 s9, s9, s23
 ; GCN-NEXT:    s_addc_u32 s22, 0, s22
-; GCN-NEXT:    s_add_u32 s9, s9, s21
-; GCN-NEXT:    s_mul_hi_u32 s16, s17, s19
-; GCN-NEXT:    s_addc_u32 s9, s22, s20
-; GCN-NEXT:    s_addc_u32 s16, s16, 0
-; GCN-NEXT:    s_mul_i32 s19, s17, s19
-; GCN-NEXT:    s_add_u32 s9, s9, s19
-; GCN-NEXT:    s_addc_u32 s16, 0, s16
-; GCN-NEXT:    s_add_u32 s9, s18, s9
-; GCN-NEXT:    s_addc_u32 s20, s17, s16
+; GCN-NEXT:    s_add_u32 s9, s9, s20
+; GCN-NEXT:    s_mul_hi_u32 s17, s19, s16
+; GCN-NEXT:    s_addc_u32 s9, s22, s18
+; GCN-NEXT:    s_addc_u32 s17, s17, 0
+; GCN-NEXT:    s_mul_i32 s16, s19, s16
+; GCN-NEXT:    s_add_u32 s9, s9, s16
+; GCN-NEXT:    s_addc_u32 s18, 0, s17
+; GCN-NEXT:    s_add_u32 s9, s21, s9
+; GCN-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[16:17], 0
+; GCN-NEXT:    s_addc_u32 s20, s19, s18
 ; GCN-NEXT:    s_ashr_i32 s16, s11, 31
 ; GCN-NEXT:    s_add_u32 s18, s10, s16
 ; GCN-NEXT:    s_mov_b32 s17, s16
@@ -5292,9 +5344,11 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_mul_i32 s9, s14, s9
 ; GCN-NEXT:    s_sub_u32 s9, s18, s9
 ; GCN-NEXT:    s_cselect_b64 s[20:21], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[20:21], 0
 ; GCN-NEXT:    s_subb_u32 s18, s22, s15
 ; GCN-NEXT:    s_sub_u32 s24, s9, s14
 ; GCN-NEXT:    s_cselect_b64 s[22:23], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[22:23], 0
 ; GCN-NEXT:    s_subb_u32 s25, s18, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s25, s15
 ; GCN-NEXT:    s_cselect_b32 s26, -1, 0
@@ -5304,10 +5358,12 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_cselect_b32 s26, s27, s26
 ; GCN-NEXT:    s_cmp_lg_u64 s[22:23], 0
 ; GCN-NEXT:    s_subb_u32 s18, s18, s15
-; GCN-NEXT:    s_sub_u32 s22, s24, s14
+; GCN-NEXT:    s_sub_u32 s27, s24, s14
+; GCN-NEXT:    s_cselect_b64 s[22:23], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[22:23], 0
 ; GCN-NEXT:    s_subb_u32 s18, s18, 0
 ; GCN-NEXT:    s_cmp_lg_u32 s26, 0
-; GCN-NEXT:    s_cselect_b32 s22, s22, s24
+; GCN-NEXT:    s_cselect_b32 s22, s27, s24
 ; GCN-NEXT:    s_cselect_b32 s18, s18, s25
 ; GCN-NEXT:    s_cmp_lg_u64 s[20:21], 0
 ; GCN-NEXT:    s_subb_u32 s11, s19, s11
@@ -5364,7 +5420,7 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v0, s10
 ; GCN-NEXT:    v_cvt_f32_u32_e32 v1, s11
 ; GCN-NEXT:    s_sub_u32 s3, 0, s10
-; GCN-NEXT:    s_subb_u32 s12, 0, s11
+; GCN-NEXT:    s_subb_u32 s14, 0, s11
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GCN-NEXT:    v_rcp_f32_e32 v0, v0
 ; GCN-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -5373,52 +5429,56 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    v_madmk_f32 v0, v1, 0xcf800000, v0
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GCN-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT:    v_readfirstlane_b32 s13, v1
-; GCN-NEXT:    v_readfirstlane_b32 s14, v0
-; GCN-NEXT:    s_mul_i32 s15, s3, s13
-; GCN-NEXT:    s_mul_hi_u32 s17, s3, s14
-; GCN-NEXT:    s_mul_i32 s16, s12, s14
-; GCN-NEXT:    s_add_i32 s15, s17, s15
-; GCN-NEXT:    s_add_i32 s15, s15, s16
-; GCN-NEXT:    s_mul_i32 s18, s3, s14
-; GCN-NEXT:    s_mul_i32 s17, s14, s15
-; GCN-NEXT:    s_mul_hi_u32 s19, s14, s18
-; GCN-NEXT:    s_mul_hi_u32 s16, s14, s15
+; GCN-NEXT:    v_readfirstlane_b32 s15, v1
+; GCN-NEXT:    v_readfirstlane_b32 s12, v0
+; GCN-NEXT:    s_mul_i32 s13, s3, s15
+; GCN-NEXT:    s_mul_hi_u32 s17, s3, s12
+; GCN-NEXT:    s_mul_i32 s16, s14, s12
+; GCN-NEXT:    s_add_i32 s13, s17, s13
+; GCN-NEXT:    s_add_i32 s13, s13, s16
+; GCN-NEXT:    s_mul_i32 s18, s3, s12
+; GCN-NEXT:    s_mul_i32 s17, s12, s13
+; GCN-NEXT:    s_mul_hi_u32 s19, s12, s18
+; GCN-NEXT:    s_mul_hi_u32 s16, s12, s13
 ; GCN-NEXT:    s_add_u32 s17, s19, s17
 ; GCN-NEXT:    s_addc_u32 s16, 0, s16
-; GCN-NEXT:    s_mul_hi_u32 s20, s13, s18
-; GCN-NEXT:    s_mul_i32 s18, s13, s18
+; GCN-NEXT:    s_mul_hi_u32 s20, s15, s18
+; GCN-NEXT:    s_mul_i32 s18, s15, s18
 ; GCN-NEXT:    s_add_u32 s17, s17, s18
-; GCN-NEXT:    s_mul_hi_u32 s19, s13, s15
+; GCN-NEXT:    s_mul_hi_u32 s19, s15, s13
 ; GCN-NEXT:    s_addc_u32 s16, s16, s20
 ; GCN-NEXT:    s_addc_u32 s17, s19, 0
-; GCN-NEXT:    s_mul_i32 s15, s13, s15
-; GCN-NEXT:    s_add_u32 s15, s16, s15
+; GCN-NEXT:    s_mul_i32 s13, s15, s13
+; GCN-NEXT:    s_add_u32 s13, s16, s13
 ; GCN-NEXT:    s_addc_u32 s16, 0, s17
-; GCN-NEXT:    s_add_u32 s14, s14, s15
-; GCN-NEXT:    s_addc_u32 s13, s13, s16
-; GCN-NEXT:    s_mul_i32 s15, s3, s13
-; GCN-NEXT:    s_mul_hi_u32 s16, s3, s14
-; GCN-NEXT:    s_add_i32 s15, s16, s15
-; GCN-NEXT:    s_mul_i32 s12, s12, s14
-; GCN-NEXT:    s_add_i32 s15, s15, s12
-; GCN-NEXT:    s_mul_i32 s3, s3, s14
-; GCN-NEXT:    s_mul_hi_u32 s16, s13, s3
-; GCN-NEXT:    s_mul_i32 s17, s13, s3
-; GCN-NEXT:    s_mul_i32 s19, s14, s15
-; GCN-NEXT:    s_mul_hi_u32 s3, s14, s3
-; GCN-NEXT:    s_mul_hi_u32 s18, s14, s15
+; GCN-NEXT:    s_add_u32 s17, s12, s13
+; GCN-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[12:13], 0
+; GCN-NEXT:    s_addc_u32 s15, s15, s16
+; GCN-NEXT:    s_mul_i32 s12, s3, s15
+; GCN-NEXT:    s_mul_hi_u32 s13, s3, s17
+; GCN-NEXT:    s_add_i32 s12, s13, s12
+; GCN-NEXT:    s_mul_i32 s14, s14, s17
+; GCN-NEXT:    s_add_i32 s12, s12, s14
+; GCN-NEXT:    s_mul_i32 s3, s3, s17
+; GCN-NEXT:    s_mul_hi_u32 s14, s15, s3
+; GCN-NEXT:    s_mul_i32 s16, s15, s3
+; GCN-NEXT:    s_mul_i32 s19, s17, s12
+; GCN-NEXT:    s_mul_hi_u32 s3, s17, s3
+; GCN-NEXT:    s_mul_hi_u32 s18, s17, s12
 ; GCN-NEXT:    s_add_u32 s3, s3, s19
 ; GCN-NEXT:    s_addc_u32 s18, 0, s18
-; GCN-NEXT:    s_add_u32 s3, s3, s17
-; GCN-NEXT:    s_mul_hi_u32 s12, s13, s15
-; GCN-NEXT:    s_addc_u32 s3, s18, s16
-; GCN-NEXT:    s_addc_u32 s12, s12, 0
-; GCN-NEXT:    s_mul_i32 s15, s13, s15
-; GCN-NEXT:    s_add_u32 s3, s3, s15
-; GCN-NEXT:    s_addc_u32 s12, 0, s12
-; GCN-NEXT:    s_add_u32 s3, s14, s3
-; GCN-NEXT:    s_addc_u32 s16, s13, s12
+; GCN-NEXT:    s_add_u32 s3, s3, s16
+; GCN-NEXT:    s_mul_hi_u32 s13, s15, s12
+; GCN-NEXT:    s_addc_u32 s3, s18, s14
+; GCN-NEXT:    s_addc_u32 s13, s13, 0
+; GCN-NEXT:    s_mul_i32 s12, s15, s12
+; GCN-NEXT:    s_add_u32 s3, s3, s12
+; GCN-NEXT:    s_addc_u32 s14, 0, s13
+; GCN-NEXT:    s_add_u32 s3, s17, s3
+; GCN-NEXT:    s_cselect_b64 s[12:13], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[12:13], 0
+; GCN-NEXT:    s_addc_u32 s16, s15, s14
 ; GCN-NEXT:    s_ashr_i32 s12, s5, 31
 ; GCN-NEXT:    s_add_u32 s14, s4, s12
 ; GCN-NEXT:    s_mov_b32 s13, s12
@@ -5447,9 +5507,11 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_mul_i32 s3, s10, s3
 ; GCN-NEXT:    s_sub_u32 s3, s14, s3
 ; GCN-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; GCN-NEXT:    s_subb_u32 s14, s18, s11
 ; GCN-NEXT:    s_sub_u32 s20, s3, s10
 ; GCN-NEXT:    s_cselect_b64 s[18:19], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[18:19], 0
 ; GCN-NEXT:    s_subb_u32 s21, s14, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s21, s11
 ; GCN-NEXT:    s_cselect_b32 s22, -1, 0
@@ -5459,10 +5521,12 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; GCN-NEXT:    s_cselect_b32 s22, s23, s22
 ; GCN-NEXT:    s_cmp_lg_u64 s[18:19], 0
 ; GCN-NEXT:    s_subb_u32 s14, s14, s11
-; GCN-NEXT:    s_sub_u32 s18, s20, s10
+; GCN-NEXT:    s_sub_u32 s23, s20, s10
+; GCN-NEXT:    s_cselect_b64 s[18:19], -1, 0
+; GCN-NEXT:    s_cmp_lg_u64 s[18:19], 0
 ; GCN-NEXT:    s_subb_u32 s14, s14, 0
 ; GCN-NEXT:    s_cmp_lg_u32 s22, 0
-; GCN-NEXT:    s_cselect_b32 s18, s18, s20
+; GCN-NEXT:    s_cselect_b32 s18, s23, s20
 ; GCN-NEXT:    s_cselect_b32 s14, s14, s21
 ; GCN-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; GCN-NEXT:    s_subb_u32 s5, s15, s5
@@ -6235,9 +6299,11 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; TONGA-NEXT:    v_readfirstlane_b32 s14, v8
 ; TONGA-NEXT:    s_sub_u32 s12, s12, s14
 ; TONGA-NEXT:    s_cselect_b64 s[14:15], -1, 0
+; TONGA-NEXT:    s_cmp_lg_u64 s[14:15], 0
 ; TONGA-NEXT:    s_subb_u32 s1, s1, s7
 ; TONGA-NEXT:    s_sub_u32 s18, s12, s6
 ; TONGA-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; TONGA-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; TONGA-NEXT:    s_subb_u32 s19, s1, 0
 ; TONGA-NEXT:    s_cmp_ge_u32 s19, s7
 ; TONGA-NEXT:    s_cselect_b32 s20, -1, 0
@@ -6247,10 +6313,12 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
 ; TONGA-NEXT:    s_cselect_b32 s20, s21, s20
 ; TONGA-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; TONGA-NEXT:    s_subb_u32 s1, s1, s7
-; TONGA-NEXT:    s_sub_u32 s16, s18, s6
+; TONGA-NEXT:    s_sub_u32 s21, s18, s6
+; TONGA-NEXT:    s_cselect_b64 s[16:17], -1, 0
+; TONGA-NEXT:    s_cmp_lg_u64 s[16:17], 0
 ; TONGA-NEXT:    s_subb_u32 s1, s1, 0
 ; TONGA-NEXT:    s_cmp_lg_u32 s20, 0
-; TONGA-NEXT:    s_cselect_b32 s16, s16, s18
+; TONGA-NEXT:    s_cselect_b32 s16, s21, s18
 ; TONGA-NEXT:    s_cselect_b32 s1, s1, s19
 ; TONGA-NEXT:    s_cmp_lg_u64 s[14:15], 0
 ; TONGA-NEXT:    s_subb_u32 s3, s13, s3
diff --git a/llvm/test/CodeGen/AMDGPU/srem64.ll b/llvm/test/CodeGen/AMDGPU/srem64.ll
index ea9bb04..33b0a5d 100644
--- a/llvm/test/CodeGen/AMDGPU/srem64.ll
+++ b/llvm/test/CodeGen/AMDGPU/srem64.ll
@@ -51,9 +51,10 @@ define amdgpu_kernel void @s_test_srem(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GCN-NEXT:    s_addc_u32 s13, 0, s14
 ; GCN-NEXT:    s_add_u32 s14, s0, s1
 ; GCN-NEXT:    v_mov_b32_e32 v0, s14
-; GCN-NEXT:    v_mul_hi_u32 v0, s10, v0
 ; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GCN-NEXT:    v_mul_hi_u32 v0, s10, v0
 ; GCN-NEXT:    s_or_b32 s0, s0, s1
+; GCN-NEXT:    s_cmp_lg_u32 s0, 0
 ; GCN-NEXT:    s_addc_u32 s12, s12, s13
 ; GCN-NEXT:    s_mul_i32 s0, s10, s12
 ; GCN-NEXT:    v_readfirstlane_b32 s1, v0
@@ -84,6 +85,7 @@ define amdgpu_kernel void @s_test_srem(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GCN-NEXT:    s_add_u32 s11, s14, s0
 ; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GCN-NEXT:    s_or_b32 s0, s0, s1
+; GCN-NEXT:    s_cmp_lg_u32 s0, 0
 ; GCN-NEXT:    s_addc_u32 s1, s12, s10
 ; GCN-NEXT:    v_mov_b32_e32 v0, s1
 ; GCN-NEXT:    v_mul_hi_u32 v1, s6, v0
@@ -113,43 +115,46 @@ define amdgpu_kernel void @s_test_srem(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GCN-NEXT:    v_readfirstlane_b32 s10, v0
 ; GCN-NEXT:    s_add_i32 s5, s10, s5
 ; GCN-NEXT:    s_mul_i32 s10, s9, s4
-; GCN-NEXT:    s_add_i32 s12, s5, s10
-; GCN-NEXT:    s_sub_i32 s10, s7, s12
+; GCN-NEXT:    s_add_i32 s10, s5, s10
+; GCN-NEXT:    s_sub_i32 s11, s7, s10
 ; GCN-NEXT:    s_mul_i32 s4, s8, s4
 ; GCN-NEXT:    s_sub_u32 s6, s6, s4
 ; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
-; GCN-NEXT:    s_or_b32 s11, s4, s5
-; GCN-NEXT:    s_subb_u32 s13, s10, s9
-; GCN-NEXT:    s_sub_u32 s14, s6, s8
-; GCN-NEXT:    s_cselect_b64 s[10:11], -1, 0
-; GCN-NEXT:    s_or_b32 s15, s10, s11
-; GCN-NEXT:    s_subb_u32 s15, s13, 0
-; GCN-NEXT:    s_cmp_ge_u32 s15, s9
-; GCN-NEXT:    s_cselect_b32 s16, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s14, s8
-; GCN-NEXT:    s_cselect_b32 s17, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s15, s9
-; GCN-NEXT:    s_cselect_b32 s16, s17, s16
-; GCN-NEXT:    s_or_b32 s10, s10, s11
-; GCN-NEXT:    s_subb_u32 s13, s13, s9
-; GCN-NEXT:    s_sub_u32 s17, s14, s8
-; GCN-NEXT:    s_cselect_b64 s[10:11], -1, 0
-; GCN-NEXT:    s_or_b32 s10, s10, s11
-; GCN-NEXT:    s_subb_u32 s10, s13, 0
-; GCN-NEXT:    s_cmp_lg_u32 s16, 0
-; GCN-NEXT:    s_cselect_b32 s11, s17, s14
-; GCN-NEXT:    s_cselect_b32 s10, s10, s15
+; GCN-NEXT:    s_or_b32 s12, s4, s5
+; GCN-NEXT:    s_cmp_lg_u32 s12, 0
+; GCN-NEXT:    s_subb_u32 s11, s11, s9
+; GCN-NEXT:    s_sub_u32 s13, s6, s8
+; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GCN-NEXT:    s_or_b32 s4, s4, s5
-; GCN-NEXT:    s_subb_u32 s4, s7, s12
-; GCN-NEXT:    s_cmp_ge_u32 s4, s9
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
+; GCN-NEXT:    s_subb_u32 s14, s11, 0
+; GCN-NEXT:    s_cmp_ge_u32 s14, s9
 ; GCN-NEXT:    s_cselect_b32 s5, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s13, s8
+; GCN-NEXT:    s_cselect_b32 s15, -1, 0
+; GCN-NEXT:    s_cmp_eq_u32 s14, s9
+; GCN-NEXT:    s_cselect_b32 s15, s15, s5
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
+; GCN-NEXT:    s_subb_u32 s11, s11, s9
+; GCN-NEXT:    s_sub_u32 s16, s13, s8
+; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GCN-NEXT:    s_or_b32 s4, s4, s5
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
+; GCN-NEXT:    s_subb_u32 s4, s11, 0
+; GCN-NEXT:    s_cmp_lg_u32 s15, 0
+; GCN-NEXT:    s_cselect_b32 s5, s16, s13
+; GCN-NEXT:    s_cselect_b32 s4, s4, s14
+; GCN-NEXT:    s_cmp_lg_u32 s12, 0
+; GCN-NEXT:    s_subb_u32 s7, s7, s10
+; GCN-NEXT:    s_cmp_ge_u32 s7, s9
+; GCN-NEXT:    s_cselect_b32 s10, -1, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s6, s8
-; GCN-NEXT:    s_cselect_b32 s7, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s4, s9
-; GCN-NEXT:    s_cselect_b32 s5, s7, s5
-; GCN-NEXT:    s_cmp_lg_u32 s5, 0
-; GCN-NEXT:    s_cselect_b32 s4, s10, s4
-; GCN-NEXT:    s_cselect_b32 s5, s11, s6
+; GCN-NEXT:    s_cselect_b32 s8, -1, 0
+; GCN-NEXT:    s_cmp_eq_u32 s7, s9
+; GCN-NEXT:    s_cselect_b32 s8, s8, s10
+; GCN-NEXT:    s_cmp_lg_u32 s8, 0
+; GCN-NEXT:    s_cselect_b32 s4, s4, s7
+; GCN-NEXT:    s_cselect_b32 s5, s5, s6
 ; GCN-NEXT:    v_mov_b32_e32 v0, s5
 ; GCN-NEXT:    v_mov_b32_e32 v1, s4
 ; GCN-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
@@ -182,6 +187,7 @@ define amdgpu_kernel void @s_test_srem(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GCN-IR-NEXT:    s_add_u32 s14, s12, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s8, s8, s9
+; GCN-IR-NEXT:    s_cmp_lg_u32 s8, 0
 ; GCN-IR-NEXT:    s_addc_u32 s8, s13, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s12, 63, s12
@@ -215,6 +221,7 @@ define amdgpu_kernel void @s_test_srem(ptr addrspace(1) %out, i64 %x, i64 %y) {
 ; GCN-IR-NEXT:    s_add_u32 s16, s16, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[18:19], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s18, s18, s19
+; GCN-IR-NEXT:    s_cmp_lg_u32 s18, 0
 ; GCN-IR-NEXT:    s_addc_u32 s17, s17, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[18:19], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[10:11], s[4:5]
@@ -1009,9 +1016,10 @@ define amdgpu_kernel void @s_test_srem33_64(ptr addrspace(1) %out, i64 %x, i64 %
 ; GCN-NEXT:    s_addc_u32 s13, 0, s14
 ; GCN-NEXT:    s_add_u32 s14, s8, s9
 ; GCN-NEXT:    v_mov_b32_e32 v0, s14
-; GCN-NEXT:    v_mul_hi_u32 v0, s10, v0
 ; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; GCN-NEXT:    v_mul_hi_u32 v0, s10, v0
 ; GCN-NEXT:    s_or_b32 s8, s8, s9
+; GCN-NEXT:    s_cmp_lg_u32 s8, 0
 ; GCN-NEXT:    s_addc_u32 s12, s12, s13
 ; GCN-NEXT:    s_mul_i32 s8, s10, s12
 ; GCN-NEXT:    v_readfirstlane_b32 s9, v0
@@ -1042,6 +1050,7 @@ define amdgpu_kernel void @s_test_srem33_64(ptr addrspace(1) %out, i64 %x, i64 %
 ; GCN-NEXT:    s_add_u32 s11, s14, s8
 ; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GCN-NEXT:    s_or_b32 s8, s8, s9
+; GCN-NEXT:    s_cmp_lg_u32 s8, 0
 ; GCN-NEXT:    s_addc_u32 s10, s12, s10
 ; GCN-NEXT:    s_ashr_i32 s8, s7, 31
 ; GCN-NEXT:    s_add_u32 s6, s6, s8
@@ -1074,43 +1083,46 @@ define amdgpu_kernel void @s_test_srem33_64(ptr addrspace(1) %out, i64 %x, i64 %
 ; GCN-NEXT:    v_readfirstlane_b32 s12, v0
 ; GCN-NEXT:    s_add_i32 s11, s12, s11
 ; GCN-NEXT:    s_mul_i32 s12, s5, s10
-; GCN-NEXT:    s_add_i32 s14, s11, s12
-; GCN-NEXT:    s_sub_i32 s12, s7, s14
+; GCN-NEXT:    s_add_i32 s12, s11, s12
+; GCN-NEXT:    s_sub_i32 s13, s7, s12
 ; GCN-NEXT:    s_mul_i32 s10, s4, s10
 ; GCN-NEXT:    s_sub_u32 s6, s6, s10
 ; GCN-NEXT:    s_cselect_b64 s[10:11], -1, 0
-; GCN-NEXT:    s_or_b32 s13, s10, s11
-; GCN-NEXT:    s_subb_u32 s15, s12, s5
-; GCN-NEXT:    s_sub_u32 s16, s6, s4
-; GCN-NEXT:    s_cselect_b64 s[12:13], -1, 0
-; GCN-NEXT:    s_or_b32 s17, s12, s13
-; GCN-NEXT:    s_subb_u32 s17, s15, 0
-; GCN-NEXT:    s_cmp_ge_u32 s17, s5
-; GCN-NEXT:    s_cselect_b32 s18, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s16, s4
-; GCN-NEXT:    s_cselect_b32 s19, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s17, s5
-; GCN-NEXT:    s_cselect_b32 s18, s19, s18
-; GCN-NEXT:    s_or_b32 s12, s12, s13
-; GCN-NEXT:    s_subb_u32 s15, s15, s5
-; GCN-NEXT:    s_sub_u32 s19, s16, s4
-; GCN-NEXT:    s_cselect_b64 s[12:13], -1, 0
-; GCN-NEXT:    s_or_b32 s12, s12, s13
-; GCN-NEXT:    s_subb_u32 s12, s15, 0
-; GCN-NEXT:    s_cmp_lg_u32 s18, 0
-; GCN-NEXT:    s_cselect_b32 s13, s19, s16
-; GCN-NEXT:    s_cselect_b32 s12, s12, s17
+; GCN-NEXT:    s_or_b32 s14, s10, s11
+; GCN-NEXT:    s_cmp_lg_u32 s14, 0
+; GCN-NEXT:    s_subb_u32 s13, s13, s5
+; GCN-NEXT:    s_sub_u32 s15, s6, s4
+; GCN-NEXT:    s_cselect_b64 s[10:11], -1, 0
+; GCN-NEXT:    s_or_b32 s10, s10, s11
+; GCN-NEXT:    s_cmp_lg_u32 s10, 0
+; GCN-NEXT:    s_subb_u32 s16, s13, 0
+; GCN-NEXT:    s_cmp_ge_u32 s16, s5
+; GCN-NEXT:    s_cselect_b32 s11, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s15, s4
+; GCN-NEXT:    s_cselect_b32 s17, -1, 0
+; GCN-NEXT:    s_cmp_eq_u32 s16, s5
+; GCN-NEXT:    s_cselect_b32 s17, s17, s11
+; GCN-NEXT:    s_cmp_lg_u32 s10, 0
+; GCN-NEXT:    s_subb_u32 s13, s13, s5
+; GCN-NEXT:    s_sub_u32 s18, s15, s4
+; GCN-NEXT:    s_cselect_b64 s[10:11], -1, 0
 ; GCN-NEXT:    s_or_b32 s10, s10, s11
-; GCN-NEXT:    s_subb_u32 s7, s7, s14
+; GCN-NEXT:    s_cmp_lg_u32 s10, 0
+; GCN-NEXT:    s_subb_u32 s10, s13, 0
+; GCN-NEXT:    s_cmp_lg_u32 s17, 0
+; GCN-NEXT:    s_cselect_b32 s11, s18, s15
+; GCN-NEXT:    s_cselect_b32 s10, s10, s16
+; GCN-NEXT:    s_cmp_lg_u32 s14, 0
+; GCN-NEXT:    s_subb_u32 s7, s7, s12
 ; GCN-NEXT:    s_cmp_ge_u32 s7, s5
-; GCN-NEXT:    s_cselect_b32 s10, -1, 0
+; GCN-NEXT:    s_cselect_b32 s12, -1, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s6, s4
 ; GCN-NEXT:    s_cselect_b32 s4, -1, 0
 ; GCN-NEXT:    s_cmp_eq_u32 s7, s5
-; GCN-NEXT:    s_cselect_b32 s4, s4, s10
+; GCN-NEXT:    s_cselect_b32 s4, s4, s12
 ; GCN-NEXT:    s_cmp_lg_u32 s4, 0
-; GCN-NEXT:    s_cselect_b32 s5, s12, s7
-; GCN-NEXT:    s_cselect_b32 s4, s13, s6
+; GCN-NEXT:    s_cselect_b32 s5, s10, s7
+; GCN-NEXT:    s_cselect_b32 s4, s11, s6
 ; GCN-NEXT:    s_xor_b64 s[4:5], s[4:5], s[8:9]
 ; GCN-NEXT:    s_sub_u32 s4, s4, s8
 ; GCN-NEXT:    s_subb_u32 s5, s5, s8
@@ -1158,6 +1170,7 @@ define amdgpu_kernel void @s_test_srem33_64(ptr addrspace(1) %out, i64 %x, i64 %
 ; GCN-IR-NEXT:    s_add_u32 s16, s14, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[10:11], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s10, s10, s11
+; GCN-IR-NEXT:    s_cmp_lg_u32 s10, 0
 ; GCN-IR-NEXT:    s_addc_u32 s10, s15, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[10:11], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s14, 63, s14
@@ -1191,6 +1204,7 @@ define amdgpu_kernel void @s_test_srem33_64(ptr addrspace(1) %out, i64 %x, i64 %
 ; GCN-IR-NEXT:    s_add_u32 s18, s18, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[20:21], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s20, s20, s21
+; GCN-IR-NEXT:    s_cmp_lg_u32 s20, 0
 ; GCN-IR-NEXT:    s_addc_u32 s19, s19, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[20:21], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[12:13], s[2:3]
@@ -1355,9 +1369,10 @@ define amdgpu_kernel void @s_test_srem_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    s_addc_u32 s10, 0, s11
 ; GCN-NEXT:    s_add_u32 s11, s6, s7
 ; GCN-NEXT:    v_mov_b32_e32 v0, s11
-; GCN-NEXT:    v_mul_hi_u32 v0, s2, v0
 ; GCN-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; GCN-NEXT:    v_mul_hi_u32 v0, s2, v0
 ; GCN-NEXT:    s_or_b32 s6, s6, s7
+; GCN-NEXT:    s_cmp_lg_u32 s6, 0
 ; GCN-NEXT:    s_addc_u32 s9, s9, s10
 ; GCN-NEXT:    s_mul_i32 s6, s2, s9
 ; GCN-NEXT:    v_readfirstlane_b32 s7, v0
@@ -1388,6 +1403,7 @@ define amdgpu_kernel void @s_test_srem_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    s_add_u32 s2, s11, s2
 ; GCN-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GCN-NEXT:    s_or_b32 s6, s6, s7
+; GCN-NEXT:    s_cmp_lg_u32 s6, 0
 ; GCN-NEXT:    s_addc_u32 s6, s9, s8
 ; GCN-NEXT:    v_mul_hi_u32 v1, s2, 24
 ; GCN-NEXT:    v_mul_hi_u32 v0, s6, 24
@@ -1402,42 +1418,45 @@ define amdgpu_kernel void @s_test_srem_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    s_mul_i32 s7, s5, s6
 ; GCN-NEXT:    s_mul_i32 s6, s4, s6
 ; GCN-NEXT:    v_readfirstlane_b32 s8, v0
-; GCN-NEXT:    s_add_i32 s10, s8, s7
-; GCN-NEXT:    s_sub_i32 s8, 0, s10
-; GCN-NEXT:    s_sub_u32 s11, 24, s6
+; GCN-NEXT:    s_add_i32 s8, s8, s7
+; GCN-NEXT:    s_sub_i32 s9, 0, s8
+; GCN-NEXT:    s_sub_u32 s10, 24, s6
+; GCN-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; GCN-NEXT:    s_or_b32 s11, s6, s7
+; GCN-NEXT:    s_cmp_lg_u32 s11, 0
+; GCN-NEXT:    s_subb_u32 s9, s9, s5
+; GCN-NEXT:    s_sub_u32 s12, s10, s4
 ; GCN-NEXT:    s_cselect_b64 s[6:7], -1, 0
-; GCN-NEXT:    s_or_b32 s9, s6, s7
-; GCN-NEXT:    s_subb_u32 s12, s8, s5
-; GCN-NEXT:    s_sub_u32 s13, s11, s4
-; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
-; GCN-NEXT:    s_or_b32 s14, s8, s9
-; GCN-NEXT:    s_subb_u32 s14, s12, 0
-; GCN-NEXT:    s_cmp_ge_u32 s14, s5
-; GCN-NEXT:    s_cselect_b32 s15, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s13, s4
-; GCN-NEXT:    s_cselect_b32 s16, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s14, s5
-; GCN-NEXT:    s_cselect_b32 s15, s16, s15
-; GCN-NEXT:    s_or_b32 s8, s8, s9
-; GCN-NEXT:    s_subb_u32 s12, s12, s5
-; GCN-NEXT:    s_sub_u32 s16, s13, s4
-; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
-; GCN-NEXT:    s_or_b32 s8, s8, s9
-; GCN-NEXT:    s_subb_u32 s8, s12, 0
-; GCN-NEXT:    s_cmp_lg_u32 s15, 0
-; GCN-NEXT:    s_cselect_b32 s9, s16, s13
-; GCN-NEXT:    s_cselect_b32 s8, s8, s14
 ; GCN-NEXT:    s_or_b32 s6, s6, s7
-; GCN-NEXT:    s_subb_u32 s6, 0, s10
-; GCN-NEXT:    s_cmp_ge_u32 s6, s5
+; GCN-NEXT:    s_cmp_lg_u32 s6, 0
+; GCN-NEXT:    s_subb_u32 s13, s9, 0
+; GCN-NEXT:    s_cmp_ge_u32 s13, s5
 ; GCN-NEXT:    s_cselect_b32 s7, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s11, s4
+; GCN-NEXT:    s_cmp_ge_u32 s12, s4
+; GCN-NEXT:    s_cselect_b32 s14, -1, 0
+; GCN-NEXT:    s_cmp_eq_u32 s13, s5
+; GCN-NEXT:    s_cselect_b32 s14, s14, s7
+; GCN-NEXT:    s_cmp_lg_u32 s6, 0
+; GCN-NEXT:    s_subb_u32 s9, s9, s5
+; GCN-NEXT:    s_sub_u32 s15, s12, s4
+; GCN-NEXT:    s_cselect_b64 s[6:7], -1, 0
+; GCN-NEXT:    s_or_b32 s6, s6, s7
+; GCN-NEXT:    s_cmp_lg_u32 s6, 0
+; GCN-NEXT:    s_subb_u32 s6, s9, 0
+; GCN-NEXT:    s_cmp_lg_u32 s14, 0
+; GCN-NEXT:    s_cselect_b32 s7, s15, s12
+; GCN-NEXT:    s_cselect_b32 s6, s6, s13
+; GCN-NEXT:    s_cmp_lg_u32 s11, 0
+; GCN-NEXT:    s_subb_u32 s8, 0, s8
+; GCN-NEXT:    s_cmp_ge_u32 s8, s5
+; GCN-NEXT:    s_cselect_b32 s9, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s10, s4
 ; GCN-NEXT:    s_cselect_b32 s4, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s6, s5
-; GCN-NEXT:    s_cselect_b32 s4, s4, s7
+; GCN-NEXT:    s_cmp_eq_u32 s8, s5
+; GCN-NEXT:    s_cselect_b32 s4, s4, s9
 ; GCN-NEXT:    s_cmp_lg_u32 s4, 0
-; GCN-NEXT:    s_cselect_b32 s4, s8, s6
-; GCN-NEXT:    s_cselect_b32 s5, s9, s11
+; GCN-NEXT:    s_cselect_b32 s4, s6, s8
+; GCN-NEXT:    s_cselect_b32 s5, s7, s10
 ; GCN-NEXT:    v_mov_b32_e32 v0, s5
 ; GCN-NEXT:    v_mov_b32_e32 v1, s4
 ; GCN-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
@@ -1470,6 +1489,7 @@ define amdgpu_kernel void @s_test_srem_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s8, s2, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[10:11], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s9, s10, s11
+; GCN-IR-NEXT:    s_cmp_lg_u32 s9, 0
 ; GCN-IR-NEXT:    s_addc_u32 s3, s3, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[10:11], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s2, 63, s2
@@ -1502,6 +1522,7 @@ define amdgpu_kernel void @s_test_srem_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s14, s14, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[16:17], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s16, s16, s17
+; GCN-IR-NEXT:    s_cmp_lg_u32 s16, 0
 ; GCN-IR-NEXT:    s_addc_u32 s15, s15, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[16:17], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[8:9], s[6:7]
diff --git a/llvm/test/CodeGen/AMDGPU/uaddo.ll b/llvm/test/CodeGen/AMDGPU/uaddo.ll
index bdd22f25..bb5918b2 100644
--- a/llvm/test/CodeGen/AMDGPU/uaddo.ll
+++ b/llvm/test/CodeGen/AMDGPU/uaddo.ll
@@ -18,6 +18,7 @@ define amdgpu_kernel void @s_uaddo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
 ; SI-NEXT:    s_mov_b32 s5, s1
 ; SI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; SI-NEXT:    s_or_b32 s0, s0, s1
+; SI-NEXT:    s_cmp_lg_u32 s0, 0
 ; SI-NEXT:    s_addc_u32 s3, s3, s9
 ; SI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; SI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -34,9 +35,11 @@ define amdgpu_kernel void @s_uaddo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
 ; VI-NEXT:    s_add_u32 s2, s2, s4
-; VI-NEXT:    s_addc_u32 s3, s3, s5
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
 ; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; VI-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; VI-NEXT:    s_addc_u32 s3, s3, s5
+; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; VI-NEXT:    v_cndmask_b32_e64 v2, 0, 1, s[0:1]
 ; VI-NEXT:    v_mov_b32_e32 v3, s3
 ; VI-NEXT:    v_add_u32_e32 v2, vcc, s2, v2
@@ -50,12 +53,14 @@ define amdgpu_kernel void @s_uaddo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
 ; GFX9-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_add_u32 s4, s2, s6
-; GFX9-NEXT:    s_addc_u32 s5, s3, s7
+; GFX9-NEXT:    s_add_u32 s6, s2, s6
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX9-NEXT:    s_addc_u32 s4, s3, s7
 ; GFX9-NEXT:    s_cselect_b64 s[2:3], -1, 0
 ; GFX9-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[2:3]
-; GFX9-NEXT:    v_mov_b32_e32 v1, s5
-; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s4, v0
+; GFX9-NEXT:    v_mov_b32_e32 v1, s4
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s6, v0
 ; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
 ; GFX9-NEXT:    global_store_dwordx2 v2, v[0:1], s[0:1]
 ; GFX9-NEXT:    s_endpgm
@@ -68,6 +73,8 @@ define amdgpu_kernel void @s_uaddo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
 ; GFX10-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    s_add_u32 s2, s2, s6
+; GFX10-NEXT:    s_cselect_b32 s4, -1, 0
+; GFX10-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX10-NEXT:    s_addc_u32 s3, s3, s7
 ; GFX10-NEXT:    s_cselect_b32 s4, -1, 0
 ; GFX10-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s4
@@ -84,12 +91,14 @@ define amdgpu_kernel void @s_uaddo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
 ; GFX11-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-NEXT:    s_add_u32 s2, s2, s4
+; GFX11-NEXT:    s_cselect_b32 s4, -1, 0
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
+; GFX11-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX11-NEXT:    s_addc_u32 s3, s3, s5
 ; GFX11-NEXT:    s_cselect_b32 s4, -1, 0
-; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
 ; GFX11-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s4
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1)
 ; GFX11-NEXT:    v_add_co_u32 v0, s2, s2, v0
-; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX11-NEXT:    v_add_co_ci_u32_e64 v1, null, s3, 0, s2
 ; GFX11-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
 ; GFX11-NEXT:    s_endpgm
@@ -435,6 +444,7 @@ define amdgpu_kernel void @s_uaddo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
 ; SI-NEXT:    s_add_u32 s4, s4, s6
 ; SI-NEXT:    s_cselect_b64 s[12:13], -1, 0
 ; SI-NEXT:    s_or_b32 s6, s12, s13
+; SI-NEXT:    s_cmp_lg_u32 s6, 0
 ; SI-NEXT:    s_addc_u32 s5, s5, s7
 ; SI-NEXT:    s_mov_b32 s8, s0
 ; SI-NEXT:    s_mov_b32 s9, s1
@@ -455,14 +465,16 @@ define amdgpu_kernel void @s_uaddo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx8 s[0:7], s[4:5], 0x24
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v2, s2
+; VI-NEXT:    s_add_u32 s2, s4, s6
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
-; VI-NEXT:    s_add_u32 s0, s4, s6
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_addc_u32 s1, s5, s7
-; VI-NEXT:    v_mov_b32_e32 v4, s0
-; VI-NEXT:    v_mov_b32_e32 v5, s1
 ; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
-; VI-NEXT:    v_mov_b32_e32 v2, s2
+; VI-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; VI-NEXT:    s_addc_u32 s0, s5, s7
+; VI-NEXT:    v_mov_b32_e32 v4, s2
+; VI-NEXT:    v_mov_b32_e32 v5, s0
+; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; VI-NEXT:    v_mov_b32_e32 v3, s3
 ; VI-NEXT:    flat_store_dwordx2 v[0:1], v[4:5]
 ; VI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -474,10 +486,12 @@ define amdgpu_kernel void @s_uaddo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
 ; GFX9-NEXT:    s_load_dwordx8 s[8:15], s[4:5], 0x24
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_add_u32 s0, s12, s14
-; GFX9-NEXT:    s_addc_u32 s1, s13, s15
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    s_add_u32 s2, s12, s14
+; GFX9-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    s_addc_u32 s0, s13, s15
+; GFX9-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s0
 ; GFX9-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GFX9-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s[0:1]
 ; GFX9-NEXT:    global_store_dwordx2 v2, v[0:1], s[8:9]
@@ -490,8 +504,10 @@ define amdgpu_kernel void @s_uaddo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
 ; GFX10-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    s_add_u32 s0, s12, s14
-; GFX10-NEXT:    s_addc_u32 s1, s13, s15
+; GFX10-NEXT:    s_cselect_b32 s1, -1, 0
 ; GFX10-NEXT:    v_mov_b32_e32 v0, s0
+; GFX10-NEXT:    s_cmp_lg_u32 s1, 0
+; GFX10-NEXT:    s_addc_u32 s1, s13, s15
 ; GFX10-NEXT:    s_cselect_b32 s0, -1, 0
 ; GFX10-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX10-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s0
@@ -504,8 +520,10 @@ define amdgpu_kernel void @s_uaddo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
 ; GFX11-NEXT:    s_load_b256 s[0:7], s[4:5], 0x24
 ; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-NEXT:    s_add_u32 s4, s4, s6
-; GFX11-NEXT:    s_addc_u32 s5, s5, s7
+; GFX11-NEXT:    s_cselect_b32 s6, -1, 0
 ; GFX11-NEXT:    v_mov_b32_e32 v0, s4
+; GFX11-NEXT:    s_cmp_lg_u32 s6, 0
+; GFX11-NEXT:    s_addc_u32 s5, s5, s7
 ; GFX11-NEXT:    s_cselect_b32 s4, -1, 0
 ; GFX11-NEXT:    v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s5
 ; GFX11-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s4
diff --git a/llvm/test/CodeGen/AMDGPU/udiv64.ll b/llvm/test/CodeGen/AMDGPU/udiv64.ll
index fd461ac..41199b0 100644
--- a/llvm/test/CodeGen/AMDGPU/udiv64.ll
+++ b/llvm/test/CodeGen/AMDGPU/udiv64.ll
@@ -148,6 +148,7 @@ define amdgpu_kernel void @s_test_udiv_i64(ptr addrspace(1) %out, i64 %x, i64 %y
 ; GCN-IR-NEXT:    s_add_u32 s14, s12, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s8, s8, s9
+; GCN-IR-NEXT:    s_cmp_lg_u32 s8, 0
 ; GCN-IR-NEXT:    s_addc_u32 s8, s13, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s12, 63, s12
@@ -181,6 +182,7 @@ define amdgpu_kernel void @s_test_udiv_i64(ptr addrspace(1) %out, i64 %x, i64 %y
 ; GCN-IR-NEXT:    s_add_u32 s10, s10, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[16:17], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s16, s16, s17
+; GCN-IR-NEXT:    s_cmp_lg_u32 s16, 0
 ; GCN-IR-NEXT:    s_addc_u32 s11, s11, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[16:17], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[2:3], s[4:5]
@@ -829,9 +831,10 @@ define amdgpu_kernel void @s_test_udiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    s_addc_u32 s10, 0, s11
 ; GCN-NEXT:    s_add_u32 s11, s4, s5
 ; GCN-NEXT:    v_mov_b32_e32 v0, s11
-; GCN-NEXT:    v_mul_hi_u32 v0, s6, v0
 ; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GCN-NEXT:    v_mul_hi_u32 v0, s6, v0
 ; GCN-NEXT:    s_or_b32 s4, s4, s5
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
 ; GCN-NEXT:    s_addc_u32 s9, s9, s10
 ; GCN-NEXT:    s_mul_i32 s4, s6, s9
 ; GCN-NEXT:    v_readfirstlane_b32 s5, v0
@@ -862,6 +865,7 @@ define amdgpu_kernel void @s_test_udiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    s_add_u32 s8, s11, s4
 ; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GCN-NEXT:    s_or_b32 s4, s4, s5
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
 ; GCN-NEXT:    s_addc_u32 s4, s9, s6
 ; GCN-NEXT:    v_mul_hi_u32 v1, s8, 24
 ; GCN-NEXT:    v_mul_hi_u32 v0, s4, 24
@@ -870,50 +874,52 @@ define amdgpu_kernel void @s_test_udiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    v_readfirstlane_b32 s8, v1
 ; GCN-NEXT:    v_readfirstlane_b32 s5, v0
 ; GCN-NEXT:    s_add_u32 s4, s8, s4
-; GCN-NEXT:    s_addc_u32 s10, 0, s5
-; GCN-NEXT:    v_mov_b32_e32 v0, s10
+; GCN-NEXT:    s_addc_u32 s8, 0, s5
+; GCN-NEXT:    v_mov_b32_e32 v0, s8
 ; GCN-NEXT:    v_mul_hi_u32 v0, s2, v0
 ; GCN-NEXT:    s_mov_b32 s4, s0
 ; GCN-NEXT:    s_mov_b32 s5, s1
-; GCN-NEXT:    s_mul_i32 s0, s3, s10
+; GCN-NEXT:    s_mul_i32 s0, s3, s8
 ; GCN-NEXT:    v_readfirstlane_b32 s1, v0
-; GCN-NEXT:    s_add_i32 s11, s1, s0
-; GCN-NEXT:    s_sub_i32 s8, 0, s11
-; GCN-NEXT:    s_mul_i32 s0, s2, s10
-; GCN-NEXT:    s_sub_u32 s12, 24, s0
+; GCN-NEXT:    s_add_i32 s9, s1, s0
+; GCN-NEXT:    s_sub_i32 s10, 0, s9
+; GCN-NEXT:    s_mul_i32 s0, s2, s8
+; GCN-NEXT:    s_sub_u32 s11, 24, s0
+; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GCN-NEXT:    s_or_b32 s12, s0, s1
+; GCN-NEXT:    s_cmp_lg_u32 s12, 0
+; GCN-NEXT:    s_subb_u32 s10, s10, s3
+; GCN-NEXT:    s_sub_u32 s13, s11, s2
 ; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
-; GCN-NEXT:    s_or_b32 s9, s0, s1
-; GCN-NEXT:    s_subb_u32 s13, s8, s3
-; GCN-NEXT:    s_sub_u32 s14, s12, s2
-; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
-; GCN-NEXT:    s_or_b32 s8, s8, s9
-; GCN-NEXT:    s_subb_u32 s8, s13, 0
-; GCN-NEXT:    s_cmp_ge_u32 s8, s3
-; GCN-NEXT:    s_cselect_b32 s9, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s14, s2
-; GCN-NEXT:    s_cselect_b32 s13, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s8, s3
-; GCN-NEXT:    s_cselect_b32 s8, s13, s9
-; GCN-NEXT:    s_add_u32 s9, s10, 1
-; GCN-NEXT:    s_addc_u32 s13, 0, 0
-; GCN-NEXT:    s_add_u32 s14, s10, 2
-; GCN-NEXT:    s_addc_u32 s15, 0, 0
-; GCN-NEXT:    s_cmp_lg_u32 s8, 0
-; GCN-NEXT:    s_cselect_b32 s8, s14, s9
-; GCN-NEXT:    s_cselect_b32 s9, s15, s13
 ; GCN-NEXT:    s_or_b32 s0, s0, s1
-; GCN-NEXT:    s_subb_u32 s0, 0, s11
+; GCN-NEXT:    s_cmp_lg_u32 s0, 0
+; GCN-NEXT:    s_subb_u32 s0, s10, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s0, s3
 ; GCN-NEXT:    s_cselect_b32 s1, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s12, s2
-; GCN-NEXT:    s_cselect_b32 s2, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s13, s2
+; GCN-NEXT:    s_cselect_b32 s10, -1, 0
 ; GCN-NEXT:    s_cmp_eq_u32 s0, s3
-; GCN-NEXT:    s_cselect_b32 s0, s2, s1
+; GCN-NEXT:    s_cselect_b32 s0, s10, s1
+; GCN-NEXT:    s_add_u32 s1, s8, 1
+; GCN-NEXT:    s_addc_u32 s10, 0, 0
+; GCN-NEXT:    s_add_u32 s13, s8, 2
+; GCN-NEXT:    s_addc_u32 s14, 0, 0
 ; GCN-NEXT:    s_cmp_lg_u32 s0, 0
-; GCN-NEXT:    s_cselect_b32 s0, s9, 0
-; GCN-NEXT:    s_cselect_b32 s1, s8, s10
-; GCN-NEXT:    v_mov_b32_e32 v0, s1
-; GCN-NEXT:    v_mov_b32_e32 v1, s0
+; GCN-NEXT:    s_cselect_b32 s0, s13, s1
+; GCN-NEXT:    s_cselect_b32 s1, s14, s10
+; GCN-NEXT:    s_cmp_lg_u32 s12, 0
+; GCN-NEXT:    s_subb_u32 s9, 0, s9
+; GCN-NEXT:    s_cmp_ge_u32 s9, s3
+; GCN-NEXT:    s_cselect_b32 s10, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s11, s2
+; GCN-NEXT:    s_cselect_b32 s2, -1, 0
+; GCN-NEXT:    s_cmp_eq_u32 s9, s3
+; GCN-NEXT:    s_cselect_b32 s2, s2, s10
+; GCN-NEXT:    s_cmp_lg_u32 s2, 0
+; GCN-NEXT:    s_cselect_b32 s1, s1, 0
+; GCN-NEXT:    s_cselect_b32 s0, s0, s8
+; GCN-NEXT:    v_mov_b32_e32 v0, s0
+; GCN-NEXT:    v_mov_b32_e32 v1, s1
 ; GCN-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; GCN-NEXT:    s_endpgm
 ;
@@ -939,6 +945,7 @@ define amdgpu_kernel void @s_test_udiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s10, s8, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s6, s6, s7
+; GCN-IR-NEXT:    s_cmp_lg_u32 s6, 0
 ; GCN-IR-NEXT:    s_addc_u32 s6, s9, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s8, 63, s8
@@ -971,6 +978,7 @@ define amdgpu_kernel void @s_test_udiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s14, s14, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[16:17], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s16, s16, s17
+; GCN-IR-NEXT:    s_cmp_lg_u32 s16, 0
 ; GCN-IR-NEXT:    s_addc_u32 s15, s15, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[16:17], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[8:9], s[4:5]
@@ -1309,6 +1317,7 @@ define amdgpu_kernel void @s_test_udiv_k_den_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s11, s8, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s6, s6, s7
+; GCN-IR-NEXT:    s_cmp_lg_u32 s6, 0
 ; GCN-IR-NEXT:    s_addc_u32 s6, s9, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s8, 63, s8
@@ -1338,6 +1347,7 @@ define amdgpu_kernel void @s_test_udiv_k_den_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s10, s10, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[12:13], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s12, s12, s13
+; GCN-IR-NEXT:    s_cmp_lg_u32 s12, 0
 ; GCN-IR-NEXT:    s_addc_u32 s11, s11, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[12:13], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[8:9], s[4:5]
diff --git a/llvm/test/CodeGen/AMDGPU/urem64.ll b/llvm/test/CodeGen/AMDGPU/urem64.ll
index 137dc1f..cdcc914 100644
--- a/llvm/test/CodeGen/AMDGPU/urem64.ll
+++ b/llvm/test/CodeGen/AMDGPU/urem64.ll
@@ -51,9 +51,10 @@ define amdgpu_kernel void @s_test_urem_i64(ptr addrspace(1) %out, i64 %x, i64 %y
 ; GCN-NEXT:    s_addc_u32 s13, 0, s14
 ; GCN-NEXT:    s_add_u32 s14, s0, s1
 ; GCN-NEXT:    v_mov_b32_e32 v0, s14
-; GCN-NEXT:    v_mul_hi_u32 v0, s10, v0
 ; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GCN-NEXT:    v_mul_hi_u32 v0, s10, v0
 ; GCN-NEXT:    s_or_b32 s0, s0, s1
+; GCN-NEXT:    s_cmp_lg_u32 s0, 0
 ; GCN-NEXT:    s_addc_u32 s12, s12, s13
 ; GCN-NEXT:    s_mul_i32 s0, s10, s12
 ; GCN-NEXT:    v_readfirstlane_b32 s1, v0
@@ -84,6 +85,7 @@ define amdgpu_kernel void @s_test_urem_i64(ptr addrspace(1) %out, i64 %x, i64 %y
 ; GCN-NEXT:    s_add_u32 s11, s14, s0
 ; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GCN-NEXT:    s_or_b32 s0, s0, s1
+; GCN-NEXT:    s_cmp_lg_u32 s0, 0
 ; GCN-NEXT:    s_addc_u32 s1, s12, s10
 ; GCN-NEXT:    v_mov_b32_e32 v0, s1
 ; GCN-NEXT:    v_mul_hi_u32 v1, s6, v0
@@ -113,43 +115,46 @@ define amdgpu_kernel void @s_test_urem_i64(ptr addrspace(1) %out, i64 %x, i64 %y
 ; GCN-NEXT:    v_readfirstlane_b32 s10, v0
 ; GCN-NEXT:    s_add_i32 s5, s10, s5
 ; GCN-NEXT:    s_mul_i32 s10, s9, s4
-; GCN-NEXT:    s_add_i32 s12, s5, s10
-; GCN-NEXT:    s_sub_i32 s10, s7, s12
+; GCN-NEXT:    s_add_i32 s10, s5, s10
+; GCN-NEXT:    s_sub_i32 s11, s7, s10
 ; GCN-NEXT:    s_mul_i32 s4, s8, s4
 ; GCN-NEXT:    s_sub_u32 s6, s6, s4
 ; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
-; GCN-NEXT:    s_or_b32 s11, s4, s5
-; GCN-NEXT:    s_subb_u32 s13, s10, s9
-; GCN-NEXT:    s_sub_u32 s14, s6, s8
-; GCN-NEXT:    s_cselect_b64 s[10:11], -1, 0
-; GCN-NEXT:    s_or_b32 s15, s10, s11
-; GCN-NEXT:    s_subb_u32 s15, s13, 0
-; GCN-NEXT:    s_cmp_ge_u32 s15, s9
-; GCN-NEXT:    s_cselect_b32 s16, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s14, s8
-; GCN-NEXT:    s_cselect_b32 s17, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s15, s9
-; GCN-NEXT:    s_cselect_b32 s16, s17, s16
-; GCN-NEXT:    s_or_b32 s10, s10, s11
-; GCN-NEXT:    s_subb_u32 s13, s13, s9
-; GCN-NEXT:    s_sub_u32 s17, s14, s8
-; GCN-NEXT:    s_cselect_b64 s[10:11], -1, 0
-; GCN-NEXT:    s_or_b32 s10, s10, s11
-; GCN-NEXT:    s_subb_u32 s10, s13, 0
-; GCN-NEXT:    s_cmp_lg_u32 s16, 0
-; GCN-NEXT:    s_cselect_b32 s11, s17, s14
-; GCN-NEXT:    s_cselect_b32 s10, s10, s15
+; GCN-NEXT:    s_or_b32 s12, s4, s5
+; GCN-NEXT:    s_cmp_lg_u32 s12, 0
+; GCN-NEXT:    s_subb_u32 s11, s11, s9
+; GCN-NEXT:    s_sub_u32 s13, s6, s8
+; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GCN-NEXT:    s_or_b32 s4, s4, s5
-; GCN-NEXT:    s_subb_u32 s4, s7, s12
-; GCN-NEXT:    s_cmp_ge_u32 s4, s9
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
+; GCN-NEXT:    s_subb_u32 s14, s11, 0
+; GCN-NEXT:    s_cmp_ge_u32 s14, s9
 ; GCN-NEXT:    s_cselect_b32 s5, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s13, s8
+; GCN-NEXT:    s_cselect_b32 s15, -1, 0
+; GCN-NEXT:    s_cmp_eq_u32 s14, s9
+; GCN-NEXT:    s_cselect_b32 s15, s15, s5
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
+; GCN-NEXT:    s_subb_u32 s11, s11, s9
+; GCN-NEXT:    s_sub_u32 s16, s13, s8
+; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GCN-NEXT:    s_or_b32 s4, s4, s5
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
+; GCN-NEXT:    s_subb_u32 s4, s11, 0
+; GCN-NEXT:    s_cmp_lg_u32 s15, 0
+; GCN-NEXT:    s_cselect_b32 s5, s16, s13
+; GCN-NEXT:    s_cselect_b32 s4, s4, s14
+; GCN-NEXT:    s_cmp_lg_u32 s12, 0
+; GCN-NEXT:    s_subb_u32 s7, s7, s10
+; GCN-NEXT:    s_cmp_ge_u32 s7, s9
+; GCN-NEXT:    s_cselect_b32 s10, -1, 0
 ; GCN-NEXT:    s_cmp_ge_u32 s6, s8
-; GCN-NEXT:    s_cselect_b32 s7, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s4, s9
-; GCN-NEXT:    s_cselect_b32 s5, s7, s5
-; GCN-NEXT:    s_cmp_lg_u32 s5, 0
-; GCN-NEXT:    s_cselect_b32 s4, s10, s4
-; GCN-NEXT:    s_cselect_b32 s5, s11, s6
+; GCN-NEXT:    s_cselect_b32 s8, -1, 0
+; GCN-NEXT:    s_cmp_eq_u32 s7, s9
+; GCN-NEXT:    s_cselect_b32 s8, s8, s10
+; GCN-NEXT:    s_cmp_lg_u32 s8, 0
+; GCN-NEXT:    s_cselect_b32 s4, s4, s7
+; GCN-NEXT:    s_cselect_b32 s5, s5, s6
 ; GCN-NEXT:    v_mov_b32_e32 v0, s5
 ; GCN-NEXT:    v_mov_b32_e32 v1, s4
 ; GCN-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
@@ -182,6 +187,7 @@ define amdgpu_kernel void @s_test_urem_i64(ptr addrspace(1) %out, i64 %x, i64 %y
 ; GCN-IR-NEXT:    s_add_u32 s14, s12, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s8, s8, s9
+; GCN-IR-NEXT:    s_cmp_lg_u32 s8, 0
 ; GCN-IR-NEXT:    s_addc_u32 s8, s13, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[8:9], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s12, 63, s12
@@ -215,6 +221,7 @@ define amdgpu_kernel void @s_test_urem_i64(ptr addrspace(1) %out, i64 %x, i64 %y
 ; GCN-IR-NEXT:    s_add_u32 s16, s16, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[18:19], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s18, s18, s19
+; GCN-IR-NEXT:    s_cmp_lg_u32 s18, 0
 ; GCN-IR-NEXT:    s_addc_u32 s17, s17, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[18:19], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[10:11], s[4:5]
@@ -846,9 +853,10 @@ define amdgpu_kernel void @s_test_urem_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    s_addc_u32 s10, 0, s11
 ; GCN-NEXT:    s_add_u32 s11, s4, s5
 ; GCN-NEXT:    v_mov_b32_e32 v0, s11
-; GCN-NEXT:    v_mul_hi_u32 v0, s6, v0
 ; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GCN-NEXT:    v_mul_hi_u32 v0, s6, v0
 ; GCN-NEXT:    s_or_b32 s4, s4, s5
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
 ; GCN-NEXT:    s_addc_u32 s9, s9, s10
 ; GCN-NEXT:    s_mul_i32 s4, s6, s9
 ; GCN-NEXT:    v_readfirstlane_b32 s5, v0
@@ -879,6 +887,7 @@ define amdgpu_kernel void @s_test_urem_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    s_add_u32 s8, s11, s4
 ; GCN-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; GCN-NEXT:    s_or_b32 s4, s4, s5
+; GCN-NEXT:    s_cmp_lg_u32 s4, 0
 ; GCN-NEXT:    s_addc_u32 s4, s9, s6
 ; GCN-NEXT:    v_mul_hi_u32 v1, s8, 24
 ; GCN-NEXT:    v_mul_hi_u32 v0, s4, 24
@@ -894,43 +903,46 @@ define amdgpu_kernel void @s_test_urem_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-NEXT:    s_mov_b32 s5, s1
 ; GCN-NEXT:    s_mul_i32 s0, s3, s8
 ; GCN-NEXT:    v_readfirstlane_b32 s1, v0
-; GCN-NEXT:    s_add_i32 s10, s1, s0
-; GCN-NEXT:    s_sub_i32 s9, 0, s10
+; GCN-NEXT:    s_add_i32 s9, s1, s0
+; GCN-NEXT:    s_sub_i32 s10, 0, s9
 ; GCN-NEXT:    s_mul_i32 s0, s2, s8
-; GCN-NEXT:    s_sub_u32 s11, 24, s0
+; GCN-NEXT:    s_sub_u32 s8, 24, s0
+; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GCN-NEXT:    s_or_b32 s11, s0, s1
+; GCN-NEXT:    s_cmp_lg_u32 s11, 0
+; GCN-NEXT:    s_subb_u32 s10, s10, s3
+; GCN-NEXT:    s_sub_u32 s12, s8, s2
 ; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
-; GCN-NEXT:    s_or_b32 s8, s0, s1
-; GCN-NEXT:    s_subb_u32 s12, s9, s3
-; GCN-NEXT:    s_sub_u32 s13, s11, s2
-; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
-; GCN-NEXT:    s_or_b32 s14, s8, s9
-; GCN-NEXT:    s_subb_u32 s14, s12, 0
-; GCN-NEXT:    s_cmp_ge_u32 s14, s3
-; GCN-NEXT:    s_cselect_b32 s15, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s13, s2
-; GCN-NEXT:    s_cselect_b32 s16, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s14, s3
-; GCN-NEXT:    s_cselect_b32 s15, s16, s15
-; GCN-NEXT:    s_or_b32 s8, s8, s9
-; GCN-NEXT:    s_subb_u32 s12, s12, s3
-; GCN-NEXT:    s_sub_u32 s16, s13, s2
-; GCN-NEXT:    s_cselect_b64 s[8:9], -1, 0
-; GCN-NEXT:    s_or_b32 s8, s8, s9
-; GCN-NEXT:    s_subb_u32 s8, s12, 0
-; GCN-NEXT:    s_cmp_lg_u32 s15, 0
-; GCN-NEXT:    s_cselect_b32 s9, s16, s13
-; GCN-NEXT:    s_cselect_b32 s8, s8, s14
 ; GCN-NEXT:    s_or_b32 s0, s0, s1
-; GCN-NEXT:    s_subb_u32 s0, 0, s10
-; GCN-NEXT:    s_cmp_ge_u32 s0, s3
+; GCN-NEXT:    s_cmp_lg_u32 s0, 0
+; GCN-NEXT:    s_subb_u32 s13, s10, 0
+; GCN-NEXT:    s_cmp_ge_u32 s13, s3
 ; GCN-NEXT:    s_cselect_b32 s1, -1, 0
-; GCN-NEXT:    s_cmp_ge_u32 s11, s2
+; GCN-NEXT:    s_cmp_ge_u32 s12, s2
+; GCN-NEXT:    s_cselect_b32 s14, -1, 0
+; GCN-NEXT:    s_cmp_eq_u32 s13, s3
+; GCN-NEXT:    s_cselect_b32 s14, s14, s1
+; GCN-NEXT:    s_cmp_lg_u32 s0, 0
+; GCN-NEXT:    s_subb_u32 s10, s10, s3
+; GCN-NEXT:    s_sub_u32 s15, s12, s2
+; GCN-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GCN-NEXT:    s_or_b32 s0, s0, s1
+; GCN-NEXT:    s_cmp_lg_u32 s0, 0
+; GCN-NEXT:    s_subb_u32 s0, s10, 0
+; GCN-NEXT:    s_cmp_lg_u32 s14, 0
+; GCN-NEXT:    s_cselect_b32 s1, s15, s12
+; GCN-NEXT:    s_cselect_b32 s0, s0, s13
+; GCN-NEXT:    s_cmp_lg_u32 s11, 0
+; GCN-NEXT:    s_subb_u32 s9, 0, s9
+; GCN-NEXT:    s_cmp_ge_u32 s9, s3
+; GCN-NEXT:    s_cselect_b32 s10, -1, 0
+; GCN-NEXT:    s_cmp_ge_u32 s8, s2
 ; GCN-NEXT:    s_cselect_b32 s2, -1, 0
-; GCN-NEXT:    s_cmp_eq_u32 s0, s3
-; GCN-NEXT:    s_cselect_b32 s1, s2, s1
-; GCN-NEXT:    s_cmp_lg_u32 s1, 0
-; GCN-NEXT:    s_cselect_b32 s0, s8, s0
-; GCN-NEXT:    s_cselect_b32 s1, s9, s11
+; GCN-NEXT:    s_cmp_eq_u32 s9, s3
+; GCN-NEXT:    s_cselect_b32 s2, s2, s10
+; GCN-NEXT:    s_cmp_lg_u32 s2, 0
+; GCN-NEXT:    s_cselect_b32 s0, s0, s9
+; GCN-NEXT:    s_cselect_b32 s1, s1, s8
 ; GCN-NEXT:    v_mov_b32_e32 v0, s1
 ; GCN-NEXT:    v_mov_b32_e32 v1, s0
 ; GCN-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
@@ -958,6 +970,7 @@ define amdgpu_kernel void @s_test_urem_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s10, s8, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s6, s6, s7
+; GCN-IR-NEXT:    s_cmp_lg_u32 s6, 0
 ; GCN-IR-NEXT:    s_addc_u32 s6, s9, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s8, 63, s8
@@ -990,6 +1003,7 @@ define amdgpu_kernel void @s_test_urem_k_num_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s14, s14, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[16:17], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s16, s16, s17
+; GCN-IR-NEXT:    s_cmp_lg_u32 s16, 0
 ; GCN-IR-NEXT:    s_addc_u32 s15, s15, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[16:17], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[8:9], s[4:5]
@@ -1079,6 +1093,7 @@ define amdgpu_kernel void @s_test_urem_k_den_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s11, s8, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s6, s6, s7
+; GCN-IR-NEXT:    s_cmp_lg_u32 s6, 0
 ; GCN-IR-NEXT:    s_addc_u32 s6, s9, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[6:7], -1, 0
 ; GCN-IR-NEXT:    s_sub_i32 s8, 63, s8
@@ -1108,6 +1123,7 @@ define amdgpu_kernel void @s_test_urem_k_den_i64(ptr addrspace(1) %out, i64 %x)
 ; GCN-IR-NEXT:    s_add_u32 s12, s12, 1
 ; GCN-IR-NEXT:    s_cselect_b64 s[14:15], -1, 0
 ; GCN-IR-NEXT:    s_or_b32 s14, s14, s15
+; GCN-IR-NEXT:    s_cmp_lg_u32 s14, 0
 ; GCN-IR-NEXT:    s_addc_u32 s13, s13, 0
 ; GCN-IR-NEXT:    s_cselect_b64 s[14:15], -1, 0
 ; GCN-IR-NEXT:    s_mov_b64 s[10:11], s[4:5]
diff --git a/llvm/test/CodeGen/AMDGPU/usubo.ll b/llvm/test/CodeGen/AMDGPU/usubo.ll
index e8db647..d67a7b1 100644
--- a/llvm/test/CodeGen/AMDGPU/usubo.ll
+++ b/llvm/test/CodeGen/AMDGPU/usubo.ll
@@ -18,6 +18,7 @@ define amdgpu_kernel void @s_usubo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
 ; SI-NEXT:    s_mov_b32 s5, s1
 ; SI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; SI-NEXT:    s_or_b32 s0, s0, s1
+; SI-NEXT:    s_cmp_lg_u32 s0, 0
 ; SI-NEXT:    s_subb_u32 s3, s3, s9
 ; SI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; SI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -34,9 +35,11 @@ define amdgpu_kernel void @s_usubo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
 ; VI-NEXT:    s_sub_u32 s2, s2, s4
-; VI-NEXT:    s_subb_u32 s3, s3, s5
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
 ; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; VI-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; VI-NEXT:    s_subb_u32 s3, s3, s5
+; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; VI-NEXT:    v_cndmask_b32_e64 v2, 0, 1, s[0:1]
 ; VI-NEXT:    v_mov_b32_e32 v3, s3
 ; VI-NEXT:    v_add_u32_e32 v2, vcc, s2, v2
@@ -50,12 +53,14 @@ define amdgpu_kernel void @s_usubo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
 ; GFX9-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_sub_u32 s4, s2, s6
-; GFX9-NEXT:    s_subb_u32 s5, s3, s7
+; GFX9-NEXT:    s_sub_u32 s6, s2, s6
+; GFX9-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX9-NEXT:    s_subb_u32 s4, s3, s7
 ; GFX9-NEXT:    s_cselect_b64 s[2:3], -1, 0
 ; GFX9-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[2:3]
-; GFX9-NEXT:    v_mov_b32_e32 v1, s5
-; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s4, v0
+; GFX9-NEXT:    v_mov_b32_e32 v1, s4
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s6, v0
 ; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
 ; GFX9-NEXT:    global_store_dwordx2 v2, v[0:1], s[0:1]
 ; GFX9-NEXT:    s_endpgm
@@ -68,6 +73,8 @@ define amdgpu_kernel void @s_usubo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
 ; GFX10-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    s_sub_u32 s2, s2, s6
+; GFX10-NEXT:    s_cselect_b32 s4, -1, 0
+; GFX10-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX10-NEXT:    s_subb_u32 s3, s3, s7
 ; GFX10-NEXT:    s_cselect_b32 s4, -1, 0
 ; GFX10-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s4
@@ -84,12 +91,14 @@ define amdgpu_kernel void @s_usubo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
 ; GFX11-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-NEXT:    s_sub_u32 s2, s2, s4
+; GFX11-NEXT:    s_cselect_b32 s4, -1, 0
+; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
+; GFX11-NEXT:    s_cmp_lg_u32 s4, 0
 ; GFX11-NEXT:    s_subb_u32 s3, s3, s5
 ; GFX11-NEXT:    s_cselect_b32 s4, -1, 0
-; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
 ; GFX11-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s4
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1)
 ; GFX11-NEXT:    v_add_co_u32 v0, s2, s2, v0
-; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX11-NEXT:    v_add_co_ci_u32_e64 v1, null, s3, 0, s2
 ; GFX11-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
 ; GFX11-NEXT:    s_endpgm
@@ -434,6 +443,7 @@ define amdgpu_kernel void @s_usubo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
 ; SI-NEXT:    s_sub_u32 s4, s4, s6
 ; SI-NEXT:    s_cselect_b64 s[12:13], -1, 0
 ; SI-NEXT:    s_or_b32 s6, s12, s13
+; SI-NEXT:    s_cmp_lg_u32 s6, 0
 ; SI-NEXT:    s_subb_u32 s5, s5, s7
 ; SI-NEXT:    s_mov_b32 s8, s0
 ; SI-NEXT:    s_mov_b32 s9, s1
@@ -454,14 +464,16 @@ define amdgpu_kernel void @s_usubo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx8 s[0:7], s[4:5], 0x24
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v2, s2
+; VI-NEXT:    s_sub_u32 s2, s4, s6
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
-; VI-NEXT:    s_sub_u32 s0, s4, s6
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_subb_u32 s1, s5, s7
-; VI-NEXT:    v_mov_b32_e32 v4, s0
-; VI-NEXT:    v_mov_b32_e32 v5, s1
 ; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
-; VI-NEXT:    v_mov_b32_e32 v2, s2
+; VI-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; VI-NEXT:    s_subb_u32 s0, s5, s7
+; VI-NEXT:    v_mov_b32_e32 v4, s2
+; VI-NEXT:    v_mov_b32_e32 v5, s0
+; VI-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; VI-NEXT:    v_mov_b32_e32 v3, s3
 ; VI-NEXT:    flat_store_dwordx2 v[0:1], v[4:5]
 ; VI-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -473,10 +485,12 @@ define amdgpu_kernel void @s_usubo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
 ; GFX9-NEXT:    s_load_dwordx8 s[8:15], s[4:5], 0x24
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_sub_u32 s0, s12, s14
-; GFX9-NEXT:    s_subb_u32 s1, s13, s15
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    s_sub_u32 s2, s12, s14
+; GFX9-NEXT:    s_cselect_b64 s[0:1], -1, 0
+; GFX9-NEXT:    s_cmp_lg_u64 s[0:1], 0
+; GFX9-NEXT:    s_subb_u32 s0, s13, s15
+; GFX9-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s0
 ; GFX9-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; GFX9-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s[0:1]
 ; GFX9-NEXT:    global_store_dwordx2 v2, v[0:1], s[8:9]
@@ -489,8 +503,10 @@ define amdgpu_kernel void @s_usubo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
 ; GFX10-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    s_sub_u32 s0, s12, s14
-; GFX10-NEXT:    s_subb_u32 s1, s13, s15
+; GFX10-NEXT:    s_cselect_b32 s1, -1, 0
 ; GFX10-NEXT:    v_mov_b32_e32 v0, s0
+; GFX10-NEXT:    s_cmp_lg_u32 s1, 0
+; GFX10-NEXT:    s_subb_u32 s1, s13, s15
 ; GFX10-NEXT:    s_cselect_b32 s0, -1, 0
 ; GFX10-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX10-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s0
@@ -503,8 +519,10 @@ define amdgpu_kernel void @s_usubo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
 ; GFX11-NEXT:    s_load_b256 s[0:7], s[4:5], 0x24
 ; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-NEXT:    s_sub_u32 s4, s4, s6
-; GFX11-NEXT:    s_subb_u32 s5, s5, s7
+; GFX11-NEXT:    s_cselect_b32 s6, -1, 0
 ; GFX11-NEXT:    v_mov_b32_e32 v0, s4
+; GFX11-NEXT:    s_cmp_lg_u32 s6, 0
+; GFX11-NEXT:    s_subb_u32 s5, s5, s7
 ; GFX11-NEXT:    s_cselect_b32 s4, -1, 0
 ; GFX11-NEXT:    v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s5
 ; GFX11-NEXT:    v_cndmask_b32_e64 v3, 0, 1, s4
diff --git a/llvm/test/CodeGen/AMDGPU/wave32.ll b/llvm/test/CodeGen/AMDGPU/wave32.ll
index 28c6b40..75db387 100644
--- a/llvm/test/CodeGen/AMDGPU/wave32.ll
+++ b/llvm/test/CodeGen/AMDGPU/wave32.ll
@@ -774,40 +774,44 @@ define amdgpu_kernel void @test_udiv64(ptr addrspace(1) %arg) #0 {
 ; GFX1032-NEXT:    s_add_u32 s11, s12, s11
 ; GFX1032-NEXT:    s_addc_u32 s12, 0, s13
 ; GFX1032-NEXT:    s_add_u32 s8, s8, s11
+; GFX1032-NEXT:    s_cselect_b32 s11, -1, 0
+; GFX1032-NEXT:    s_mul_hi_u32 s13, s9, s8
+; GFX1032-NEXT:    s_cmp_lg_u32 s11, 0
+; GFX1032-NEXT:    s_mul_i32 s11, s9, s8
 ; GFX1032-NEXT:    s_addc_u32 s5, s5, s12
-; GFX1032-NEXT:    s_mul_hi_u32 s11, s9, s8
-; GFX1032-NEXT:    s_mul_i32 s12, s9, s8
-; GFX1032-NEXT:    s_mul_i32 s9, s9, s5
 ; GFX1032-NEXT:    s_mul_i32 s10, s10, s8
-; GFX1032-NEXT:    s_add_i32 s9, s11, s9
-; GFX1032-NEXT:    s_mul_i32 s11, s5, s12
+; GFX1032-NEXT:    s_mul_i32 s9, s9, s5
+; GFX1032-NEXT:    s_mul_hi_u32 s12, s8, s11
+; GFX1032-NEXT:    s_add_i32 s9, s13, s9
+; GFX1032-NEXT:    s_mul_hi_u32 s13, s5, s11
 ; GFX1032-NEXT:    s_add_i32 s9, s9, s10
-; GFX1032-NEXT:    s_mul_hi_u32 s10, s8, s12
+; GFX1032-NEXT:    s_mul_i32 s10, s5, s11
 ; GFX1032-NEXT:    s_mul_i32 s15, s8, s9
 ; GFX1032-NEXT:    s_mul_hi_u32 s14, s8, s9
-; GFX1032-NEXT:    s_add_u32 s10, s10, s15
-; GFX1032-NEXT:    s_mul_hi_u32 s13, s5, s12
+; GFX1032-NEXT:    s_add_u32 s12, s12, s15
 ; GFX1032-NEXT:    s_addc_u32 s14, 0, s14
-; GFX1032-NEXT:    s_mul_hi_u32 s12, s5, s9
-; GFX1032-NEXT:    s_add_u32 s10, s10, s11
+; GFX1032-NEXT:    s_mul_hi_u32 s11, s5, s9
+; GFX1032-NEXT:    s_add_u32 s10, s12, s10
 ; GFX1032-NEXT:    s_mul_i32 s9, s5, s9
 ; GFX1032-NEXT:    s_addc_u32 s10, s14, s13
-; GFX1032-NEXT:    s_addc_u32 s11, s12, 0
+; GFX1032-NEXT:    s_addc_u32 s11, s11, 0
 ; GFX1032-NEXT:    s_add_u32 s9, s10, s9
 ; GFX1032-NEXT:    s_addc_u32 s10, 0, s11
 ; GFX1032-NEXT:    s_add_u32 s8, s8, s9
+; GFX1032-NEXT:    s_cselect_b32 s9, -1, 0
+; GFX1032-NEXT:    s_mul_hi_u32 s11, s2, s8
+; GFX1032-NEXT:    s_cmp_lg_u32 s9, 0
+; GFX1032-NEXT:    s_mul_hi_u32 s9, s3, s8
 ; GFX1032-NEXT:    s_addc_u32 s5, s5, s10
-; GFX1032-NEXT:    s_mul_hi_u32 s9, s2, s8
-; GFX1032-NEXT:    s_mul_i32 s12, s2, s5
-; GFX1032-NEXT:    s_mul_hi_u32 s11, s2, s5
-; GFX1032-NEXT:    s_mul_hi_u32 s10, s3, s8
 ; GFX1032-NEXT:    s_mul_i32 s8, s3, s8
-; GFX1032-NEXT:    s_add_u32 s9, s9, s12
-; GFX1032-NEXT:    s_addc_u32 s11, 0, s11
+; GFX1032-NEXT:    s_mul_i32 s12, s2, s5
+; GFX1032-NEXT:    s_mul_hi_u32 s10, s2, s5
+; GFX1032-NEXT:    s_add_u32 s11, s11, s12
+; GFX1032-NEXT:    s_addc_u32 s10, 0, s10
 ; GFX1032-NEXT:    s_mul_hi_u32 s13, s3, s5
-; GFX1032-NEXT:    s_add_u32 s8, s9, s8
+; GFX1032-NEXT:    s_add_u32 s8, s11, s8
 ; GFX1032-NEXT:    s_mul_i32 s5, s3, s5
-; GFX1032-NEXT:    s_addc_u32 s8, s11, s10
+; GFX1032-NEXT:    s_addc_u32 s8, s10, s9
 ; GFX1032-NEXT:    s_addc_u32 s9, s13, 0
 ; GFX1032-NEXT:    s_add_u32 s5, s8, s5
 ; GFX1032-NEXT:    s_addc_u32 s8, 0, s9
@@ -820,8 +824,11 @@ define amdgpu_kernel void @test_udiv64(ptr addrspace(1) %arg) #0 {
 ; GFX1032-NEXT:    s_sub_i32 s11, s3, s9
 ; GFX1032-NEXT:    s_sub_u32 s10, s2, s10
 ; GFX1032-NEXT:    s_cselect_b32 s12, -1, 0
+; GFX1032-NEXT:    s_cmp_lg_u32 s12, 0
 ; GFX1032-NEXT:    s_subb_u32 s11, s11, s1
 ; GFX1032-NEXT:    s_sub_u32 s13, s10, s0
+; GFX1032-NEXT:    s_cselect_b32 s14, -1, 0
+; GFX1032-NEXT:    s_cmp_lg_u32 s14, 0
 ; GFX1032-NEXT:    s_subb_u32 s11, s11, 0
 ; GFX1032-NEXT:    s_cmp_ge_u32 s11, s1
 ; GFX1032-NEXT:    s_cselect_b32 s14, -1, 0
@@ -894,8 +901,8 @@ define amdgpu_kernel void @test_udiv64(ptr addrspace(1) %arg) #0 {
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    v_cvt_f32_u32_e32 v0, s0
 ; GFX1064-NEXT:    v_cvt_f32_u32_e32 v1, s1
-; GFX1064-NEXT:    s_sub_u32 s8, 0, s0
-; GFX1064-NEXT:    s_subb_u32 s9, 0, s1
+; GFX1064-NEXT:    s_sub_u32 s9, 0, s0
+; GFX1064-NEXT:    s_subb_u32 s10, 0, s1
 ; GFX1064-NEXT:    v_madmk_f32 v0, v1, 0x4f800000, v0
 ; GFX1064-NEXT:    v_rcp_f32_e32 v0, v0
 ; GFX1064-NEXT:    v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -904,102 +911,109 @@ define amdgpu_kernel void @test_udiv64(ptr addrspace(1) %arg) #0 {
 ; GFX1064-NEXT:    v_madmk_f32 v0, v1, 0xcf800000, v0
 ; GFX1064-NEXT:    v_cvt_u32_f32_e32 v1, v1
 ; GFX1064-NEXT:    v_cvt_u32_f32_e32 v0, v0
-; GFX1064-NEXT:    v_readfirstlane_b32 s4, v1
-; GFX1064-NEXT:    v_readfirstlane_b32 s5, v0
-; GFX1064-NEXT:    s_mul_i32 s10, s8, s4
-; GFX1064-NEXT:    s_mul_hi_u32 s12, s8, s5
-; GFX1064-NEXT:    s_mul_i32 s11, s9, s5
-; GFX1064-NEXT:    s_add_i32 s10, s12, s10
-; GFX1064-NEXT:    s_mul_i32 s13, s8, s5
-; GFX1064-NEXT:    s_add_i32 s10, s10, s11
-; GFX1064-NEXT:    s_mul_hi_u32 s12, s5, s13
-; GFX1064-NEXT:    s_mul_i32 s15, s5, s10
-; GFX1064-NEXT:    s_mul_hi_u32 s14, s4, s13
-; GFX1064-NEXT:    s_mul_i32 s11, s4, s13
-; GFX1064-NEXT:    s_mul_hi_u32 s13, s5, s10
+; GFX1064-NEXT:    v_readfirstlane_b32 s8, v1
+; GFX1064-NEXT:    v_readfirstlane_b32 s4, v0
+; GFX1064-NEXT:    s_mul_i32 s5, s9, s8
+; GFX1064-NEXT:    s_mul_hi_u32 s12, s9, s4
+; GFX1064-NEXT:    s_mul_i32 s11, s10, s4
+; GFX1064-NEXT:    s_add_i32 s5, s12, s5
+; GFX1064-NEXT:    s_mul_i32 s13, s9, s4
+; GFX1064-NEXT:    s_add_i32 s5, s5, s11
+; GFX1064-NEXT:    s_mul_hi_u32 s12, s4, s13
+; GFX1064-NEXT:    s_mul_i32 s15, s4, s5
+; GFX1064-NEXT:    s_mul_hi_u32 s14, s8, s13
+; GFX1064-NEXT:    s_mul_i32 s11, s8, s13
+; GFX1064-NEXT:    s_mul_hi_u32 s13, s4, s5
 ; GFX1064-NEXT:    s_add_u32 s12, s12, s15
 ; GFX1064-NEXT:    s_addc_u32 s13, 0, s13
-; GFX1064-NEXT:    s_mul_hi_u32 s16, s4, s10
+; GFX1064-NEXT:    s_mul_hi_u32 s16, s8, s5
 ; GFX1064-NEXT:    s_add_u32 s11, s12, s11
-; GFX1064-NEXT:    s_mul_i32 s10, s4, s10
+; GFX1064-NEXT:    s_mul_i32 s5, s8, s5
 ; GFX1064-NEXT:    s_addc_u32 s11, s13, s14
 ; GFX1064-NEXT:    s_addc_u32 s12, s16, 0
-; GFX1064-NEXT:    s_add_u32 s10, s11, s10
+; GFX1064-NEXT:    s_add_u32 s5, s11, s5
 ; GFX1064-NEXT:    s_addc_u32 s11, 0, s12
-; GFX1064-NEXT:    s_add_u32 s5, s5, s10
-; GFX1064-NEXT:    s_addc_u32 s4, s4, s11
-; GFX1064-NEXT:    s_mul_hi_u32 s10, s8, s5
-; GFX1064-NEXT:    s_mul_i32 s11, s8, s5
-; GFX1064-NEXT:    s_mul_i32 s8, s8, s4
-; GFX1064-NEXT:    s_mul_i32 s9, s9, s5
-; GFX1064-NEXT:    s_add_i32 s8, s10, s8
-; GFX1064-NEXT:    s_mul_i32 s10, s4, s11
-; GFX1064-NEXT:    s_add_i32 s8, s8, s9
-; GFX1064-NEXT:    s_mul_hi_u32 s9, s5, s11
-; GFX1064-NEXT:    s_mul_i32 s14, s5, s8
-; GFX1064-NEXT:    s_mul_hi_u32 s13, s5, s8
-; GFX1064-NEXT:    s_add_u32 s9, s9, s14
-; GFX1064-NEXT:    s_mul_hi_u32 s12, s4, s11
+; GFX1064-NEXT:    s_add_u32 s12, s4, s5
+; GFX1064-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX1064-NEXT:    s_mul_hi_u32 s13, s9, s12
+; GFX1064-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX1064-NEXT:    s_mul_i32 s4, s9, s12
+; GFX1064-NEXT:    s_addc_u32 s8, s8, s11
+; GFX1064-NEXT:    s_mul_i32 s10, s10, s12
+; GFX1064-NEXT:    s_mul_i32 s9, s9, s8
+; GFX1064-NEXT:    s_mul_hi_u32 s5, s12, s4
+; GFX1064-NEXT:    s_add_i32 s9, s13, s9
+; GFX1064-NEXT:    s_mul_hi_u32 s11, s8, s4
+; GFX1064-NEXT:    s_add_i32 s9, s9, s10
+; GFX1064-NEXT:    s_mul_i32 s4, s8, s4
+; GFX1064-NEXT:    s_mul_i32 s14, s12, s9
+; GFX1064-NEXT:    s_mul_hi_u32 s13, s12, s9
+; GFX1064-NEXT:    s_add_u32 s5, s5, s14
 ; GFX1064-NEXT:    s_addc_u32 s13, 0, s13
-; GFX1064-NEXT:    s_mul_hi_u32 s11, s4, s8
-; GFX1064-NEXT:    s_add_u32 s9, s9, s10
-; GFX1064-NEXT:    s_mul_i32 s8, s4, s8
-; GFX1064-NEXT:    s_addc_u32 s9, s13, s12
-; GFX1064-NEXT:    s_addc_u32 s10, s11, 0
-; GFX1064-NEXT:    s_add_u32 s8, s9, s8
-; GFX1064-NEXT:    s_addc_u32 s9, 0, s10
-; GFX1064-NEXT:    s_add_u32 s5, s5, s8
-; GFX1064-NEXT:    s_addc_u32 s4, s4, s9
-; GFX1064-NEXT:    s_mul_hi_u32 s8, s2, s5
-; GFX1064-NEXT:    s_mul_i32 s11, s2, s4
-; GFX1064-NEXT:    s_mul_hi_u32 s10, s2, s4
-; GFX1064-NEXT:    s_mul_hi_u32 s9, s3, s5
+; GFX1064-NEXT:    s_mul_hi_u32 s10, s8, s9
+; GFX1064-NEXT:    s_add_u32 s4, s5, s4
+; GFX1064-NEXT:    s_mul_i32 s9, s8, s9
+; GFX1064-NEXT:    s_addc_u32 s4, s13, s11
+; GFX1064-NEXT:    s_addc_u32 s5, s10, 0
+; GFX1064-NEXT:    s_add_u32 s4, s4, s9
+; GFX1064-NEXT:    s_addc_u32 s9, 0, s5
+; GFX1064-NEXT:    s_add_u32 s10, s12, s4
+; GFX1064-NEXT:    s_cselect_b64 s[4:5], -1, 0
+; GFX1064-NEXT:    s_mul_hi_u32 s11, s2, s10
+; GFX1064-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX1064-NEXT:    s_mul_hi_u32 s4, s3, s10
+; GFX1064-NEXT:    s_addc_u32 s5, s8, s9
+; GFX1064-NEXT:    s_mul_i32 s8, s3, s10
+; GFX1064-NEXT:    s_mul_i32 s10, s2, s5
+; GFX1064-NEXT:    s_mul_hi_u32 s9, s2, s5
+; GFX1064-NEXT:    s_add_u32 s10, s11, s10
+; GFX1064-NEXT:    s_addc_u32 s9, 0, s9
+; GFX1064-NEXT:    s_mul_hi_u32 s12, s3, s5
+; GFX1064-NEXT:    s_add_u32 s8, s10, s8
 ; GFX1064-NEXT:    s_mul_i32 s5, s3, s5
-; GFX1064-NEXT:    s_add_u32 s8, s8, s11
-; GFX1064-NEXT:    s_addc_u32 s10, 0, s10
-; GFX1064-NEXT:    s_mul_hi_u32 s12, s3, s4
-; GFX1064-NEXT:    s_add_u32 s5, s8, s5
-; GFX1064-NEXT:    s_mul_i32 s4, s3, s4
-; GFX1064-NEXT:    s_addc_u32 s5, s10, s9
+; GFX1064-NEXT:    s_addc_u32 s4, s9, s4
 ; GFX1064-NEXT:    s_addc_u32 s8, s12, 0
-; GFX1064-NEXT:    s_add_u32 s10, s5, s4
+; GFX1064-NEXT:    s_add_u32 s10, s4, s5
 ; GFX1064-NEXT:    s_addc_u32 s11, 0, s8
 ; GFX1064-NEXT:    s_mul_hi_u32 s4, s0, s10
 ; GFX1064-NEXT:    s_mul_i32 s5, s0, s11
 ; GFX1064-NEXT:    s_mul_i32 s8, s1, s10
 ; GFX1064-NEXT:    s_add_i32 s4, s4, s5
-; GFX1064-NEXT:    s_add_i32 s8, s4, s8
+; GFX1064-NEXT:    s_add_i32 s12, s4, s8
 ; GFX1064-NEXT:    s_mul_i32 s4, s0, s10
-; GFX1064-NEXT:    s_sub_i32 s9, s3, s8
-; GFX1064-NEXT:    s_sub_u32 s12, s2, s4
+; GFX1064-NEXT:    s_sub_i32 s8, s3, s12
+; GFX1064-NEXT:    s_sub_u32 s13, s2, s4
 ; GFX1064-NEXT:    s_cselect_b64 s[4:5], -1, 0
-; GFX1064-NEXT:    s_subb_u32 s9, s9, s1
-; GFX1064-NEXT:    s_sub_u32 s13, s12, s0
-; GFX1064-NEXT:    s_subb_u32 s9, s9, 0
-; GFX1064-NEXT:    s_cmp_ge_u32 s9, s1
+; GFX1064-NEXT:    s_cmp_lg_u64 s[4:5], 0
+; GFX1064-NEXT:    s_subb_u32 s14, s8, s1
+; GFX1064-NEXT:    s_sub_u32 s15, s13, s0
+; GFX1064-NEXT:    s_cselect_b64 s[8:9], -1, 0
+; GFX1064-NEXT:    s_cmp_lg_u64 s[8:9], 0
+; GFX1064-NEXT:    s_subb_u32 s8, s14, 0
+; GFX1064-NEXT:    s_cmp_ge_u32 s8, s1
+; GFX1064-NEXT:    s_cselect_b32 s9, -1, 0
+; GFX1064-NEXT:    s_cmp_ge_u32 s15, s0
 ; GFX1064-NEXT:    s_cselect_b32 s14, -1, 0
-; GFX1064-NEXT:    s_cmp_ge_u32 s13, s0
-; GFX1064-NEXT:    s_cselect_b32 s13, -1, 0
-; GFX1064-NEXT:    s_cmp_eq_u32 s9, s1
-; GFX1064-NEXT:    s_cselect_b32 s9, s13, s14
-; GFX1064-NEXT:    s_add_u32 s13, s10, 1
+; GFX1064-NEXT:    s_cmp_eq_u32 s8, s1
+; GFX1064-NEXT:    s_cselect_b32 s8, s14, s9
+; GFX1064-NEXT:    s_add_u32 s9, s10, 1
 ; GFX1064-NEXT:    s_addc_u32 s14, s11, 0
 ; GFX1064-NEXT:    s_add_u32 s15, s10, 2
 ; GFX1064-NEXT:    s_addc_u32 s16, s11, 0
-; GFX1064-NEXT:    s_cmp_lg_u32 s9, 0
-; GFX1064-NEXT:    s_cselect_b32 s13, s15, s13
+; GFX1064-NEXT:    s_cmp_lg_u32 s8, 0
+; GFX1064-NEXT:    s_cselect_b32 s15, s15, s9
 ; GFX1064-NEXT:    s_cselect_b32 s14, s16, s14
 ; GFX1064-NEXT:    s_cmp_lg_u64 s[4:5], 0
-; GFX1064-NEXT:    s_subb_u32 s3, s3, s8
+; GFX1064-NEXT:    s_subb_u32 s3, s3, s12
 ; GFX1064-NEXT:    s_cmp_ge_u32 s3, s1
 ; GFX1064-NEXT:    s_cselect_b32 s4, -1, 0
-; GFX1064-NEXT:    s_cmp_ge_u32 s12, s0
+; GFX1064-NEXT:    s_cmp_ge_u32 s13, s0
 ; GFX1064-NEXT:    s_cselect_b32 s5, -1, 0
 ; GFX1064-NEXT:    s_cmp_eq_u32 s3, s1
 ; GFX1064-NEXT:    s_cselect_b32 s1, s5, s4
 ; GFX1064-NEXT:    s_cmp_lg_u32 s1, 0
 ; GFX1064-NEXT:    s_cselect_b32 s5, s14, s11
-; GFX1064-NEXT:    s_cselect_b32 s4, s13, s10
+; GFX1064-NEXT:    s_cselect_b32 s4, s15, s10
 ; GFX1064-NEXT:    s_cbranch_execnz .LBB15_3
 ; GFX1064-NEXT:  .LBB15_2:
 ; GFX1064-NEXT:    v_cvt_f32_u32_e32 v0, s0
diff --git a/llvm/test/CodeGen/AMDGPU/workitem-intrinsic-opts.ll b/llvm/test/CodeGen/AMDGPU/workitem-intrinsic-opts.ll
index 4445383..64d055b 100644
--- a/llvm/test/CodeGen/AMDGPU/workitem-intrinsic-opts.ll
+++ b/llvm/test/CodeGen/AMDGPU/workitem-intrinsic-opts.ll
@@ -271,6 +271,7 @@ define i1 @workgroup_nonzero() {
 ; DAGISEL-GFX8-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; DAGISEL-GFX8-NEXT:    s_or_b32 s4, s12, s13
 ; DAGISEL-GFX8-NEXT:    s_or_b32 s4, s4, s14
+; DAGISEL-GFX8-NEXT:    s_cmp_lg_u32 s4, 0
 ; DAGISEL-GFX8-NEXT:    s_cselect_b64 s[4:5], -1, 0
 ; DAGISEL-GFX8-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[4:5]
 ; DAGISEL-GFX8-NEXT:    s_setpc_b64 s[30:31]
@@ -280,6 +281,7 @@ define i1 @workgroup_nonzero() {
 ; DAGISEL-GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; DAGISEL-GFX942-NEXT:    s_or_b32 s0, s12, s13
 ; DAGISEL-GFX942-NEXT:    s_or_b32 s0, s0, s14
+; DAGISEL-GFX942-NEXT:    s_cmp_lg_u32 s0, 0
 ; DAGISEL-GFX942-NEXT:    s_cselect_b64 s[0:1], -1, 0
 ; DAGISEL-GFX942-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s[0:1]
 ; DAGISEL-GFX942-NEXT:    s_setpc_b64 s[30:31]
@@ -297,6 +299,8 @@ define i1 @workgroup_nonzero() {
 ; DAGISEL-GFX12-NEXT:    s_or_b32 s0, ttmp9, s0
 ; DAGISEL-GFX12-NEXT:    s_wait_alu 0xfffe
 ; DAGISEL-GFX12-NEXT:    s_or_b32 s0, s0, s1
+; DAGISEL-GFX12-NEXT:    s_wait_alu 0xfffe
+; DAGISEL-GFX12-NEXT:    s_cmp_lg_u32 s0, 0
 ; DAGISEL-GFX12-NEXT:    s_cselect_b32 s0, -1, 0
 ; DAGISEL-GFX12-NEXT:    s_wait_alu 0xfffe
 ; DAGISEL-GFX12-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s0
@@ -307,6 +311,7 @@ define i1 @workgroup_nonzero() {
 ; GISEL-GFX8-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GISEL-GFX8-NEXT:    s_or_b32 s4, s12, s13
 ; GISEL-GFX8-NEXT:    s_or_b32 s4, s4, s14
+; GISEL-GFX8-NEXT:    s_cmp_lg_u32 s4, 0
 ; GISEL-GFX8-NEXT:    s_cselect_b32 s4, 1, 0
 ; GISEL-GFX8-NEXT:    v_mov_b32_e32 v0, s4
 ; GISEL-GFX8-NEXT:    s_setpc_b64 s[30:31]
@@ -316,6 +321,7 @@ define i1 @workgroup_nonzero() {
 ; GISEL-GFX942-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GISEL-GFX942-NEXT:    s_or_b32 s0, s12, s13
 ; GISEL-GFX942-NEXT:    s_or_b32 s0, s0, s14
+; GISEL-GFX942-NEXT:    s_cmp_lg_u32 s0, 0
 ; GISEL-GFX942-NEXT:    s_cselect_b32 s0, 1, 0
 ; GISEL-GFX942-NEXT:    v_mov_b32_e32 v0, s0
 ; GISEL-GFX942-NEXT:    s_setpc_b64 s[30:31]
@@ -333,6 +339,8 @@ define i1 @workgroup_nonzero() {
 ; GISEL-GFX12-NEXT:    s_or_b32 s0, ttmp9, s0
 ; GISEL-GFX12-NEXT:    s_wait_alu 0xfffe
 ; GISEL-GFX12-NEXT:    s_or_b32 s0, s0, s1
+; GISEL-GFX12-NEXT:    s_wait_alu 0xfffe
+; GISEL-GFX12-NEXT:    s_cmp_lg_u32 s0, 0
 ; GISEL-GFX12-NEXT:    s_cselect_b32 s0, 1, 0
 ; GISEL-GFX12-NEXT:    s_wait_alu 0xfffe
 ; GISEL-GFX12-NEXT:    v_mov_b32_e32 v0, s0
diff --git a/llvm/test/CodeGen/RISCV/rvv/mixed-float-bf16-arith.ll b/llvm/test/CodeGen/RISCV/rvv/mixed-float-bf16-arith.ll
new file mode 100644
index 0000000..489323b
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/mixed-float-bf16-arith.ll
@@ -0,0 +1,186 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+zvfh,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+zvfh,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.riscv.vfadd.nxv1f16.nxv1f16(
+  <vscale x 1 x half>,
+  <vscale x 1 x half>,
+  <vscale x 1 x half>,
+  iXLen, iXLen);
+
+declare <vscale x 1 x i32> @llvm.riscv.vadd.nxv1i32.nxv1i32(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>,
+  iXLen);
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @test_half_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2, <vscale x 1 x half> %3, <vscale x 1 x half> %4, ptr %ptr) nounwind {
+; CHECK-LABEL: test_half_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a2, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v10, v10, v11
+; CHECK-NEXT:    vsetvli zero, zero, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    fsrm a2
+; CHECK-NEXT:    vse16.v v10, (a1)
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x half> @llvm.riscv.vfadd.nxv1f16.nxv1f16(
+    <vscale x 1 x half> poison,
+    <vscale x 1 x half> %3,
+    <vscale x 1 x half> %4,
+    iXLen 0, iXLen %2)
+
+  %b = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  call void @llvm.riscv.vse(<vscale x 1 x half> %a, ptr %ptr, iXLen %2)
+
+  ret <vscale x 1 x bfloat> %b
+}
+
+define <vscale x 1 x bfloat> @test_i32_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2, <vscale x 1 x i32> %3, <vscale x 1 x i32> %4, ptr %ptr) nounwind {
+; CHECK-LABEL: test_i32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vadd.vv v10, v10, v11
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vsetvli zero, zero, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    vse32.v v10, (a1)
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vadd.nxv1i32.nxv1i32(
+    <vscale x 1 x i32> poison,
+    <vscale x 1 x i32> %3,
+    <vscale x 1 x i32> %4,
+    iXLen %2)
+
+  %b = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  call void @llvm.riscv.vse(<vscale x 1 x i32> %a, ptr %ptr, iXLen %2)
+
+  ret <vscale x 1 x bfloat> %b
+}
+
+define <vscale x 1 x bfloat> @test_half_bf16_half(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2, <vscale x 1 x half> %3, <vscale x 1 x half> %4, ptr %ptr) nounwind {
+; CHECK-LABEL: test_half_bf16_half:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a2, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v10, v10, v11
+; CHECK-NEXT:    vsetvli zero, zero, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v9, v10, v11
+; CHECK-NEXT:    fsrm a2
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vse16.v v9, (a1)
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x half> @llvm.riscv.vfadd.nxv1f16.nxv1f16(
+    <vscale x 1 x half> poison,
+    <vscale x 1 x half> %3,
+    <vscale x 1 x half> %4,
+    iXLen 0, iXLen %2)
+
+  %b = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  %c = call <vscale x 1 x half> @llvm.riscv.vfadd.nxv1f16.nxv1f16(
+    <vscale x 1 x half> poison,
+    <vscale x 1 x half> %a,
+    <vscale x 1 x half> %4,
+    iXLen 0, iXLen %2)
+
+  store <vscale x 1 x half> %c, ptr %ptr
+
+  ret <vscale x 1 x bfloat> %b
+}
+
+define <vscale x 1 x bfloat> @test_bf16_half_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2, <vscale x 1 x half> %3, <vscale x 1 x half> %4, ptr %ptr) nounwind {
+; CHECK-LABEL: test_bf16_half_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a2, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v10, v10, v11
+; CHECK-NEXT:    vsetvli zero, zero, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    fsrm a2
+; CHECK-NEXT:    vsetvli a0, zero, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vse16.v v10, (a1)
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  %b = call <vscale x 1 x half> @llvm.riscv.vfadd.nxv1f16.nxv1f16(
+    <vscale x 1 x half> poison,
+    <vscale x 1 x half> %3,
+    <vscale x 1 x half> %4,
+    iXLen 0, iXLen %2)
+
+  %c = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %a,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  store <vscale x 1 x half> %b, ptr %ptr
+
+  ret <vscale x 1 x bfloat> %c
+}
+
+define <vscale x 1 x bfloat> @test_bf16_i16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2, <vscale x 1 x i16> %3, <vscale x 1 x i16> %4, ptr %ptr) nounwind {
+; CHECK-LABEL: test_bf16_i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a2, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    vadd.vv v9, v10, v11
+; CHECK-NEXT:    fsrm a2
+; CHECK-NEXT:    vsetvli a0, zero, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vse16.v v9, (a1)
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  %b = call <vscale x 1 x i16> @llvm.riscv.vadd.nxv1i16.nxv1i16(
+    <vscale x 1 x i16> poison,
+    <vscale x 1 x i16> %3,
+    <vscale x 1 x i16> %4,
+    iXLen %2)
+
+  store <vscale x 1 x i16> %b, ptr %ptr
+
+  ret <vscale x 1 x bfloat> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfadd-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfadd-bf.ll
new file mode 100644
index 0000000..db1b081
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfadd-bf.ll
@@ -0,0 +1,607 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfadd_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfadd_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfadd_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfadd_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfadd_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfadd_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfadd_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfadd_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfadd_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfadd_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfadd_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfadd_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x bfloat> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vsetvli zero, a1, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v16, v24, v0.t
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x bfloat> %2,
+    <vscale x 32 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfadd_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfadd_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfadd.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfadd_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfadd_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfadd.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfadd_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfadd_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfadd.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfadd_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfadd_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfadd.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfadd_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfadd_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfadd.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfadd_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfadd_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfadd_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfadd.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfclass-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfclass-bf.ll
new file mode 100644
index 0000000..d7d49b3
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfclass-bf.ll
@@ -0,0 +1,294 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i16> @llvm.riscv.vfclass.nxv1i16.nxv1bf16(
+  <vscale x 1 x i16>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x i16> @intrinsic_vfclass_v_nxv1i16_nxv1bf16(
+; CHECK-LABEL: intrinsic_vfclass_v_nxv1i16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfclass.v v8, v8
+; CHECK-NEXT:    ret
+  <vscale x 1 x bfloat> %0,
+  iXLen %1) nounwind {
+entry:
+  %a = call <vscale x 1 x i16> @llvm.riscv.vfclass.nxv1i16.nxv1bf16(
+    <vscale x 1 x i16> poison,
+    <vscale x 1 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 1 x i16> %a
+}
+
+declare <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1i16.nxv1bf16(
+  <vscale x 1 x i16>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 1 x i16> @intrinsic_vfclass_mask_v_nxv1i16_nxv1bf16(
+; CHECK-LABEL: intrinsic_vfclass_mask_v_nxv1i16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfclass.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+  <vscale x 1 x i16> %0,
+  <vscale x 1 x bfloat> %1,
+  <vscale x 1 x i1> %2,
+  iXLen %3) nounwind {
+entry:
+  %a = call <vscale x 1 x i16> @llvm.riscv.vfclass.mask.nxv1i16.nxv1bf16(
+    <vscale x 1 x i16> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x i1> %2,
+    iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x i16> %a
+}
+
+declare <vscale x 2 x i16> @llvm.riscv.vfclass.nxv2i16.nxv2bf16(
+  <vscale x 2 x i16>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x i16> @intrinsic_vfclass_v_nxv2i16_nxv2bf16(
+; CHECK-LABEL: intrinsic_vfclass_v_nxv2i16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfclass.v v8, v8
+; CHECK-NEXT:    ret
+  <vscale x 2 x bfloat> %0,
+  iXLen %1) nounwind {
+entry:
+  %a = call <vscale x 2 x i16> @llvm.riscv.vfclass.nxv2i16.nxv2bf16(
+    <vscale x 2 x i16> poison,
+    <vscale x 2 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 2 x i16> %a
+}
+
+declare <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2i16.nxv2bf16(
+  <vscale x 2 x i16>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 2 x i16> @intrinsic_vfclass_mask_v_nxv2i16_nxv2bf16(
+; CHECK-LABEL: intrinsic_vfclass_mask_v_nxv2i16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfclass.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+  <vscale x 2 x i16> %0,
+  <vscale x 2 x bfloat> %1,
+  <vscale x 2 x i1> %2,
+  iXLen %3) nounwind {
+entry:
+  %a = call <vscale x 2 x i16> @llvm.riscv.vfclass.mask.nxv2i16.nxv2bf16(
+    <vscale x 2 x i16> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x i1> %2,
+    iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x i16> %a
+}
+
+declare <vscale x 4 x i16> @llvm.riscv.vfclass.nxv4i16.nxv4bf16(
+  <vscale x 4 x i16>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x i16> @intrinsic_vfclass_v_nxv4i16_nxv4bf16(
+; CHECK-LABEL: intrinsic_vfclass_v_nxv4i16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfclass.v v8, v8
+; CHECK-NEXT:    ret
+  <vscale x 4 x bfloat> %0,
+  iXLen %1) nounwind {
+entry:
+  %a = call <vscale x 4 x i16> @llvm.riscv.vfclass.nxv4i16.nxv4bf16(
+    <vscale x 4 x i16> poison,
+    <vscale x 4 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 4 x i16> %a
+}
+
+declare <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4i16.nxv4bf16(
+  <vscale x 4 x i16>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 4 x i16> @intrinsic_vfclass_mask_v_nxv4i16_nxv4bf16(
+; CHECK-LABEL: intrinsic_vfclass_mask_v_nxv4i16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfclass.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+  <vscale x 4 x i16> %0,
+  <vscale x 4 x bfloat> %1,
+  <vscale x 4 x i1> %2,
+  iXLen %3) nounwind {
+entry:
+  %a = call <vscale x 4 x i16> @llvm.riscv.vfclass.mask.nxv4i16.nxv4bf16(
+    <vscale x 4 x i16> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x i1> %2,
+    iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x i16> %a
+}
+
+declare <vscale x 8 x i16> @llvm.riscv.vfclass.nxv8i16.nxv8bf16(
+  <vscale x 8 x i16>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x i16> @intrinsic_vfclass_v_nxv8i16_nxv8bf16(
+; CHECK-LABEL: intrinsic_vfclass_v_nxv8i16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfclass.v v8, v8
+; CHECK-NEXT:    ret
+  <vscale x 8 x bfloat> %0,
+  iXLen %1) nounwind {
+entry:
+  %a = call <vscale x 8 x i16> @llvm.riscv.vfclass.nxv8i16.nxv8bf16(
+    <vscale x 8 x i16> poison,
+    <vscale x 8 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 8 x i16> %a
+}
+
+declare <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8i16.nxv8bf16(
+  <vscale x 8 x i16>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 8 x i16> @intrinsic_vfclass_mask_v_nxv8i16_nxv8bf16(
+; CHECK-LABEL: intrinsic_vfclass_mask_v_nxv8i16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfclass.v v8, v10, v0.t
+; CHECK-NEXT:    ret
+  <vscale x 8 x i16> %0,
+  <vscale x 8 x bfloat> %1,
+  <vscale x 8 x i1> %2,
+  iXLen %3) nounwind {
+entry:
+  %a = call <vscale x 8 x i16> @llvm.riscv.vfclass.mask.nxv8i16.nxv8bf16(
+    <vscale x 8 x i16> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x i1> %2,
+    iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x i16> %a
+}
+
+declare <vscale x 16 x i16> @llvm.riscv.vfclass.nxv16i16.nxv16bf16(
+  <vscale x 16 x i16>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x i16> @intrinsic_vfclass_v_nxv16i16_nxv16bf16(
+; CHECK-LABEL: intrinsic_vfclass_v_nxv16i16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfclass.v v8, v8
+; CHECK-NEXT:    ret
+  <vscale x 16 x bfloat> %0,
+  iXLen %1) nounwind {
+entry:
+  %a = call <vscale x 16 x i16> @llvm.riscv.vfclass.nxv16i16.nxv16bf16(
+    <vscale x 16 x i16> poison,
+    <vscale x 16 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 16 x i16> %a
+}
+
+declare <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16i16.nxv16bf16(
+  <vscale x 16 x i16>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 16 x i16> @intrinsic_vfclass_mask_v_nxv16i16_nxv16bf16(
+; CHECK-LABEL: intrinsic_vfclass_mask_v_nxv16i16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfclass.v v8, v12, v0.t
+; CHECK-NEXT:    ret
+  <vscale x 16 x i16> %0,
+  <vscale x 16 x bfloat> %1,
+  <vscale x 16 x i1> %2,
+  iXLen %3) nounwind {
+entry:
+  %a = call <vscale x 16 x i16> @llvm.riscv.vfclass.mask.nxv16i16.nxv16bf16(
+    <vscale x 16 x i16> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x i1> %2,
+    iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x i16> %a
+}
+
+declare <vscale x 32 x i16> @llvm.riscv.vfclass.nxv32i16.nxv32bf16(
+  <vscale x 32 x i16>,
+  <vscale x 32 x bfloat>,
+  iXLen);
+
+define <vscale x 32 x i16> @intrinsic_vfclass_v_nxv32i16_nxv32bf16(
+; CHECK-LABEL: intrinsic_vfclass_v_nxv32i16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfclass.v v8, v8
+; CHECK-NEXT:    ret
+  <vscale x 32 x bfloat> %0,
+  iXLen %1) nounwind {
+entry:
+  %a = call <vscale x 32 x i16> @llvm.riscv.vfclass.nxv32i16.nxv32bf16(
+    <vscale x 32 x i16> poison,
+    <vscale x 32 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 32 x i16> %a
+}
+
+declare <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32i16.nxv32bf16(
+  <vscale x 32 x i16>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 32 x i16> @intrinsic_vfclass_mask_v_nxv32i16_nxv32bf16(
+; CHECK-LABEL: intrinsic_vfclass_mask_v_nxv32i16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, tu, mu
+; CHECK-NEXT:    vfclass.v v8, v16, v0.t
+; CHECK-NEXT:    ret
+  <vscale x 32 x i16> %0,
+  <vscale x 32 x bfloat> %1,
+  <vscale x 32 x i1> %2,
+  iXLen %3) nounwind {
+entry:
+  %a = call <vscale x 32 x i16> @llvm.riscv.vfclass.mask.nxv32i16.nxv32bf16(
+    <vscale x 32 x i16> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x i1> %2,
+    iXLen %3, iXLen 0)
+
+  ret <vscale x 32 x i16> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmacc-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfmacc-bf.ll
new file mode 100644
index 0000000..13821d7
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmacc-bf.ll
@@ -0,0 +1,553 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmacc_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmacc_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmacc_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmacc_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmacc_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmacc_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfmacc.vv v8, v10, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmacc_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfmacc.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmacc_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfmacc.vv v8, v12, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmacc_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfmacc.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmacc_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmacc_mask_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_mask_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmacc_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmacc_mask_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_mask_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmacc.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmacc_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmacc_mask_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_mask_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmacc.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmacc_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmacc_mask_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_mask_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmacc.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmacc_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmacc_mask_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_mask_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmacc.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %2,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmacc_vf_nxv1bf16_bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmacc_vf_nxv1bf16_bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmacc.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %2,
+    bfloat %1,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmadd-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfmadd-bf.ll
new file mode 100644
index 0000000..09fc199
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmadd-bf.ll
@@ -0,0 +1,553 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmadd_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmadd_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmadd_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmadd_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmadd_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmadd_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfmadd.vv v8, v10, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmadd_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfmadd.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmadd_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfmadd.vv v8, v12, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmadd_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfmadd.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmadd_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmadd_mask_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_mask_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmadd_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmadd_mask_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_mask_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmadd.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmadd_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmadd_mask_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_mask_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmadd.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmadd_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmadd_mask_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_mask_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmadd.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmadd_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmadd_mask_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_mask_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmadd.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %2,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmacc.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmadd_vf_nxv1bf16_bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmadd_vf_nxv1bf16_bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmadd.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %2,
+    bfloat %1,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmax-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfmax-bf.ll
new file mode 100644
index 0000000..a337d30
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmax-bf.ll
@@ -0,0 +1,571 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmax_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmax.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmax_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfmax.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmax_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfmax.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmax_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfmax.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmax_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfmax.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmax_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfmax.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmax_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfmax.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmax_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfmax.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmax_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfmax.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmax_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfmax.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmax_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfmax.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmax_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x bfloat> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    vsetvli zero, a1, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfmax.vv v8, v16, v24, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x bfloat> %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmax_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmax.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmax_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfmax.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmax.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmax_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfmax.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmax_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfmax.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmax.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmax_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfmax.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmax_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfmax.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmax.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmax_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfmax.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmax_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfmax.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmax.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmax_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfmax.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmax_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfmax.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmax.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmax_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfmax.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmax_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmax_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfmax.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmax.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmerge-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfmerge-bf.ll
new file mode 100644
index 0000000..86ba7c7
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmerge-bf.ll
@@ -0,0 +1,258 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmerge_vfm_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vfm_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa0, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x i1> %2,
+    iXLen %3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmerge_vfm_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vfm_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa0, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x i1> %2,
+    iXLen %3)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmerge_vfm_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vfm_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa0, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x i1> %2,
+    iXLen %3)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmerge_vfm_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vfm_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa0, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x i1> %2,
+    iXLen %3)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmerge_vfm_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vfm_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa0, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x i1> %2,
+    iXLen %3)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmerge_vfm_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, <vscale x 32 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vfm_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa0, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    <vscale x 32 x i1> %2,
+    iXLen %3)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat> @intrinsic_vfmerge_vzm_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x i1> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vzm_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fmv.h.x fa5, zero
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa5, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmerge.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat zeroinitializer,
+    <vscale x 1 x i1> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 2 x bfloat> @intrinsic_vfmerge_vzm_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x i1> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vzm_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fmv.h.x fa5, zero
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa5, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmerge.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat zeroinitializer,
+    <vscale x 2 x i1> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+define <vscale x 4 x bfloat> @intrinsic_vfmerge_vzm_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x i1> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vzm_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fmv.h.x fa5, zero
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa5, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmerge.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat zeroinitializer,
+    <vscale x 4 x i1> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+define <vscale x 8 x bfloat> @intrinsic_vfmerge_vzm_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x i1> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vzm_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fmv.h.x fa5, zero
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa5, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmerge.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat zeroinitializer,
+    <vscale x 8 x i1> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+define <vscale x 16 x bfloat> @intrinsic_vfmerge_vzm_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x i1> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vzm_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fmv.h.x fa5, zero
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa5, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmerge.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat zeroinitializer,
+    <vscale x 16 x i1> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 32 x bfloat> @intrinsic_vfmerge_vzm_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x i1> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmerge_vzm_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fmv.h.x fa5, zero
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfmerge.vfm v8, v8, fa5, v0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmerge.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat zeroinitializer,
+    <vscale x 32 x i1> %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmin-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfmin-bf.ll
new file mode 100644
index 0000000..37c0cf5
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmin-bf.ll
@@ -0,0 +1,571 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmin_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmin.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmin_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfmin.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmin_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfmin.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmin_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfmin.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmin_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfmin.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmin_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfmin.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmin_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfmin.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmin_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfmin.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmin_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfmin.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmin_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfmin.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmin_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfmin.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmin_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x bfloat> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    vsetvli zero, a1, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfmin.vv v8, v16, v24, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x bfloat> %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmin_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmin.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmin_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfmin.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmin.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmin_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfmin.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmin_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfmin.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmin.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmin_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfmin.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmin_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfmin.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmin.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmin_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfmin.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmin_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfmin.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmin.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmin_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfmin.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmin_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfmin.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmin.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmin_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfmin.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmin_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmin_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfmin.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmin.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmsac-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfmsac-bf.ll
new file mode 100644
index 0000000..948d219
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmsac-bf.ll
@@ -0,0 +1,553 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsac_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmsac_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmsac_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmsac_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmsac_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmsac_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfmsac.vv v8, v10, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmsac_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfmsac.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmsac_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfmsac.vv v8, v12, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmsac_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfmsac.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsac_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmsac_mask_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_mask_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmsac_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmsac_mask_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_mask_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmsac.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmsac_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmsac_mask_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_mask_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmsac.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmsac_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmsac_mask_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_mask_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmsac.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmsac_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmsac_mask_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_mask_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmsac.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %2,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsac_vf_nxv1bf16_bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsac_vf_nxv1bf16_bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsac.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %2,
+    bfloat %1,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmsub-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfmsub-bf.ll
new file mode 100644
index 0000000..6838f37
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmsub-bf.ll
@@ -0,0 +1,553 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsub_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmsub_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmsub_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmsub_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmsub_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmsub_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfmsub.vv v8, v10, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmsub_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfmsub.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmsub_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfmsub.vv v8, v12, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmsub_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfmsub.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsub_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmsub_mask_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_mask_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfmsub_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmsub_mask_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_mask_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmsub.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfmsub_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmsub_mask_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_mask_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmsub.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfmsub_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmsub_mask_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_mask_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmsub.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfmsub_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmsub_mask_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_mask_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmsub.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %2,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmsac.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfmsub_vf_nxv1bf16_bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfmsub_vf_nxv1bf16_bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmsub.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %2,
+    bfloat %1,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmul-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfmul-bf.ll
new file mode 100644
index 0000000..44bce72
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmul-bf.ll
@@ -0,0 +1,607 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmul_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmul_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmul_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmul_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmul_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmul_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmul_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmul_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmul_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmul_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmul_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmul_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x bfloat> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vsetvli zero, a1, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v16, v24, v0.t
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x bfloat> %2,
+    <vscale x 32 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmul_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmul_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmul.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmul_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmul_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmul.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmul_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmul_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmul.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmul_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmul_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmul.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmul_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmul_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmul.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmul_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmul_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfmul_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmul.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmv-bf-s.ll b/llvm/test/CodeGen/RISCV/rvv/vfmv-bf-s.ll
new file mode 100644
index 0000000..fbc73119
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmv-bf-s.ll
@@ -0,0 +1,88 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv64 -mattr=+d,+v,+experimental-zvfbfa -target-abi lp64d -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv32 -mattr=+d,+v,+experimental-zvfbfa -target-abi ilp32d -verify-machineinstrs < %s | FileCheck %s
+
+declare bfloat @llvm.riscv.vfmv.f.s.nxv1bf16(<vscale x 1 x bfloat>)
+
+define bfloat @intrinsic_vfmv.f.s_s_nxv1bf16(<vscale x 1 x bfloat> %0) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.f.s_s_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.x.s a0, v8
+; CHECK-NEXT:    fmv.h.x fa0, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call bfloat @llvm.riscv.vfmv.f.s.nxv1bf16(<vscale x 1 x bfloat> %0)
+  ret bfloat %a
+}
+
+declare bfloat @llvm.riscv.vfmv.f.s.nxv2bf16(<vscale x 2 x bfloat>)
+
+define bfloat @intrinsic_vfmv.f.s_s_nxv2bf16(<vscale x 2 x bfloat> %0) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.f.s_s_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.x.s a0, v8
+; CHECK-NEXT:    fmv.h.x fa0, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call bfloat @llvm.riscv.vfmv.f.s.nxv2bf16(<vscale x 2 x bfloat> %0)
+  ret bfloat %a
+}
+
+declare bfloat @llvm.riscv.vfmv.f.s.nxv4bf16(<vscale x 4 x bfloat>)
+
+define bfloat @intrinsic_vfmv.f.s_s_nxv4bf16(<vscale x 4 x bfloat> %0) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.f.s_s_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.x.s a0, v8
+; CHECK-NEXT:    fmv.h.x fa0, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call bfloat @llvm.riscv.vfmv.f.s.nxv4bf16(<vscale x 4 x bfloat> %0)
+  ret bfloat %a
+}
+
+declare bfloat @llvm.riscv.vfmv.f.s.nxv8bf16(<vscale x 8 x bfloat>)
+
+define bfloat @intrinsic_vfmv.f.s_s_nxv8bf16(<vscale x 8 x bfloat> %0) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.f.s_s_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.x.s a0, v8
+; CHECK-NEXT:    fmv.h.x fa0, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call bfloat @llvm.riscv.vfmv.f.s.nxv8bf16(<vscale x 8 x bfloat> %0)
+  ret bfloat %a
+}
+
+declare bfloat @llvm.riscv.vfmv.f.s.nxv16bf16(<vscale x 16 x bfloat>)
+
+define bfloat @intrinsic_vfmv.f.s_s_nxv16bf16(<vscale x 16 x bfloat> %0) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.f.s_s_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.x.s a0, v8
+; CHECK-NEXT:    fmv.h.x fa0, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call bfloat @llvm.riscv.vfmv.f.s.nxv16bf16(<vscale x 16 x bfloat> %0)
+  ret bfloat %a
+}
+
+declare bfloat @llvm.riscv.vfmv.f.s.nxv32bf16(<vscale x 32 x bfloat>)
+
+define bfloat @intrinsic_vfmv.f.s_s_nxv32bf16(<vscale x 32 x bfloat> %0) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.f.s_s_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.x.s a0, v8
+; CHECK-NEXT:    fmv.h.x fa0, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call bfloat @llvm.riscv.vfmv.f.s.nxv32bf16(<vscale x 32 x bfloat> %0)
+  ret bfloat %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmv-s-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfmv-s-bf.ll
new file mode 100644
index 0000000..a810809
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmv-s-bf.ll
@@ -0,0 +1,161 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s --check-prefixes=CHECK
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s --check-prefixes=CHECK
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmv.s.f.nxv1bf16(<vscale x 1 x bfloat>, bfloat, iXLen)
+
+define <vscale x 1 x bfloat> @intrinsic_vfmv.s.f_f_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmv.s.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.s.f.nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2)
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmv.s.f.nxv2bf16(<vscale x 2 x bfloat>, bfloat, iXLen)
+
+define <vscale x 2 x bfloat> @intrinsic_vfmv.s.f_f_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmv.s.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.s.f.nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2)
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmv.s.f.nxv4bf16(<vscale x 4 x bfloat>, bfloat, iXLen)
+
+define <vscale x 4 x bfloat> @intrinsic_vfmv.s.f_f_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmv.s.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.s.f.nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2)
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmv.s.f.nxv8bf16(<vscale x 8 x bfloat>, bfloat, iXLen)
+
+define <vscale x 8 x bfloat> @intrinsic_vfmv.s.f_f_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmv.s.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.s.f.nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2)
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmv.s.f.nxv16bf16(<vscale x 16 x bfloat>, bfloat, iXLen)
+
+define <vscale x 16 x bfloat> @intrinsic_vfmv.s.f_f_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmv.s.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.s.f.nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2)
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmv.s.f.nxv32bf16(<vscale x 32 x bfloat>, bfloat, iXLen)
+
+define <vscale x 32 x bfloat> @intrinsic_vfmv.s.f_f_nxv32bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfmv.s.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.s.f.nxv32bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2)
+  ret <vscale x 32 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat> @intrinsic_vfmv.s.f_f_zero_nxv1bf16(<vscale x 1 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_zero_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    vmv.s.x v8, zero
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.s.f.nxv1bf16(<vscale x 1 x bfloat> %0, bfloat 0.0, iXLen %1)
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 2 x bfloat> @intrinsic_vfmv.s.f_f_zero_nxv2bf16(<vscale x 2 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_zero_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    vmv.s.x v8, zero
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.s.f.nxv2bf16(<vscale x 2 x bfloat> %0, bfloat 0.0, iXLen %1)
+  ret <vscale x 2 x bfloat> %a
+}
+
+define <vscale x 4 x bfloat> @intrinsic_vfmv.s.f_f_zero_nxv4bf16(<vscale x 4 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_zero_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    vmv.s.x v8, zero
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.s.f.nxv4bf16(<vscale x 4 x bfloat> %0, bfloat 0.0, iXLen %1)
+  ret <vscale x 4 x bfloat> %a
+}
+
+define <vscale x 8 x bfloat> @intrinsic_vfmv.s.f_f_zero_nxv8bf16(<vscale x 8 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_zero_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    vmv.s.x v8, zero
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.s.f.nxv8bf16(<vscale x 8 x bfloat> %0, bfloat 0.0, iXLen %1)
+  ret <vscale x 8 x bfloat> %a
+}
+
+define <vscale x 16 x bfloat> @intrinsic_vfmv.s.f_f_zero_nxv16bf16(<vscale x 16 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_zero_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    vmv.s.x v8, zero
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.s.f.nxv16bf16(<vscale x 16 x bfloat> %0, bfloat 0.0, iXLen %1)
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 32 x bfloat> @intrinsic_vfmv.s.f_f_zero_nxv32bf16(<vscale x 32 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_zero_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    vmv.s.x v8, zero
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.s.f.nxv32bf16(<vscale x 32 x bfloat> %0, bfloat 0.0, iXLen %1)
+  ret <vscale x 32 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat> @intrinsic_vfmv.s.f_f_nxv1bf16_negzero(<vscale x 1 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.s.f_f_nxv1bf16_negzero:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    lui a1, 1048568
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    vmv.s.x v8, a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.s.f.nxv1bf16(<vscale x 1 x bfloat> %0, bfloat -0.0, iXLen %1)
+  ret <vscale x 1 x bfloat> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfmv-v-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfmv-v-bf.ll
new file mode 100644
index 0000000..f3293dd
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfmv-v-bf.ll
@@ -0,0 +1,216 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfmv.v.f.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfmv.v.f_f_nxv1bf16(bfloat %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.v.f_f_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfmv.v.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.v.f.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    bfloat %0,
+    iXLen %1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfmv.v.f.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfmv.v.f_f_nxv2bf16(bfloat %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.v.f_f_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfmv.v.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.v.f.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    bfloat %0,
+    iXLen %1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfmv.v.f.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfmv.v.f_f_nxv4bf16(bfloat %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.v.f_f_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.v.f.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    bfloat %0,
+    iXLen %1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfmv.v.f.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfmv.v.f_f_nxv8bf16(bfloat %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.v.f_f_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfmv.v.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.v.f.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    bfloat %0,
+    iXLen %1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfmv.v.f.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfmv.v.f_f_nxv16bf16(bfloat %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.v.f_f_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfmv.v.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.v.f.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    bfloat %0,
+    iXLen %1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfmv.v.f.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfmv.v.f_f_nxv32bf16(bfloat %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.v.f_f_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfmv.v.f v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.v.f.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    bfloat %0,
+    iXLen %1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat> @intrinsic_vfmv.v.f_zero_nxv1bf16(iXLen %0) nounwind {
+; CHECK-LABEL: intrinsic_vfmv.v.f_zero_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v8, 0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfmv.v.f.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    bfloat 0.0,
+    iXLen %0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 2 x bfloat> @intrinsic_vmv.v.i_zero_nxv2bf16(iXLen %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.v.i_zero_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v8, 0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfmv.v.f.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    bfloat 0.0,
+    iXLen %0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+define <vscale x 4 x bfloat> @intrinsic_vmv.v.i_zero_nxv4bf16(iXLen %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.v.i_zero_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v8, 0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfmv.v.f.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    bfloat 0.0,
+    iXLen %0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+define <vscale x 8 x bfloat> @intrinsic_vmv.v.i_zero_nxv8bf16(iXLen %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.v.i_zero_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, ma
+; CHECK-NEXT:    vmv.v.i v8, 0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfmv.v.f.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    bfloat 0.0,
+    iXLen %0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+define <vscale x 16 x bfloat> @intrinsic_vmv.v.i_zero_nxv16bf16(iXLen %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.v.i_zero_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, ma
+; CHECK-NEXT:    vmv.v.i v8, 0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfmv.v.f.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    bfloat 0.0,
+    iXLen %0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 32 x bfloat> @intrinsic_vmv.v.i_zero_nxv32bf16(iXLen %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.v.i_zero_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, ma
+; CHECK-NEXT:    vmv.v.i v8, 0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfmv.v.f.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    bfloat 0.0,
+    iXLen %0)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfncvt-rod-bf-f.ll b/llvm/test/CodeGen/RISCV/rvv/vfncvt-rod-bf-f.ll
new file mode 100644
index 0000000..7d587fd
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfncvt-rod-bf-f.ll
@@ -0,0 +1,226 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv1bf16.nxv1f32(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x float>,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfncvt_rod.f.f.w_nxv1bf16_nxv1f32(<vscale x 1 x float> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rod.f.f.w_nxv1bf16_nxv1f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv1bf16.nxv1f32(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x float> %0,
+    iXLen %1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x float>,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfncvt_mask_rod.f.f.w_nxv1bf16_nxv1f32(<vscale x 1 x bfloat> %0, <vscale x 1 x float> %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rod.f.f.w_nxv1bf16_nxv1f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv1bf16.nxv1f32(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x float> %1,
+    <vscale x 1 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv2bf16.nxv2f32(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x float>,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfncvt_rod.f.f.w_nxv2bf16_nxv2f32(<vscale x 2 x float> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rod.f.f.w_nxv2bf16_nxv2f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv2bf16.nxv2f32(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x float> %0,
+    iXLen %1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x float>,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfncvt_mask_rod.f.f.w_nxv2bf16_nxv2f32(<vscale x 2 x bfloat> %0, <vscale x 2 x float> %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rod.f.f.w_nxv2bf16_nxv2f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv2bf16.nxv2f32(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x float> %1,
+    <vscale x 2 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv4bf16.nxv4f32(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x float>,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfncvt_rod.f.f.w_nxv4bf16_nxv4f32(<vscale x 4 x float> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rod.f.f.w_nxv4bf16_nxv4f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v10, v8
+; CHECK-NEXT:    vmv.v.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv4bf16.nxv4f32(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x float> %0,
+    iXLen %1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x float>,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfncvt_mask_rod.f.f.w_nxv4bf16_nxv4f32(<vscale x 4 x bfloat> %0, <vscale x 4 x float> %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rod.f.f.w_nxv4bf16_nxv4f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v8, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv4bf16.nxv4f32(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x float> %1,
+    <vscale x 4 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv8bf16.nxv8f32(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x float>,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfncvt_rod.f.f.w_nxv8bf16_nxv8f32(<vscale x 8 x float> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rod.f.f.w_nxv8bf16_nxv8f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v12, v8
+; CHECK-NEXT:    vmv.v.v v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv8bf16.nxv8f32(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x float> %0,
+    iXLen %1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x float>,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfncvt_mask_rod.f.f.w_nxv8bf16_nxv8f32(<vscale x 8 x bfloat> %0, <vscale x 8 x float> %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rod.f.f.w_nxv8bf16_nxv8f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v8, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv8bf16.nxv8f32(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x float> %1,
+    <vscale x 8 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv16bf16.nxv16f32(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x float>,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfncvt_rod.f.f.w_nxv16bf16_nxv16f32(<vscale x 16 x float> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rod.f.f.w_nxv16bf16_nxv16f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v16, v8
+; CHECK-NEXT:    vmv.v.v v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.nxv16bf16.nxv16f32(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x float> %0,
+    iXLen %1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x float>,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfncvt_mask_rod.f.f.w_nxv16bf16_nxv16f32(<vscale x 16 x bfloat> %0, <vscale x 16 x float> %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rod.f.f.w_nxv16bf16_nxv16f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v8, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfncvt.rod.f.f.w.mask.nxv16bf16.nxv16f32(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x float> %1,
+    <vscale x 16 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfncvt-rtz-x-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfncvt-rtz-x-bf.ll
new file mode 100644
index 0000000..ee9e3d1
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfncvt-rtz-x-bf.ll
@@ -0,0 +1,270 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv1i8.nxv1bf16(
+  <vscale x 1 x i8>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x i8> @intrinsic_vfncvt_rtz.x.f.w_nxv1i8_nxv1bf16(<vscale x 1 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.x.f.w_nxv1i8_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv1i8.nxv1bf16(
+    <vscale x 1 x i8> poison,
+    <vscale x 1 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 1 x i8> %a
+}
+
+declare <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16(
+  <vscale x 1 x i8>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x i8> @intrinsic_vfncvt_mask_rtz.x.f.w_nxv1i8_nxv1bf16(<vscale x 1 x i8> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.x.f.w_nxv1i8_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv1i8.nxv1bf16(
+    <vscale x 1 x i8> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x i8> %a
+}
+
+declare <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv2i8.nxv2bf16(
+  <vscale x 2 x i8>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x i8> @intrinsic_vfncvt_rtz.x.f.w_nxv2i8_nxv2bf16(<vscale x 2 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.x.f.w_nxv2i8_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv2i8.nxv2bf16(
+    <vscale x 2 x i8> poison,
+    <vscale x 2 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 2 x i8> %a
+}
+
+declare <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16(
+  <vscale x 2 x i8>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x i8> @intrinsic_vfncvt_mask_rtz.x.f.w_nxv2i8_nxv2bf16(<vscale x 2 x i8> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.x.f.w_nxv2i8_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv2i8.nxv2bf16(
+    <vscale x 2 x i8> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x i8> %a
+}
+
+declare <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv4i8.nxv4bf16(
+  <vscale x 4 x i8>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x i8> @intrinsic_vfncvt_rtz.x.f.w_nxv4i8_nxv4bf16(<vscale x 4 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.x.f.w_nxv4i8_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv4i8.nxv4bf16(
+    <vscale x 4 x i8> poison,
+    <vscale x 4 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 4 x i8> %a
+}
+
+declare <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16(
+  <vscale x 4 x i8>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x i8> @intrinsic_vfncvt_mask_rtz.x.f.w_nxv4i8_nxv4bf16(<vscale x 4 x i8> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.x.f.w_nxv4i8_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv4i8.nxv4bf16(
+    <vscale x 4 x i8> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x i8> %a
+}
+
+declare <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv8i8.nxv8bf16(
+  <vscale x 8 x i8>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x i8> @intrinsic_vfncvt_rtz.x.f.w_nxv8i8_nxv8bf16(<vscale x 8 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.x.f.w_nxv8i8_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v10, v8
+; CHECK-NEXT:    vmv.v.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv8i8.nxv8bf16(
+    <vscale x 8 x i8> poison,
+    <vscale x 8 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 8 x i8> %a
+}
+
+declare <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16(
+  <vscale x 8 x i8>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x i8> @intrinsic_vfncvt_mask_rtz.x.f.w_nxv8i8_nxv8bf16(<vscale x 8 x i8> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.x.f.w_nxv8i8_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v8, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv8i8.nxv8bf16(
+    <vscale x 8 x i8> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x i8> %a
+}
+
+declare <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv16i8.nxv16bf16(
+  <vscale x 16 x i8>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x i8> @intrinsic_vfncvt_rtz.x.f.w_nxv16i8_nxv16bf16(<vscale x 16 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.x.f.w_nxv16i8_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v12, v8
+; CHECK-NEXT:    vmv.v.v v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv16i8.nxv16bf16(
+    <vscale x 16 x i8> poison,
+    <vscale x 16 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 16 x i8> %a
+}
+
+declare <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16(
+  <vscale x 16 x i8>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x i8> @intrinsic_vfncvt_mask_rtz.x.f.w_nxv16i8_nxv16bf16(<vscale x 16 x i8> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.x.f.w_nxv16i8_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v8, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv16i8.nxv16bf16(
+    <vscale x 16 x i8> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x i8> %a
+}
+
+declare <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv32i8.nxv32bf16(
+  <vscale x 32 x i8>,
+  <vscale x 32 x bfloat>,
+  iXLen);
+
+define <vscale x 32 x i8> @intrinsic_vfncvt_rtz.x.f.w_nxv32i8_nxv32bf16(<vscale x 32 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.x.f.w_nxv32i8_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v16, v8
+; CHECK-NEXT:    vmv.v.v v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.nxv32i8.nxv32bf16(
+    <vscale x 32 x i8> poison,
+    <vscale x 32 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 32 x i8> %a
+}
+
+declare <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16(
+  <vscale x 32 x i8>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x i8> @intrinsic_vfncvt_mask_rtz.x.f.w_nxv32i8_nxv32bf16(<vscale x 32 x i8> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.x.f.w_nxv32i8_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.x.f.w v8, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.x.f.w.mask.nxv32i8.nxv32bf16(
+    <vscale x 32 x i8> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 32 x i8> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfncvt-rtz-xu-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfncvt-rtz-xu-bf.ll
new file mode 100644
index 0000000..521f727
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfncvt-rtz-xu-bf.ll
@@ -0,0 +1,270 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv1i8.nxv1bf16(
+  <vscale x 1 x i8>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x i8> @intrinsic_vfncvt_rtz.xu.f.w_nxv1i8_nxv1bf16(<vscale x 1 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.xu.f.w_nxv1i8_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv1i8.nxv1bf16(
+    <vscale x 1 x i8> poison,
+    <vscale x 1 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 1 x i8> %a
+}
+
+declare <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16(
+  <vscale x 1 x i8>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x i8> @intrinsic_vfncvt_mask_rtz.xu.f.w_nxv1i8_nxv1bf16(<vscale x 1 x i8> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.xu.f.w_nxv1i8_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv1i8.nxv1bf16(
+    <vscale x 1 x i8> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x i8> %a
+}
+
+declare <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv2i8.nxv2bf16(
+  <vscale x 2 x i8>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x i8> @intrinsic_vfncvt_rtz.xu.f.w_nxv2i8_nxv2bf16(<vscale x 2 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.xu.f.w_nxv2i8_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv2i8.nxv2bf16(
+    <vscale x 2 x i8> poison,
+    <vscale x 2 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 2 x i8> %a
+}
+
+declare <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16(
+  <vscale x 2 x i8>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x i8> @intrinsic_vfncvt_mask_rtz.xu.f.w_nxv2i8_nxv2bf16(<vscale x 2 x i8> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.xu.f.w_nxv2i8_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv2i8.nxv2bf16(
+    <vscale x 2 x i8> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x i8> %a
+}
+
+declare <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv4i8.nxv4bf16(
+  <vscale x 4 x i8>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x i8> @intrinsic_vfncvt_rtz.xu.f.w_nxv4i8_nxv4bf16(<vscale x 4 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.xu.f.w_nxv4i8_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv4i8.nxv4bf16(
+    <vscale x 4 x i8> poison,
+    <vscale x 4 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 4 x i8> %a
+}
+
+declare <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16(
+  <vscale x 4 x i8>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x i8> @intrinsic_vfncvt_mask_rtz.xu.f.w_nxv4i8_nxv4bf16(<vscale x 4 x i8> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.xu.f.w_nxv4i8_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv4i8.nxv4bf16(
+    <vscale x 4 x i8> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x i8> %a
+}
+
+declare <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv8i8.nxv8bf16(
+  <vscale x 8 x i8>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x i8> @intrinsic_vfncvt_rtz.xu.f.w_nxv8i8_nxv8bf16(<vscale x 8 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.xu.f.w_nxv8i8_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v10, v8
+; CHECK-NEXT:    vmv.v.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv8i8.nxv8bf16(
+    <vscale x 8 x i8> poison,
+    <vscale x 8 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 8 x i8> %a
+}
+
+declare <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16(
+  <vscale x 8 x i8>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x i8> @intrinsic_vfncvt_mask_rtz.xu.f.w_nxv8i8_nxv8bf16(<vscale x 8 x i8> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.xu.f.w_nxv8i8_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v8, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv8i8.nxv8bf16(
+    <vscale x 8 x i8> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x i8> %a
+}
+
+declare <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv16i8.nxv16bf16(
+  <vscale x 16 x i8>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x i8> @intrinsic_vfncvt_rtz.xu.f.w_nxv16i8_nxv16bf16(<vscale x 16 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.xu.f.w_nxv16i8_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v12, v8
+; CHECK-NEXT:    vmv.v.v v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv16i8.nxv16bf16(
+    <vscale x 16 x i8> poison,
+    <vscale x 16 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 16 x i8> %a
+}
+
+declare <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16(
+  <vscale x 16 x i8>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x i8> @intrinsic_vfncvt_mask_rtz.xu.f.w_nxv16i8_nxv16bf16(<vscale x 16 x i8> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.xu.f.w_nxv16i8_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v8, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv16i8.nxv16bf16(
+    <vscale x 16 x i8> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x i8> %a
+}
+
+declare <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv32i8.nxv32bf16(
+  <vscale x 32 x i8>,
+  <vscale x 32 x bfloat>,
+  iXLen);
+
+define <vscale x 32 x i8> @intrinsic_vfncvt_rtz.xu.f.w_nxv32i8_nxv32bf16(<vscale x 32 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_rtz.xu.f.w_nxv32i8_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, ma
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v16, v8
+; CHECK-NEXT:    vmv.v.v v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.nxv32i8.nxv32bf16(
+    <vscale x 32 x i8> poison,
+    <vscale x 32 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 32 x i8> %a
+}
+
+declare <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16(
+  <vscale x 32 x i8>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x i8> @intrinsic_vfncvt_mask_rtz.xu.f.w_nxv32i8_nxv32bf16(<vscale x 32 x i8> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_rtz.xu.f.w_nxv32i8_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, mu
+; CHECK-NEXT:    vfncvt.rtz.xu.f.w v8, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i8> @llvm.riscv.vfncvt.rtz.xu.f.w.mask.nxv32i8.nxv32bf16(
+    <vscale x 32 x i8> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 32 x i8> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfncvt-x-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfncvt-x-bf.ll
new file mode 100644
index 0000000..ab9ebad
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfncvt-x-bf.ll
@@ -0,0 +1,288 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16(
+  <vscale x 1 x i8>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x i8> @intrinsic_vfncvt_x.f.w_nxv1i8_nxv1bf16(<vscale x 1 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_x.f.w_nxv1i8_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, ma
+; CHECK-NEXT:    vfncvt.x.f.w v9, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.nxv1i8.nxv1bf16(
+    <vscale x 1 x i8> poison,
+    <vscale x 1 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 1 x i8> %a
+}
+
+declare <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16(
+  <vscale x 1 x i8>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x i8> @intrinsic_vfncvt_mask_x.f.w_nxv1i8_nxv1bf16(<vscale x 1 x i8> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_x.f.w_nxv1i8_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, mu
+; CHECK-NEXT:    vfncvt.x.f.w v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv1i8.nxv1bf16(
+    <vscale x 1 x i8> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x i8> %a
+}
+
+declare <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16(
+  <vscale x 2 x i8>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x i8> @intrinsic_vfncvt_x.f.w_nxv2i8_nxv2bf16(<vscale x 2 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_x.f.w_nxv2i8_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, ma
+; CHECK-NEXT:    vfncvt.x.f.w v9, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.nxv2i8.nxv2bf16(
+    <vscale x 2 x i8> poison,
+    <vscale x 2 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 2 x i8> %a
+}
+
+declare <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16(
+  <vscale x 2 x i8>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x i8> @intrinsic_vfncvt_mask_x.f.w_nxv2i8_nxv2bf16(<vscale x 2 x i8> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_x.f.w_nxv2i8_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, mu
+; CHECK-NEXT:    vfncvt.x.f.w v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv2i8.nxv2bf16(
+    <vscale x 2 x i8> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x i8> %a
+}
+
+declare <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16(
+  <vscale x 4 x i8>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x i8> @intrinsic_vfncvt_x.f.w_nxv4i8_nxv4bf16(<vscale x 4 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_x.f.w_nxv4i8_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, ma
+; CHECK-NEXT:    vfncvt.x.f.w v9, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.nxv4i8.nxv4bf16(
+    <vscale x 4 x i8> poison,
+    <vscale x 4 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 4 x i8> %a
+}
+
+declare <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16(
+  <vscale x 4 x i8>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x i8> @intrinsic_vfncvt_mask_x.f.w_nxv4i8_nxv4bf16(<vscale x 4 x i8> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_x.f.w_nxv4i8_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, mu
+; CHECK-NEXT:    vfncvt.x.f.w v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv4i8.nxv4bf16(
+    <vscale x 4 x i8> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x i8> %a
+}
+
+declare <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16(
+  <vscale x 8 x i8>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x i8> @intrinsic_vfncvt_x.f.w_nxv8i8_nxv8bf16(<vscale x 8 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_x.f.w_nxv8i8_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, ma
+; CHECK-NEXT:    vfncvt.x.f.w v10, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv.v.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.nxv8i8.nxv8bf16(
+    <vscale x 8 x i8> poison,
+    <vscale x 8 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 8 x i8> %a
+}
+
+declare <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16(
+  <vscale x 8 x i8>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x i8> @intrinsic_vfncvt_mask_x.f.w_nxv8i8_nxv8bf16(<vscale x 8 x i8> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_x.f.w_nxv8i8_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, mu
+; CHECK-NEXT:    vfncvt.x.f.w v8, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv8i8.nxv8bf16(
+    <vscale x 8 x i8> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x i8> %a
+}
+
+declare <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16(
+  <vscale x 16 x i8>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x i8> @intrinsic_vfncvt_x.f.w_nxv16i8_nxv16bf16(<vscale x 16 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_x.f.w_nxv16i8_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, ma
+; CHECK-NEXT:    vfncvt.x.f.w v12, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv.v.v v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.nxv16i8.nxv16bf16(
+    <vscale x 16 x i8> poison,
+    <vscale x 16 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 16 x i8> %a
+}
+
+declare <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16(
+  <vscale x 16 x i8>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x i8> @intrinsic_vfncvt_mask_x.f.w_nxv16i8_nxv16bf16(<vscale x 16 x i8> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_x.f.w_nxv16i8_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, mu
+; CHECK-NEXT:    vfncvt.x.f.w v8, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv16i8.nxv16bf16(
+    <vscale x 16 x i8> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x i8> %a
+}
+
+declare <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16(
+  <vscale x 32 x i8>,
+  <vscale x 32 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 32 x i8> @intrinsic_vfncvt_x.f.w_nxv32i8_nxv32bf16(<vscale x 32 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_x.f.w_nxv32i8_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, ma
+; CHECK-NEXT:    vfncvt.x.f.w v16, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv.v.v v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.nxv32i8.nxv32bf16(
+    <vscale x 32 x i8> poison,
+    <vscale x 32 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 32 x i8> %a
+}
+
+declare <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16(
+  <vscale x 32 x i8>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 32 x i8> @intrinsic_vfncvt_mask_x.f.w_nxv32i8_nxv32bf16(<vscale x 32 x i8> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_x.f.w_nxv32i8_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, mu
+; CHECK-NEXT:    vfncvt.x.f.w v8, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i8> @llvm.riscv.vfncvt.x.f.w.mask.nxv32i8.nxv32bf16(
+    <vscale x 32 x i8> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 32 x i8> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfncvt-xu-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfncvt-xu-bf.ll
new file mode 100644
index 0000000..61c6803
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfncvt-xu-bf.ll
@@ -0,0 +1,288 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16(
+  <vscale x 1 x i8>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x i8> @intrinsic_vfncvt_xu.f.w_nxv1i8_nxv1bf16(<vscale x 1 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_xu.f.w_nxv1i8_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, ma
+; CHECK-NEXT:    vfncvt.xu.f.w v9, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv1i8.nxv1bf16(
+    <vscale x 1 x i8> poison,
+    <vscale x 1 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 1 x i8> %a
+}
+
+declare <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16(
+  <vscale x 1 x i8>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x i8> @intrinsic_vfncvt_mask_xu.f.w_nxv1i8_nxv1bf16(<vscale x 1 x i8> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_xu.f.w_nxv1i8_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, mu
+; CHECK-NEXT:    vfncvt.xu.f.w v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv1i8.nxv1bf16(
+    <vscale x 1 x i8> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x i8> %a
+}
+
+declare <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16(
+  <vscale x 2 x i8>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x i8> @intrinsic_vfncvt_xu.f.w_nxv2i8_nxv2bf16(<vscale x 2 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_xu.f.w_nxv2i8_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, ma
+; CHECK-NEXT:    vfncvt.xu.f.w v9, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv2i8.nxv2bf16(
+    <vscale x 2 x i8> poison,
+    <vscale x 2 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 2 x i8> %a
+}
+
+declare <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16(
+  <vscale x 2 x i8>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x i8> @intrinsic_vfncvt_mask_xu.f.w_nxv2i8_nxv2bf16(<vscale x 2 x i8> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_xu.f.w_nxv2i8_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, mu
+; CHECK-NEXT:    vfncvt.xu.f.w v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv2i8.nxv2bf16(
+    <vscale x 2 x i8> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x i8> %a
+}
+
+declare <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16(
+  <vscale x 4 x i8>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x i8> @intrinsic_vfncvt_xu.f.w_nxv4i8_nxv4bf16(<vscale x 4 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_xu.f.w_nxv4i8_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, ma
+; CHECK-NEXT:    vfncvt.xu.f.w v9, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv4i8.nxv4bf16(
+    <vscale x 4 x i8> poison,
+    <vscale x 4 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 4 x i8> %a
+}
+
+declare <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16(
+  <vscale x 4 x i8>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x i8> @intrinsic_vfncvt_mask_xu.f.w_nxv4i8_nxv4bf16(<vscale x 4 x i8> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_xu.f.w_nxv4i8_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, mu
+; CHECK-NEXT:    vfncvt.xu.f.w v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv4i8.nxv4bf16(
+    <vscale x 4 x i8> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x i8> %a
+}
+
+declare <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16(
+  <vscale x 8 x i8>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x i8> @intrinsic_vfncvt_xu.f.w_nxv8i8_nxv8bf16(<vscale x 8 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_xu.f.w_nxv8i8_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, ma
+; CHECK-NEXT:    vfncvt.xu.f.w v10, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv.v.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv8i8.nxv8bf16(
+    <vscale x 8 x i8> poison,
+    <vscale x 8 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 8 x i8> %a
+}
+
+declare <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16(
+  <vscale x 8 x i8>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x i8> @intrinsic_vfncvt_mask_xu.f.w_nxv8i8_nxv8bf16(<vscale x 8 x i8> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_xu.f.w_nxv8i8_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, mu
+; CHECK-NEXT:    vfncvt.xu.f.w v8, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv8i8.nxv8bf16(
+    <vscale x 8 x i8> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x i8> %a
+}
+
+declare <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16(
+  <vscale x 16 x i8>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x i8> @intrinsic_vfncvt_xu.f.w_nxv16i8_nxv16bf16(<vscale x 16 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_xu.f.w_nxv16i8_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, ma
+; CHECK-NEXT:    vfncvt.xu.f.w v12, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv.v.v v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv16i8.nxv16bf16(
+    <vscale x 16 x i8> poison,
+    <vscale x 16 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 16 x i8> %a
+}
+
+declare <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16(
+  <vscale x 16 x i8>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x i8> @intrinsic_vfncvt_mask_xu.f.w_nxv16i8_nxv16bf16(<vscale x 16 x i8> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_xu.f.w_nxv16i8_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, mu
+; CHECK-NEXT:    vfncvt.xu.f.w v8, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv16i8.nxv16bf16(
+    <vscale x 16 x i8> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x i8> %a
+}
+
+declare <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16(
+  <vscale x 32 x i8>,
+  <vscale x 32 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 32 x i8> @intrinsic_vfncvt_xu.f.w_nxv32i8_nxv32bf16(<vscale x 32 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_xu.f.w_nxv32i8_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, ma
+; CHECK-NEXT:    vfncvt.xu.f.w v16, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv.v.v v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.nxv32i8.nxv32bf16(
+    <vscale x 32 x i8> poison,
+    <vscale x 32 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 32 x i8> %a
+}
+
+declare <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16(
+  <vscale x 32 x i8>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 32 x i8> @intrinsic_vfncvt_mask_xu.f.w_nxv32i8_nxv32bf16(<vscale x 32 x i8> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfncvt_mask_xu.f.w_nxv32i8_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, mu
+; CHECK-NEXT:    vfncvt.xu.f.w v8, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i8> @llvm.riscv.vfncvt.xu.f.w.mask.nxv32i8.nxv32bf16(
+    <vscale x 32 x i8> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 32 x i8> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfnmacc-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfnmacc-bf.ll
new file mode 100644
index 0000000..4b4091b
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfnmacc-bf.ll
@@ -0,0 +1,553 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfnmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmacc_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfnmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmacc_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfnmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmacc_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfnmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmacc_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfnmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmacc_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfnmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmacc_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfnmacc.vv v8, v10, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmacc_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfnmacc.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmacc_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfnmacc.vv v8, v12, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmacc_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfnmacc.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmacc_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfnmacc_mask_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_mask_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmacc_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfnmacc_mask_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_mask_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmacc.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmacc_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfnmacc_mask_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_mask_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmacc.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmacc_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfnmacc_mask_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_mask_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmacc.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmacc_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfnmacc_mask_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_mask_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmacc.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %2,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmacc_vf_nxv1bf16_bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmacc_vf_nxv1bf16_bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmacc.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %2,
+    bfloat %1,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfnmadd-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfnmadd-bf.ll
new file mode 100644
index 0000000..2bb6bf5
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfnmadd-bf.ll
@@ -0,0 +1,553 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmadd_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfnmadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmadd_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmadd_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfnmadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmadd_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmadd_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfnmadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmadd_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v10, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmadd_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfnmadd.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmadd_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v12, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmadd_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfnmadd.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmadd_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfnmadd_mask_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_mask_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmadd_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfnmadd_mask_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_mask_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmadd.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmadd_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfnmadd_mask_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_mask_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmadd.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmadd_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfnmadd_mask_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_mask_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmadd.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmadd_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfnmadd_mask_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_mask_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmadd.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v9, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %2,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmacc.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmadd_vf_nxv1bf16_bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmadd_vf_nxv1bf16_bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmadd.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %2,
+    bfloat %1,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfnmsac-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfnmsac-bf.ll
new file mode 100644
index 0000000..cfbaafa
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfnmsac-bf.ll
@@ -0,0 +1,553 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfnmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsac_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfnmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmsac_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfnmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmsac_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfnmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmsac_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfnmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmsac_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfnmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmsac_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfnmsac.vv v8, v10, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmsac_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfnmsac.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmsac_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfnmsac.vv v8, v12, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmsac_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfnmsac.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsac_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfnmsac_mask_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_mask_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmsac_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfnmsac_mask_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_mask_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsac.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmsac_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfnmsac_mask_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_mask_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsac.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmsac_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfnmsac_mask_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_mask_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsac.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmsac_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfnmsac_mask_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_mask_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsac.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %2,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsac_vf_nxv1bf16_bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsac_vf_nxv1bf16_bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsac.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %2,
+    bfloat %1,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfnmsub-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfnmsub-bf.ll
new file mode 100644
index 0000000..5ebbb90c
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfnmsub-bf.ll
@@ -0,0 +1,553 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsub_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfnmsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmsub_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmsub_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfnmsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmsub_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmsub_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfnmsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmsub_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v10, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmsub_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfnmsub.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmsub_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v12, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmsub_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfnmsub.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsub_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfnmsub_mask_vf_nxv1bf16_bf16_nxv1bf16(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_mask_vf_nxv1bf16_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat>  @intrinsic_vfnmsub_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfnmsub_mask_vf_nxv2bf16_bf16_nxv2bf16(<vscale x 2 x bfloat> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_mask_vf_nxv2bf16_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfnmsub.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat>  @intrinsic_vfnmsub_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfnmsub_mask_vf_nxv4bf16_bf16_nxv4bf16(<vscale x 4 x bfloat> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_mask_vf_nxv4bf16_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfnmsub.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat>  @intrinsic_vfnmsub_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfnmsub_mask_vf_nxv8bf16_bf16_nxv8bf16(<vscale x 8 x bfloat> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_mask_vf_nxv8bf16_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfnmsub.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat>  @intrinsic_vfnmsub_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfnmsub_mask_vf_nxv16bf16_bf16_nxv16bf16(<vscale x 16 x bfloat> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_mask_vf_nxv16bf16_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfnmsub.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0);
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v9, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %2,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vv_nxv1bf16_nxv1bf16_nxv1bf16_commute2:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmsac.vv v8, v10, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+define <vscale x 1 x bfloat>  @intrinsic_vfnmsub_vf_nxv1bf16_bf16_nxv1bf16_commute(<vscale x 1 x bfloat> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfnmsub_vf_nxv1bf16_bf16_nxv1bf16_commute:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfnmsub.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %2,
+    bfloat %1,
+    <vscale x 1 x bfloat> %0,
+    iXLen 7, iXLen %3, iXLen 3)
+
+  ret <vscale x 1 x bfloat> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfrec7-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfrec7-bf.ll
new file mode 100644
index 0000000..1211415
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfrec7-bf.ll
@@ -0,0 +1,282 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfrec7.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfrec7_v_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_v_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfrec7.v v8, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfrec7_mask_v_nxv1bf16_nxv1bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_mask_v_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfrec7.v v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfrec7.mask.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %0,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfrec7.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfrec7_v_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_v_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfrec7.v v8, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfrec7_mask_v_nxv2bf16_nxv2bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_mask_v_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfrec7.v v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfrec7.mask.nxv2bf16(
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %0,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfrec7.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfrec7_v_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_v_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfrec7.v v8, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfrec7_mask_v_nxv4bf16_nxv4bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_mask_v_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfrec7.v v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfrec7.mask.nxv4bf16(
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %0,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfrec7.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfrec7_v_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_v_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfrec7.v v8, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfrec7_mask_v_nxv8bf16_nxv8bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_mask_v_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfrec7.v v8, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfrec7.mask.nxv8bf16(
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %0,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfrec7.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfrec7_v_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_v_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfrec7.v v8, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfrec7_mask_v_nxv16bf16_nxv16bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_mask_v_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfrec7.v v8, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfrec7.mask.nxv16bf16(
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %0,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfrec7.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfrec7_v_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_v_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfrec7.v v8, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    iXLen 0, iXLen %1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfrec7_mask_v_nxv32bf16_nxv32bf16(<vscale x 32 x i1> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrec7_mask_v_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfrec7.v v8, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfrec7.mask.nxv32bf16(
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x bfloat> %2,
+    <vscale x 32 x i1> %0,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfrsqrt7-bf16.ll b/llvm/test/CodeGen/RISCV/rvv/vfrsqrt7-bf16.ll
new file mode 100644
index 0000000..4626b86
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfrsqrt7-bf16.ll
@@ -0,0 +1,264 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfrsqrt7_v_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_v_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfrsqrt7.v v8, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfrsqrt7_mask_v_nxv1bf16_nxv1bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_mask_v_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfrsqrt7.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %0,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfrsqrt7_v_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_v_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfrsqrt7.v v8, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfrsqrt7_mask_v_nxv2bf16_nxv2bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_mask_v_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfrsqrt7.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv2bf16(
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %0,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfrsqrt7_v_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_v_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfrsqrt7.v v8, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfrsqrt7_mask_v_nxv4bf16_nxv4bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_mask_v_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfrsqrt7.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv4bf16(
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %0,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfrsqrt7_v_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_v_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfrsqrt7.v v8, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfrsqrt7_mask_v_nxv8bf16_nxv8bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_mask_v_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfrsqrt7.v v8, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv8bf16(
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %0,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfrsqrt7_v_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_v_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfrsqrt7.v v8, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfrsqrt7_mask_v_nxv16bf16_nxv16bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_mask_v_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfrsqrt7.v v8, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv16bf16(
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %0,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfrsqrt7_v_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_v_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfrsqrt7.v v8, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    iXLen %1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfrsqrt7_mask_v_nxv32bf16_nxv32bf16(<vscale x 32 x i1> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfrsqrt7_mask_v_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfrsqrt7.v v8, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfrsqrt7.mask.nxv32bf16(
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x bfloat> %2,
+    <vscale x 32 x i1> %0,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfrsub-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfrsub-bf.ll
new file mode 100644
index 0000000..54a6d48
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfrsub-bf.ll
@@ -0,0 +1,282 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfrsub.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfrsub_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfrsub_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfrsub.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfrsub.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfrsub_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfrsub_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfrsub.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfrsub.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfrsub_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfrsub_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfrsub.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfrsub.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfrsub_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfrsub_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfrsub.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfrsub.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfrsub_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfrsub_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfrsub.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfrsub.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfrsub_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfrsub_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfrsub_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfrsub.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfsgnj-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfsgnj-bf.ll
new file mode 100644
index 0000000..2cd698d
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfsgnj-bf.ll
@@ -0,0 +1,571 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnj_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfsgnj.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnj_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfsgnj.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnj_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfsgnj.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnj_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfsgnj.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnj_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfsgnj.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnj_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfsgnj.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnj_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfsgnj.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnj_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfsgnj.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnj_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfsgnj.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnj_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfsgnj.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnj_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfsgnj.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnj_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x bfloat> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    vsetvli zero, a1, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfsgnj.vv v8, v16, v24, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x bfloat> %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnj_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfsgnj.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnj_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfsgnj.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnj.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnj_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfsgnj.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnj_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfsgnj.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnj.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnj_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfsgnj.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnj_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfsgnj.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnj.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnj_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfsgnj.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnj_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfsgnj.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnj.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnj_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfsgnj.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnj_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfsgnj.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnj.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnj_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfsgnj.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnj_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnj_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfsgnj.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnj.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfsgnjn-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfsgnjn-bf.ll
new file mode 100644
index 0000000..08340be
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfsgnjn-bf.ll
@@ -0,0 +1,571 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnjn_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfsgnjn.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnjn_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfsgnjn.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnjn_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfsgnjn.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnjn_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfsgnjn.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnjn_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfsgnjn.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnjn_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfsgnjn.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnjn_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfsgnjn.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnjn_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfsgnjn.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnjn_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfsgnjn.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnjn_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfsgnjn.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnjn_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfsgnjn.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnjn_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x bfloat> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    vsetvli zero, a1, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfsgnjn.vv v8, v16, v24, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x bfloat> %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnjn_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfsgnjn.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnjn_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfsgnjn.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnjn_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfsgnjn.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnjn_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfsgnjn.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnjn_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfsgnjn.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnjn_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfsgnjn.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnjn_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfsgnjn.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnjn_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfsgnjn.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnjn_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfsgnjn.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnjn_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfsgnjn.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnjn_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfsgnjn.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnjn_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjn_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfsgnjn.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjn.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfsgnjx-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfsgnjx-bf.ll
new file mode 100644
index 0000000..e51a42e
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfsgnjx-bf.ll
@@ -0,0 +1,571 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnjx_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfsgnjx.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnjx_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfsgnjx.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnjx_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfsgnjx.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnjx_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfsgnjx.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnjx_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfsgnjx.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnjx_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfsgnjx.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnjx_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfsgnjx.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnjx_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfsgnjx.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnjx_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfsgnjx.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnjx_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfsgnjx.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnjx_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfsgnjx.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnjx_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x bfloat> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    vsetvli zero, a1, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfsgnjx.vv v8, v16, v24, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x bfloat> %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnjx_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfsgnjx.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsgnjx_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfsgnjx.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnjx_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfsgnjx.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsgnjx_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfsgnjx.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnjx_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfsgnjx.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsgnjx_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfsgnjx.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnjx_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfsgnjx.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsgnjx_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfsgnjx.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnjx_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfsgnjx.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsgnjx_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfsgnjx.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnjx_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfsgnjx.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsgnjx_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsgnjx_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfsgnjx.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsgnjx.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfslide1down-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfslide1down-bf.ll
new file mode 100644
index 0000000..c65719c
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfslide1down-bf.ll
@@ -0,0 +1,288 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfslide1down_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfslide1down.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfslide1down_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfslide1down.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1down.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfslide1down_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfslide1down.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfslide1down_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfslide1down.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1down.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfslide1down_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfslide1down.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfslide1down_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfslide1down.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1down.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfslide1down_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfslide1down.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfslide1down_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfslide1down.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1down.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfslide1down_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfslide1down.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfslide1down_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfslide1down.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1down.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfslide1down_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfslide1down.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfslide1down_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1down_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfslide1down.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1down.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfslide1up-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfslide1up-bf.ll
new file mode 100644
index 0000000..57a4898
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfslide1up-bf.ll
@@ -0,0 +1,294 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfslide1up_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfslide1up.vf v9, v8, fa0
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfslide1up_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfslide1up.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfslide1up.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfslide1up_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfslide1up.vf v9, v8, fa0
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfslide1up_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfslide1up.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfslide1up.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfslide1up_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfslide1up.vf v9, v8, fa0
+; CHECK-NEXT:    vmv.v.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfslide1up_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfslide1up.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfslide1up.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfslide1up_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfslide1up.vf v10, v8, fa0
+; CHECK-NEXT:    vmv.v.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfslide1up_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfslide1up.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfslide1up.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfslide1up_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfslide1up.vf v12, v8, fa0
+; CHECK-NEXT:    vmv.v.v v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfslide1up_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfslide1up.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfslide1up.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfslide1up_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfslide1up.vf v16, v8, fa0
+; CHECK-NEXT:    vmv.v.v v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfslide1up_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfslide1up_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfslide1up.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfslide1up.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfsub-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfsub-bf.ll
new file mode 100644
index 0000000..aea7521
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfsub-bf.ll
@@ -0,0 +1,559 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsub_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsub_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vv_nxv1bf16_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsub_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsub_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vv_nxv2bf16_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsub_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsub_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vv_nxv4bf16_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsub_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsub_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vv_nxv8bf16_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v10, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsub_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsub_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vv_nxv16bf16_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v12, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsub_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsub_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, <vscale x 32 x bfloat> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vv_nxv32bf16_nxv32bf16_nxv32bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    vsetvli zero, a1, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v16, v24, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.nxv32bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    <vscale x 32 x bfloat> %2,
+    <vscale x 32 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsub_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfsub_mask_vf_nxv1bf16_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vf_nxv1bf16_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfsub.mask.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsub_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfsub_mask_vf_nxv2bf16_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vf_nxv2bf16_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfsub.mask.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsub_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfsub_mask_vf_nxv4bf16_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vf_nxv4bf16_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfsub.mask.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsub_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfsub_mask_vf_nxv8bf16_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vf_nxv8bf16_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfsub.mask.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsub_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfsub_mask_vf_nxv16bf16_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vf_nxv16bf16_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfsub.mask.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsub_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x bfloat> %0,
+    bfloat %1,
+    iXLen 7, iXLen %2)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x bfloat>,
+  bfloat,
+  <vscale x 32 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfsub_mask_vf_nxv32bf16_nxv32bf16_bf16(<vscale x 32 x bfloat> %0, <vscale x 32 x bfloat> %1, bfloat %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfsub_mask_vf_nxv32bf16_nxv32bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m8, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfsub.mask.nxv32bf16.bf16(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x bfloat> %1,
+    bfloat %2,
+    <vscale x 32 x i1> %3,
+    iXLen 7, iXLen %4, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwadd-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfwadd-bf.ll
new file mode 100644
index 0000000..62feac8
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwadd-bf.ll
@@ -0,0 +1,519 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwadd_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwadd.vv v10, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.nxv1bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwadd_mask_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_mask_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwadd_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.vv v10, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.nxv2bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwadd_mask_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_mask_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwadd.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwadd_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmv1r.v v10, v9
+; CHECK-NEXT:    vmv1r.v v11, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwadd.vv v8, v11, v10
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.nxv4bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwadd_mask_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_mask_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwadd.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwadd_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmv2r.v v12, v10
+; CHECK-NEXT:    vmv2r.v v14, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwadd.vv v8, v14, v12
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.nxv8bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwadd_mask_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_mask_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwadd.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwadd_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmv4r.v v16, v12
+; CHECK-NEXT:    vmv4r.v v20, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwadd.vv v8, v20, v16
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.nxv16bf16(
+    <vscale x 16 x float> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwadd_mask_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_mask_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwadd.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwadd_vf_nxv1f32_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_vf_nxv1f32_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwadd.vf v9, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.nxv1f32.nxv1bf16.bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwadd_mask_vf_nxv1f32_nxv1bf16_bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_mask_vf_nxv1f32_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwadd.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.mask.nxv1f32.nxv1bf16.bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwadd_vf_nxv2f32_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_vf_nxv2f32_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.vf v9, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.nxv2f32.nxv2bf16.bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwadd_mask_vf_nxv2f32_nxv2bf16_bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_mask_vf_nxv2f32_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwadd.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.mask.nxv2f32.nxv2bf16.bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwadd_vf_nxv4f32_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_vf_nxv4f32_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmv1r.v v10, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwadd.vf v8, v10, fa0
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.nxv4f32.nxv4bf16.bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwadd_mask_vf_nxv4f32_nxv4bf16_bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_mask_vf_nxv4f32_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwadd.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.mask.nxv4f32.nxv4bf16.bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwadd_vf_nxv8f32_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_vf_nxv8f32_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmv2r.v v12, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwadd.vf v8, v12, fa0
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.nxv8f32.nxv8bf16.bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwadd_mask_vf_nxv8f32_nxv8bf16_bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_mask_vf_nxv8f32_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwadd.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.mask.nxv8f32.nxv8bf16.bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwadd_vf_nxv16f32_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_vf_nxv16f32_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmv4r.v v16, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwadd.vf v8, v16, fa0
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwadd.nxv16f32.nxv16bf16.bf16(
+    <vscale x 16 x float> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwadd_mask_vf_nxv16f32_nxv16bf16_bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd_mask_vf_nxv16f32_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwadd.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwadd.mask.nxv16f32.nxv16bf16.bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwadd-w-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfwadd-w-bf.ll
new file mode 100644
index 0000000..c5417e8
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwadd-w-bf.ll
@@ -0,0 +1,773 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwadd.w_wv_nxv1f32_nxv1f32_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wv_nxv1f32_nxv1f32_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwadd.wv v8, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwadd.w_mask_wv_nxv1f32_nxv1f32_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x float> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wv_nxv1f32_nxv1f32_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwadd.wv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x float> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwadd.w_wv_nxv2f32_nxv2f32_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wv_nxv2f32_nxv2f32_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.wv v8, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwadd.w_mask_wv_nxv2f32_nxv2f32_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x float> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wv_nxv2f32_nxv2f32_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwadd.wv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x float> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwadd.w_wv_nxv4f32_nxv4f32_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wv_nxv4f32_nxv4f32_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfwadd.wv v8, v8, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwadd.w_mask_wv_nxv4f32_nxv4f32_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x float> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wv_nxv4f32_nxv4f32_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwadd.wv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x float> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwadd.w_wv_nxv8f32_nxv8f32_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wv_nxv8f32_nxv8f32_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfwadd.wv v8, v8, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwadd.w_mask_wv_nxv8f32_nxv8f32_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x float> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wv_nxv8f32_nxv8f32_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwadd.wv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x float> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwadd.w_wv_nxv16f32_nxv16f32_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wv_nxv16f32_nxv16f32_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfwadd.wv v8, v8, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> poison,
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwadd.w_mask_wv_nxv16f32_nxv16f32_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x float> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wv_nxv16f32_nxv16f32_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vl4re16.v v24, (a0)
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vsetvli zero, a1, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwadd.wv v8, v16, v24, v0.t
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x float> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x float>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwadd.w_wf_nxv1f32_nxv1f32_bf16(<vscale x 1 x float> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wf_nxv1f32_nxv1f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x float> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x float>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwadd.w_mask_wf_nxv1f32_nxv1f32_bf16(<vscale x 1 x float> %0, <vscale x 1 x float> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wf_nxv1f32_nxv1f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwadd.wf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x float> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x float>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwadd.w_wf_nxv2f32_nxv2f32_bf16(<vscale x 2 x float> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wf_nxv2f32_nxv2f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x float> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x float>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwadd.w_mask_wf_nxv2f32_nxv2f32_bf16(<vscale x 2 x float> %0, <vscale x 2 x float> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wf_nxv2f32_nxv2f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwadd.wf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x float> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x float>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwadd.w_wf_nxv4f32_nxv4f32_bf16(<vscale x 4 x float> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wf_nxv4f32_nxv4f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x float> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x float>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwadd.w_mask_wf_nxv4f32_nxv4f32_bf16(<vscale x 4 x float> %0, <vscale x 4 x float> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wf_nxv4f32_nxv4f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwadd.wf v8, v10, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x float> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x float>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwadd.w_wf_nxv8f32_nxv8f32_bf16(<vscale x 8 x float> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wf_nxv8f32_nxv8f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x float> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x float>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwadd.w_mask_wf_nxv8f32_nxv8f32_bf16(<vscale x 8 x float> %0, <vscale x 8 x float> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wf_nxv8f32_nxv8f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwadd.wf v8, v12, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x float> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x float>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwadd.w_wf_nxv16f32_nxv16f32_bf16(<vscale x 16 x float> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wf_nxv16f32_nxv16f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.nxv16f32.bf16(
+    <vscale x 16 x float> poison,
+    <vscale x 16 x float> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x float>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwadd.w_mask_wf_nxv16f32_nxv16f32_bf16(<vscale x 16 x float> %0, <vscale x 16 x float> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wf_nxv16f32_nxv16f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwadd.wf v8, v16, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x float> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+define <vscale x 1 x float> @intrinsic_vfwadd.w_mask_wv_tie_nxv1f32_nxv1f32_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wv_tie_nxv1f32_nxv1f32_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwadd.wv v8, v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+define <vscale x 2 x float> @intrinsic_vfwadd.w_mask_wv_tie_nxv2f32_nxv2f32_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wv_tie_nxv2f32_nxv2f32_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwadd.wv v8, v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+define <vscale x 4 x float> @intrinsic_vfwadd.w_mask_wv_tie_nxv4f32_nxv4f32_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wv_tie_nxv4f32_nxv4f32_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwadd.wv v8, v8, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+define <vscale x 8 x float> @intrinsic_vfwadd.w_mask_wv_tie_nxv8f32_nxv8f32_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wv_tie_nxv8f32_nxv8f32_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwadd.wv v8, v8, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+define <vscale x 16 x float> @intrinsic_vfwadd.w_mask_wv_tie_nxv16f32_nxv16f32_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wv_tie_nxv16f32_nxv16f32_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwadd.wv v8, v8, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+define <vscale x 1 x float> @intrinsic_vfwadd.w_mask_wf_tie_nxv1f32_nxv1f32_bf16(<vscale x 1 x float> %0, bfloat %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wf_tie_nxv1f32_nxv1f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.mask.nxv1f32.bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x float> %0,
+    bfloat %1,
+    <vscale x 1 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+define <vscale x 2 x float> @intrinsic_vfwadd.w_mask_wf_tie_nxv2f32_nxv2f32_bf16(<vscale x 2 x float> %0, bfloat %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wf_tie_nxv2f32_nxv2f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.mask.nxv2f32.bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x float> %0,
+    bfloat %1,
+    <vscale x 2 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+define <vscale x 4 x float> @intrinsic_vfwadd.w_mask_wf_tie_nxv4f32_nxv4f32_bf16(<vscale x 4 x float> %0, bfloat %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wf_tie_nxv4f32_nxv4f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.mask.nxv4f32.bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x float> %0,
+    bfloat %1,
+    <vscale x 4 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+define <vscale x 8 x float> @intrinsic_vfwadd.w_mask_wf_tie_nxv8f32_nxv8f32_bf16(<vscale x 8 x float> %0, bfloat %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wf_tie_nxv8f32_nxv8f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.mask.nxv8f32.bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x float> %0,
+    bfloat %1,
+    <vscale x 8 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+define <vscale x 16 x float> @intrinsic_vfwadd.w_mask_wf_tie_nxv16f32_nxv16f32_bf16(<vscale x 16 x float> %0, bfloat %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_mask_wf_tie_nxv16f32_nxv16f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwadd.w.mask.nxv16f32.bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x float> %0,
+    bfloat %1,
+    <vscale x 16 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+define <vscale x 1 x float> @intrinsic_vfwadd.w_wv_untie_nxv1f32_nxv1f32_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x float> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wv_untie_nxv1f32_nxv1f32_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwadd.wv v10, v9, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwadd.w.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x float> %1,
+    <vscale x 1 x bfloat> %0,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+define <vscale x 2 x float> @intrinsic_vfwadd.w_wv_untie_nxv2f32_nxv2f32_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x float> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wv_untie_nxv2f32_nxv2f32_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.wv v10, v9, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwadd.w.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x float> %1,
+    <vscale x 2 x bfloat> %0,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+define <vscale x 4 x float> @intrinsic_vfwadd.w_wv_untie_nxv4f32_nxv4f32_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x float> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wv_untie_nxv4f32_nxv4f32_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmv1r.v v12, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwadd.wv v8, v10, v12
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwadd.w.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x float> %1,
+    <vscale x 4 x bfloat> %0,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+define <vscale x 8 x float> @intrinsic_vfwadd.w_wv_untie_nxv8f32_nxv8f32_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x float> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwadd.w_wv_untie_nxv8f32_nxv8f32_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmv2r.v v16, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwadd.wv v8, v12, v16
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwadd.w.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x float> %1,
+    <vscale x 8 x bfloat> %0,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwcvt-bf-x.ll b/llvm/test/CodeGen/RISCV/rvv/vfwcvt-bf-x.ll
new file mode 100644
index 0000000..b7df45b
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwcvt-bf-x.ll
@@ -0,0 +1,264 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv1bf16.nxv1i8(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i8>,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfwcvt_f.x.v_nxv1bf16_nxv1i8(<vscale x 1 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.x.v_nxv1bf16_nxv1i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, ma
+; CHECK-NEXT:    vfwcvt.f.x.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv1bf16.nxv1i8(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i8>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfwcvt_mask_f.x.v_nxv1bf16_nxv1i8(<vscale x 1 x bfloat> %0, <vscale x 1 x i8> %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.x.v_nxv1bf16_nxv1i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, mu
+; CHECK-NEXT:    vfwcvt.f.x.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv1bf16.nxv1i8(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x i8> %1,
+    <vscale x 1 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv2bf16.nxv2i8(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i8>,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfwcvt_f.x.v_nxv2bf16_nxv2i8(<vscale x 2 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.x.v_nxv2bf16_nxv2i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, ma
+; CHECK-NEXT:    vfwcvt.f.x.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv2bf16.nxv2i8(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i8>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfwcvt_mask_f.x.v_nxv2bf16_nxv2i8(<vscale x 2 x bfloat> %0, <vscale x 2 x i8> %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.x.v_nxv2bf16_nxv2i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, mu
+; CHECK-NEXT:    vfwcvt.f.x.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv2bf16.nxv2i8(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x i8> %1,
+    <vscale x 2 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv4bf16.nxv4i8(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i8>,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfwcvt_f.x.v_nxv4bf16_nxv4i8(<vscale x 4 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.x.v_nxv4bf16_nxv4i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, ma
+; CHECK-NEXT:    vfwcvt.f.x.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv4bf16.nxv4i8(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i8>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfwcvt_mask_f.x.v_nxv4bf16_nxv4i8(<vscale x 4 x bfloat> %0, <vscale x 4 x i8> %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.x.v_nxv4bf16_nxv4i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, mu
+; CHECK-NEXT:    vfwcvt.f.x.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv4bf16.nxv4i8(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x i8> %1,
+    <vscale x 4 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv8bf16.nxv8i8(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i8>,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfwcvt_f.x.v_nxv8bf16_nxv8i8(<vscale x 8 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.x.v_nxv8bf16_nxv8i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, ma
+; CHECK-NEXT:    vmv1r.v v10, v8
+; CHECK-NEXT:    vfwcvt.f.x.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv8bf16.nxv8i8(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i8>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfwcvt_mask_f.x.v_nxv8bf16_nxv8i8(<vscale x 8 x bfloat> %0, <vscale x 8 x i8> %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.x.v_nxv8bf16_nxv8i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, mu
+; CHECK-NEXT:    vfwcvt.f.x.v v8, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv8bf16.nxv8i8(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x i8> %1,
+    <vscale x 8 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv16bf16.nxv16i8(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i8>,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfwcvt_f.x.v_nxv16bf16_nxv16i8(<vscale x 16 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.x.v_nxv16bf16_nxv16i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, ma
+; CHECK-NEXT:    vmv2r.v v12, v8
+; CHECK-NEXT:    vfwcvt.f.x.v v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv16bf16.nxv16i8(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i8>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfwcvt_mask_f.x.v_nxv16bf16_nxv16i8(<vscale x 16 x bfloat> %0, <vscale x 16 x i8> %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.x.v_nxv16bf16_nxv16i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, mu
+; CHECK-NEXT:    vfwcvt.f.x.v v8, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv16bf16.nxv16i8(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x i8> %1,
+    <vscale x 16 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv32bf16.nxv32i8(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i8>,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfwcvt_f.x.v_nxv32bf16_nxv32i8(<vscale x 32 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.x.v_nxv32bf16_nxv32i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, ma
+; CHECK-NEXT:    vmv4r.v v16, v8
+; CHECK-NEXT:    vfwcvt.f.x.v v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.nxv32bf16.nxv32i8(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i8>,
+  <vscale x 32 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfwcvt_mask_f.x.v_nxv32bf16_nxv32i8(<vscale x 32 x bfloat> %0, <vscale x 32 x i8> %1, <vscale x 32 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.x.v_nxv32bf16_nxv32i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, mu
+; CHECK-NEXT:    vfwcvt.f.x.v v8, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.x.v.mask.nxv32bf16.nxv32i8(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x i8> %1,
+    <vscale x 32 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwcvt-bf-xu.ll b/llvm/test/CodeGen/RISCV/rvv/vfwcvt-bf-xu.ll
new file mode 100644
index 0000000..c370261
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwcvt-bf-xu.ll
@@ -0,0 +1,264 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv1bf16.nxv1i8(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i8>,
+  iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfwcvt_f.xu.v_nxv1bf16_nxv1i8(<vscale x 1 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.xu.v_nxv1bf16_nxv1i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, ma
+; CHECK-NEXT:    vfwcvt.f.xu.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv1bf16.nxv1i8(
+    <vscale x 1 x bfloat> poison,
+    <vscale x 1 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i8>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 1 x bfloat> @intrinsic_vfwcvt_mask_f.xu.v_nxv1bf16_nxv1i8(<vscale x 1 x bfloat> %0, <vscale x 1 x i8> %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.xu.v_nxv1bf16_nxv1i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf8, ta, mu
+; CHECK-NEXT:    vfwcvt.f.xu.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv1bf16.nxv1i8(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x i8> %1,
+    <vscale x 1 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv2bf16.nxv2i8(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i8>,
+  iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfwcvt_f.xu.v_nxv2bf16_nxv2i8(<vscale x 2 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.xu.v_nxv2bf16_nxv2i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, ma
+; CHECK-NEXT:    vfwcvt.f.xu.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv2bf16.nxv2i8(
+    <vscale x 2 x bfloat> poison,
+    <vscale x 2 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i8>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 2 x bfloat> @intrinsic_vfwcvt_mask_f.xu.v_nxv2bf16_nxv2i8(<vscale x 2 x bfloat> %0, <vscale x 2 x i8> %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.xu.v_nxv2bf16_nxv2i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf4, ta, mu
+; CHECK-NEXT:    vfwcvt.f.xu.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv2bf16.nxv2i8(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x i8> %1,
+    <vscale x 2 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv4bf16.nxv4i8(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i8>,
+  iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfwcvt_f.xu.v_nxv4bf16_nxv4i8(<vscale x 4 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.xu.v_nxv4bf16_nxv4i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, ma
+; CHECK-NEXT:    vfwcvt.f.xu.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv4bf16.nxv4i8(
+    <vscale x 4 x bfloat> poison,
+    <vscale x 4 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i8>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 4 x bfloat> @intrinsic_vfwcvt_mask_f.xu.v_nxv4bf16_nxv4i8(<vscale x 4 x bfloat> %0, <vscale x 4 x i8> %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.xu.v_nxv4bf16_nxv4i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, mf2, ta, mu
+; CHECK-NEXT:    vfwcvt.f.xu.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv4bf16.nxv4i8(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x i8> %1,
+    <vscale x 4 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv8bf16.nxv8i8(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i8>,
+  iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfwcvt_f.xu.v_nxv8bf16_nxv8i8(<vscale x 8 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.xu.v_nxv8bf16_nxv8i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, ma
+; CHECK-NEXT:    vmv1r.v v10, v8
+; CHECK-NEXT:    vfwcvt.f.xu.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv8bf16.nxv8i8(
+    <vscale x 8 x bfloat> poison,
+    <vscale x 8 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i8>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 8 x bfloat> @intrinsic_vfwcvt_mask_f.xu.v_nxv8bf16_nxv8i8(<vscale x 8 x bfloat> %0, <vscale x 8 x i8> %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.xu.v_nxv8bf16_nxv8i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m1, ta, mu
+; CHECK-NEXT:    vfwcvt.f.xu.v v8, v10, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv8bf16.nxv8i8(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x i8> %1,
+    <vscale x 8 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv16bf16.nxv16i8(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i8>,
+  iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfwcvt_f.xu.v_nxv16bf16_nxv16i8(<vscale x 16 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.xu.v_nxv16bf16_nxv16i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, ma
+; CHECK-NEXT:    vmv2r.v v12, v8
+; CHECK-NEXT:    vfwcvt.f.xu.v v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv16bf16.nxv16i8(
+    <vscale x 16 x bfloat> poison,
+    <vscale x 16 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i8>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 16 x bfloat> @intrinsic_vfwcvt_mask_f.xu.v_nxv16bf16_nxv16i8(<vscale x 16 x bfloat> %0, <vscale x 16 x i8> %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.xu.v_nxv16bf16_nxv16i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m2, ta, mu
+; CHECK-NEXT:    vfwcvt.f.xu.v v8, v12, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv16bf16.nxv16i8(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x i8> %1,
+    <vscale x 16 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv32bf16.nxv32i8(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i8>,
+  iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfwcvt_f.xu.v_nxv32bf16_nxv32i8(<vscale x 32 x i8> %0, iXLen %1) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_f.xu.v_nxv32bf16_nxv32i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, ma
+; CHECK-NEXT:    vmv4r.v v16, v8
+; CHECK-NEXT:    vfwcvt.f.xu.v v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.nxv32bf16.nxv32i8(
+    <vscale x 32 x bfloat> poison,
+    <vscale x 32 x i8> %0,
+    iXLen %1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
+declare <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8(
+  <vscale x 32 x bfloat>,
+  <vscale x 32 x i8>,
+  <vscale x 32 x i1>,
+  iXLen, iXLen);
+
+define <vscale x 32 x bfloat> @intrinsic_vfwcvt_mask_f.xu.v_nxv32bf16_nxv32i8(<vscale x 32 x bfloat> %0, <vscale x 32 x i8> %1, <vscale x 32 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwcvt_mask_f.xu.v_nxv32bf16_nxv32i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e8alt, m4, ta, mu
+; CHECK-NEXT:    vfwcvt.f.xu.v v8, v16, v0.t
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x bfloat> @llvm.riscv.vfwcvt.f.xu.v.mask.nxv32bf16.nxv32i8(
+    <vscale x 32 x bfloat> %0,
+    <vscale x 32 x i8> %1,
+    <vscale x 32 x i1> %2,
+    iXLen %3, iXLen 1)
+
+  ret <vscale x 32 x bfloat> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwmsac-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfwmsac-bf.ll
new file mode 100644
index 0000000..a3f6678
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwmsac-bf.ll
@@ -0,0 +1,506 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float>  @intrinsic_vfwmsac_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float>  @intrinsic_vfwmsac_mask_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfwmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float>  @intrinsic_vfwmsac_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float>  @intrinsic_vfwmsac_mask_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfwmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float>  @intrinsic_vfwmsac_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float>  @intrinsic_vfwmsac_mask_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfwmsac.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float>  @intrinsic_vfwmsac_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float>  @intrinsic_vfwmsac_mask_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfwmsac.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float>  @intrinsic_vfwmsac_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float>  @intrinsic_vfwmsac_mask_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfwmsac.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16(
+  <vscale x 1 x float>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float>  @intrinsic_vfwmsac_vf_nxv1f32_bf16_nxv1bf16(<vscale x 1 x float> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv1f32_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.bf16(
+    <vscale x 1 x float> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16(
+  <vscale x 1 x float>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwmsac_mask_vf_nxv1f32_bf16_nxv1bf16(<vscale x 1 x float> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv1f32_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.bf16(
+    <vscale x 1 x float> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16(
+  <vscale x 2 x float>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float>  @intrinsic_vfwmsac_vf_nxv2f32_bf16_nxv2bf16(<vscale x 2 x float> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv2f32_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.bf16(
+    <vscale x 2 x float> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16(
+  <vscale x 2 x float>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwmsac_mask_vf_nxv2f32_bf16_nxv2bf16(<vscale x 2 x float> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv2f32_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.bf16(
+    <vscale x 2 x float> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16(
+  <vscale x 4 x float>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float>  @intrinsic_vfwmsac_vf_nxv4f32_bf16_nxv4bf16(<vscale x 4 x float> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv4f32_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.bf16(
+    <vscale x 4 x float> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16(
+  <vscale x 4 x float>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwmsac_mask_vf_nxv4f32_bf16_nxv4bf16(<vscale x 4 x float> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv4f32_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.bf16(
+    <vscale x 4 x float> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16(
+  <vscale x 8 x float>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float>  @intrinsic_vfwmsac_vf_nxv8f32_bf16_nxv8bf16(<vscale x 8 x float> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv8f32_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.bf16(
+    <vscale x 8 x float> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16(
+  <vscale x 8 x float>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwmsac_mask_vf_nxv8f32_bf16_nxv8bf16(<vscale x 8 x float> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv8f32_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.bf16(
+    <vscale x 8 x float> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16(
+  <vscale x 16 x float>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float>  @intrinsic_vfwmsac_vf_nxv16f32_bf16_nxv16bf16(<vscale x 16 x float> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv16f32_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.bf16(
+    <vscale x 16 x float> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16(
+  <vscale x 16 x float>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwmsac_mask_vf_nxv16f32_bf16_nxv16bf16(<vscale x 16 x float> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv16f32_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.bf16(
+    <vscale x 16 x float> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwmul-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfwmul-bf.ll
new file mode 100644
index 0000000..577b93a
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwmul-bf.ll
@@ -0,0 +1,519 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwmul_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwmul.vv v10, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.nxv1bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwmul_mask_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_mask_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwmul.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwmul_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwmul.vv v10, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.nxv2bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwmul_mask_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_mask_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwmul.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwmul_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmv1r.v v10, v9
+; CHECK-NEXT:    vmv1r.v v11, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwmul.vv v8, v11, v10
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.nxv4bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwmul_mask_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_mask_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwmul.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwmul_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmv2r.v v12, v10
+; CHECK-NEXT:    vmv2r.v v14, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwmul.vv v8, v14, v12
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.nxv8bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwmul_mask_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_mask_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwmul.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwmul_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmv4r.v v16, v12
+; CHECK-NEXT:    vmv4r.v v20, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwmul.vv v8, v20, v16
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.nxv16bf16(
+    <vscale x 16 x float> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwmul_mask_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_mask_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwmul.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwmul_vf_nxv1f32_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_vf_nxv1f32_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwmul.vf v9, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwmul.nxv1f32.nxv1bf16.bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwmul_mask_vf_nxv1f32_nxv1bf16_bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_mask_vf_nxv1f32_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwmul.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwmul.mask.nxv1f32.nxv1bf16.bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwmul_vf_nxv2f32_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_vf_nxv2f32_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwmul.vf v9, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwmul.nxv2f32.nxv2bf16.bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwmul_mask_vf_nxv2f32_nxv2bf16_bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_mask_vf_nxv2f32_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwmul.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwmul.mask.nxv2f32.nxv2bf16.bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwmul_vf_nxv4f32_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_vf_nxv4f32_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmv1r.v v10, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwmul.vf v8, v10, fa0
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwmul.nxv4f32.nxv4bf16.bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwmul_mask_vf_nxv4f32_nxv4bf16_bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_mask_vf_nxv4f32_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwmul.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwmul.mask.nxv4f32.nxv4bf16.bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwmul_vf_nxv8f32_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_vf_nxv8f32_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmv2r.v v12, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwmul.vf v8, v12, fa0
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwmul.nxv8f32.nxv8bf16.bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwmul_mask_vf_nxv8f32_nxv8bf16_bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_mask_vf_nxv8f32_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwmul.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwmul.mask.nxv8f32.nxv8bf16.bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwmul_vf_nxv16f32_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_vf_nxv16f32_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmv4r.v v16, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwmul.vf v8, v16, fa0
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwmul.nxv16f32.nxv16bf16.bf16(
+    <vscale x 16 x float> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwmul_mask_vf_nxv16f32_nxv16bf16_bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwmul_mask_vf_nxv16f32_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwmul.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwmul.mask.nxv16f32.nxv16bf16.bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwnmacc-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfwnmacc-bf.ll
new file mode 100644
index 0000000..1e05e4c
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwnmacc-bf.ll
@@ -0,0 +1,506 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float>  @intrinsic_vfwnmacc_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float>  @intrinsic_vfwnmacc_mask_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float>  @intrinsic_vfwnmacc_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float>  @intrinsic_vfwnmacc_mask_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float>  @intrinsic_vfwnmacc_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float>  @intrinsic_vfwnmacc_mask_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfwnmacc.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float>  @intrinsic_vfwnmacc_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float>  @intrinsic_vfwnmacc_mask_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfwnmacc.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float>  @intrinsic_vfwnmacc_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float>  @intrinsic_vfwnmacc_mask_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfwnmacc.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16(
+  <vscale x 1 x float>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float>  @intrinsic_vfwnmacc_vf_nxv1f32_bf16_nxv1bf16(<vscale x 1 x float> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv1f32_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.bf16(
+    <vscale x 1 x float> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16(
+  <vscale x 1 x float>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwnmacc_mask_vf_nxv1f32_bf16_nxv1bf16(<vscale x 1 x float> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv1f32_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.bf16(
+    <vscale x 1 x float> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16(
+  <vscale x 2 x float>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float>  @intrinsic_vfwnmacc_vf_nxv2f32_bf16_nxv2bf16(<vscale x 2 x float> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv2f32_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.bf16(
+    <vscale x 2 x float> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16(
+  <vscale x 2 x float>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwnmacc_mask_vf_nxv2f32_bf16_nxv2bf16(<vscale x 2 x float> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv2f32_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.bf16(
+    <vscale x 2 x float> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16(
+  <vscale x 4 x float>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float>  @intrinsic_vfwnmacc_vf_nxv4f32_bf16_nxv4bf16(<vscale x 4 x float> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv4f32_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.bf16(
+    <vscale x 4 x float> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16(
+  <vscale x 4 x float>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwnmacc_mask_vf_nxv4f32_bf16_nxv4bf16(<vscale x 4 x float> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv4f32_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.bf16(
+    <vscale x 4 x float> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16(
+  <vscale x 8 x float>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float>  @intrinsic_vfwnmacc_vf_nxv8f32_bf16_nxv8bf16(<vscale x 8 x float> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv8f32_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.bf16(
+    <vscale x 8 x float> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16(
+  <vscale x 8 x float>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwnmacc_mask_vf_nxv8f32_bf16_nxv8bf16(<vscale x 8 x float> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv8f32_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.bf16(
+    <vscale x 8 x float> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16(
+  <vscale x 16 x float>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float>  @intrinsic_vfwnmacc_vf_nxv16f32_bf16_nxv16bf16(<vscale x 16 x float> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv16f32_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.bf16(
+    <vscale x 16 x float> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16(
+  <vscale x 16 x float>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwnmacc_mask_vf_nxv16f32_bf16_nxv16bf16(<vscale x 16 x float> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv16f32_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.bf16(
+    <vscale x 16 x float> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwnmsac-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfwnmsac-bf.ll
new file mode 100644
index 0000000..223ad4f
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwnmsac-bf.ll
@@ -0,0 +1,506 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float>  @intrinsic_vfwnmsac_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float>  @intrinsic_vfwnmsac_mask_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float>  @intrinsic_vfwnmsac_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float>  @intrinsic_vfwnmsac_mask_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float>  @intrinsic_vfwnmsac_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float>  @intrinsic_vfwnmsac_mask_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfwnmsac.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float>  @intrinsic_vfwnmsac_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float>  @intrinsic_vfwnmsac_mask_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfwnmsac.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float>  @intrinsic_vfwnmsac_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float>  @intrinsic_vfwnmsac_mask_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfwnmsac.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16(
+  <vscale x 1 x float>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float>  @intrinsic_vfwnmsac_vf_nxv1f32_bf16_nxv1bf16(<vscale x 1 x float> %0, bfloat %1, <vscale x 1 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv1f32_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.bf16(
+    <vscale x 1 x float> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16(
+  <vscale x 1 x float>,
+  bfloat,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwnmsac_mask_vf_nxv1f32_bf16_nxv1bf16(<vscale x 1 x float> %0, bfloat %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv1f32_bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, tu, mu
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.bf16(
+    <vscale x 1 x float> %0,
+    bfloat %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16(
+  <vscale x 2 x float>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float>  @intrinsic_vfwnmsac_vf_nxv2f32_bf16_nxv2bf16(<vscale x 2 x float> %0, bfloat %1, <vscale x 2 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv2f32_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.bf16(
+    <vscale x 2 x float> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16(
+  <vscale x 2 x float>,
+  bfloat,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwnmsac_mask_vf_nxv2f32_bf16_nxv2bf16(<vscale x 2 x float> %0, bfloat %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv2f32_bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, tu, mu
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.bf16(
+    <vscale x 2 x float> %0,
+    bfloat %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16(
+  <vscale x 4 x float>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float>  @intrinsic_vfwnmsac_vf_nxv4f32_bf16_nxv4bf16(<vscale x 4 x float> %0, bfloat %1, <vscale x 4 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv4f32_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.bf16(
+    <vscale x 4 x float> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16(
+  <vscale x 4 x float>,
+  bfloat,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwnmsac_mask_vf_nxv4f32_bf16_nxv4bf16(<vscale x 4 x float> %0, bfloat %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv4f32_bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, tu, mu
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.bf16(
+    <vscale x 4 x float> %0,
+    bfloat %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16(
+  <vscale x 8 x float>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float>  @intrinsic_vfwnmsac_vf_nxv8f32_bf16_nxv8bf16(<vscale x 8 x float> %0, bfloat %1, <vscale x 8 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv8f32_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.bf16(
+    <vscale x 8 x float> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16(
+  <vscale x 8 x float>,
+  bfloat,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwnmsac_mask_vf_nxv8f32_bf16_nxv8bf16(<vscale x 8 x float> %0, bfloat %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv8f32_bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, tu, mu
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.bf16(
+    <vscale x 8 x float> %0,
+    bfloat %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16(
+  <vscale x 16 x float>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float>  @intrinsic_vfwnmsac_vf_nxv16f32_bf16_nxv16bf16(<vscale x 16 x float> %0, bfloat %1, <vscale x 16 x bfloat> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv16f32_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.bf16(
+    <vscale x 16 x float> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen 0, iXLen %3, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16(
+  <vscale x 16 x float>,
+  bfloat,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwnmsac_mask_vf_nxv16f32_bf16_nxv16bf16(<vscale x 16 x float> %0, bfloat %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv16f32_bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, tu, mu
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.bf16(
+    <vscale x 16 x float> %0,
+    bfloat %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 0)
+
+  ret <vscale x 16 x float> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwsub-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfwsub-bf.ll
new file mode 100644
index 0000000..d993e4e
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwsub-bf.ll
@@ -0,0 +1,519 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwsub_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwsub.vv v10, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.nxv1bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwsub_mask_vv_nxv1f32_nxv1bf16_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_mask_vv_nxv1f32_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwsub_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.vv v10, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.nxv2bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwsub_mask_vv_nxv2f32_nxv2bf16_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_mask_vv_nxv2f32_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwsub.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwsub_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmv1r.v v10, v9
+; CHECK-NEXT:    vmv1r.v v11, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwsub.vv v8, v11, v10
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.nxv4bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwsub_mask_vv_nxv4f32_nxv4bf16_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_mask_vv_nxv4f32_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwsub.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwsub_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmv2r.v v12, v10
+; CHECK-NEXT:    vmv2r.v v14, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwsub.vv v8, v14, v12
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.nxv8bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwsub_mask_vv_nxv8f32_nxv8bf16_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_mask_vv_nxv8f32_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwsub.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwsub_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmv4r.v v16, v12
+; CHECK-NEXT:    vmv4r.v v20, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwsub.vv v8, v20, v16
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.nxv16bf16(
+    <vscale x 16 x float> poison,
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwsub_mask_vv_nxv16f32_nxv16bf16_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_mask_vv_nxv16f32_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwsub.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwsub_vf_nxv1f32_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_vf_nxv1f32_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwsub.vf v9, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.nxv1f32.nxv1bf16.bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwsub_mask_vf_nxv1f32_nxv1bf16_bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_mask_vf_nxv1f32_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwsub.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.mask.nxv1f32.nxv1bf16.bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwsub_vf_nxv2f32_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_vf_nxv2f32_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.vf v9, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.nxv2f32.nxv2bf16.bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwsub_mask_vf_nxv2f32_nxv2bf16_bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_mask_vf_nxv2f32_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwsub.vf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.mask.nxv2f32.nxv2bf16.bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwsub_vf_nxv4f32_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_vf_nxv4f32_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmv1r.v v10, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwsub.vf v8, v10, fa0
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.nxv4f32.nxv4bf16.bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwsub_mask_vf_nxv4f32_nxv4bf16_bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_mask_vf_nxv4f32_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwsub.vf v8, v10, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.mask.nxv4f32.nxv4bf16.bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwsub_vf_nxv8f32_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_vf_nxv8f32_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmv2r.v v12, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwsub.vf v8, v12, fa0
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.nxv8f32.nxv8bf16.bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwsub_mask_vf_nxv8f32_nxv8bf16_bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_mask_vf_nxv8f32_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwsub.vf v8, v12, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.mask.nxv8f32.nxv8bf16.bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwsub_vf_nxv16f32_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_vf_nxv16f32_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmv4r.v v16, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwsub.vf v8, v16, fa0
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwsub.nxv16f32.nxv16bf16.bf16(
+    <vscale x 16 x float> poison,
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwsub_mask_vf_nxv16f32_nxv16bf16_bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub_mask_vf_nxv16f32_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwsub.vf v8, v16, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwsub.mask.nxv16f32.nxv16bf16.bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwsub-w-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vfwsub-w-bf.ll
new file mode 100644
index 0000000..b22899a
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwsub-w-bf.ll
@@ -0,0 +1,773 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwsub.w_wv_nxv1f32_nxv1f32_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wv_nxv1f32_nxv1f32_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwsub.wv v8, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x float>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwsub.w_mask_wv_nxv1f32_nxv1f32_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x float> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wv_nxv1f32_nxv1f32_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwsub.wv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x float> %1,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwsub.w_wv_nxv2f32_nxv2f32_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wv_nxv2f32_nxv2f32_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.wv v8, v8, v9
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x float>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwsub.w_mask_wv_nxv2f32_nxv2f32_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x float> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wv_nxv2f32_nxv2f32_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwsub.wv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x float> %1,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwsub.w_wv_nxv4f32_nxv4f32_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wv_nxv4f32_nxv4f32_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfwsub.wv v8, v8, v10
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x float>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwsub.w_mask_wv_nxv4f32_nxv4f32_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x float> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wv_nxv4f32_nxv4f32_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwsub.wv v8, v10, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x float> %1,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwsub.w_wv_nxv8f32_nxv8f32_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wv_nxv8f32_nxv8f32_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfwsub.wv v8, v8, v12
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x float>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwsub.w_mask_wv_nxv8f32_nxv8f32_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x float> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wv_nxv8f32_nxv8f32_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwsub.wv v8, v12, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x float> %1,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwsub.w_wv_nxv16f32_nxv16f32_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wv_nxv16f32_nxv16f32_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfwsub.wv v8, v8, v16
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> poison,
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x float>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwsub.w_mask_wv_nxv16f32_nxv16f32_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x float> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wv_nxv16f32_nxv16f32_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vl4re16.v v24, (a0)
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vsetvli zero, a1, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwsub.wv v8, v16, v24, v0.t
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x float> %1,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x float>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwsub.w_wf_nxv1f32_nxv1f32_bf16(<vscale x 1 x float> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wf_nxv1f32_nxv1f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x float> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16(
+  <vscale x 1 x float>,
+  <vscale x 1 x float>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 1 x float> @intrinsic_vfwsub.w_mask_wf_nxv1f32_nxv1f32_bf16(<vscale x 1 x float> %0, <vscale x 1 x float> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wf_nxv1f32_nxv1f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwsub.wf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x float> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x float>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwsub.w_wf_nxv2f32_nxv2f32_bf16(<vscale x 2 x float> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wf_nxv2f32_nxv2f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x float> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16(
+  <vscale x 2 x float>,
+  <vscale x 2 x float>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 2 x float> @intrinsic_vfwsub.w_mask_wf_nxv2f32_nxv2f32_bf16(<vscale x 2 x float> %0, <vscale x 2 x float> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wf_nxv2f32_nxv2f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwsub.wf v8, v9, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x float> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x float>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwsub.w_wf_nxv4f32_nxv4f32_bf16(<vscale x 4 x float> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wf_nxv4f32_nxv4f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x float> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16(
+  <vscale x 4 x float>,
+  <vscale x 4 x float>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 4 x float> @intrinsic_vfwsub.w_mask_wf_nxv4f32_nxv4f32_bf16(<vscale x 4 x float> %0, <vscale x 4 x float> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wf_nxv4f32_nxv4f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwsub.wf v8, v10, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x float> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x float>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwsub.w_wf_nxv8f32_nxv8f32_bf16(<vscale x 8 x float> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wf_nxv8f32_nxv8f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x float> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16(
+  <vscale x 8 x float>,
+  <vscale x 8 x float>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 8 x float> @intrinsic_vfwsub.w_mask_wf_nxv8f32_nxv8f32_bf16(<vscale x 8 x float> %0, <vscale x 8 x float> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wf_nxv8f32_nxv8f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwsub.wf v8, v12, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x float> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x float>,
+  bfloat,
+  iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwsub.w_wf_nxv16f32_nxv16f32_bf16(<vscale x 16 x float> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wf_nxv16f32_nxv16f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.nxv16f32.bf16(
+    <vscale x 16 x float> poison,
+    <vscale x 16 x float> %0,
+    bfloat %1,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 16 x float> %a
+}
+
+declare <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16(
+  <vscale x 16 x float>,
+  <vscale x 16 x float>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen, iXLen, iXLen);
+
+define <vscale x 16 x float> @intrinsic_vfwsub.w_mask_wf_nxv16f32_nxv16f32_bf16(<vscale x 16 x float> %0, <vscale x 16 x float> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wf_nxv16f32_nxv16f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwsub.wf v8, v16, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x float> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen 0, iXLen %4, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+define <vscale x 1 x float> @intrinsic_vfwsub.w_mask_wv_tie_nxv1f32_nxv1f32_nxv1bf16(<vscale x 1 x float> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wv_tie_nxv1f32_nxv1f32_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwsub.wv v8, v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x float> %0,
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+define <vscale x 2 x float> @intrinsic_vfwsub.w_mask_wv_tie_nxv2f32_nxv2f32_nxv2bf16(<vscale x 2 x float> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wv_tie_nxv2f32_nxv2f32_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwsub.wv v8, v8, v9, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x float> %0,
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+define <vscale x 4 x float> @intrinsic_vfwsub.w_mask_wv_tie_nxv4f32_nxv4f32_nxv4bf16(<vscale x 4 x float> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wv_tie_nxv4f32_nxv4f32_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwsub.wv v8, v8, v10, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x float> %0,
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+define <vscale x 8 x float> @intrinsic_vfwsub.w_mask_wv_tie_nxv8f32_nxv8f32_nxv8bf16(<vscale x 8 x float> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wv_tie_nxv8f32_nxv8f32_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwsub.wv v8, v8, v12, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x float> %0,
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+define <vscale x 16 x float> @intrinsic_vfwsub.w_mask_wv_tie_nxv16f32_nxv16f32_nxv16bf16(<vscale x 16 x float> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wv_tie_nxv16f32_nxv16f32_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwsub.wv v8, v8, v16, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.nxv16bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x float> %0,
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+define <vscale x 1 x float> @intrinsic_vfwsub.w_mask_wf_tie_nxv1f32_nxv1f32_bf16(<vscale x 1 x float> %0, bfloat %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wf_tie_nxv1f32_nxv1f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.mask.nxv1f32.bf16(
+    <vscale x 1 x float> %0,
+    <vscale x 1 x float> %0,
+    bfloat %1,
+    <vscale x 1 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 1 x float> %a
+}
+
+define <vscale x 2 x float> @intrinsic_vfwsub.w_mask_wf_tie_nxv2f32_nxv2f32_bf16(<vscale x 2 x float> %0, bfloat %1, <vscale x 2 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wf_tie_nxv2f32_nxv2f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.mask.nxv2f32.bf16(
+    <vscale x 2 x float> %0,
+    <vscale x 2 x float> %0,
+    bfloat %1,
+    <vscale x 2 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 2 x float> %a
+}
+
+define <vscale x 4 x float> @intrinsic_vfwsub.w_mask_wf_tie_nxv4f32_nxv4f32_bf16(<vscale x 4 x float> %0, bfloat %1, <vscale x 4 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wf_tie_nxv4f32_nxv4f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.mask.nxv4f32.bf16(
+    <vscale x 4 x float> %0,
+    <vscale x 4 x float> %0,
+    bfloat %1,
+    <vscale x 4 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 4 x float> %a
+}
+
+define <vscale x 8 x float> @intrinsic_vfwsub.w_mask_wf_tie_nxv8f32_nxv8f32_bf16(<vscale x 8 x float> %0, bfloat %1, <vscale x 8 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wf_tie_nxv8f32_nxv8f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.mask.nxv8f32.bf16(
+    <vscale x 8 x float> %0,
+    <vscale x 8 x float> %0,
+    bfloat %1,
+    <vscale x 8 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 8 x float> %a
+}
+
+define <vscale x 16 x float> @intrinsic_vfwsub.w_mask_wf_tie_nxv16f32_nxv16f32_bf16(<vscale x 16 x float> %0, bfloat %1, <vscale x 16 x i1> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_mask_wf_tie_nxv16f32_nxv16f32_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0, v0.t
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x float> @llvm.riscv.vfwsub.w.mask.nxv16f32.bf16(
+    <vscale x 16 x float> %0,
+    <vscale x 16 x float> %0,
+    bfloat %1,
+    <vscale x 16 x i1> %2,
+    iXLen 0, iXLen %3, iXLen 1)
+
+  ret <vscale x 16 x float> %a
+}
+
+define <vscale x 1 x float> @intrinsic_vfwsub.w_wv_untie_nxv1f32_nxv1f32_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x float> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wv_untie_nxv1f32_nxv1f32_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vfwsub.wv v10, v9, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vfwsub.w.nxv1f32.nxv1bf16(
+    <vscale x 1 x float> poison,
+    <vscale x 1 x float> %1,
+    <vscale x 1 x bfloat> %0,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+define <vscale x 2 x float> @intrinsic_vfwsub.w_wv_untie_nxv2f32_nxv2f32_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x float> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wv_untie_nxv2f32_nxv2f32_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    fsrmi a1, 0
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.wv v10, v9, v8
+; CHECK-NEXT:    fsrm a1
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x float> @llvm.riscv.vfwsub.w.nxv2f32.nxv2bf16(
+    <vscale x 2 x float> poison,
+    <vscale x 2 x float> %1,
+    <vscale x 2 x bfloat> %0,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 2 x float> %a
+}
+
+define <vscale x 4 x float> @intrinsic_vfwsub.w_wv_untie_nxv4f32_nxv4f32_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x float> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wv_untie_nxv4f32_nxv4f32_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmv1r.v v12, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwsub.wv v8, v10, v12
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x float> @llvm.riscv.vfwsub.w.nxv4f32.nxv4bf16(
+    <vscale x 4 x float> poison,
+    <vscale x 4 x float> %1,
+    <vscale x 4 x bfloat> %0,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 4 x float> %a
+}
+
+define <vscale x 8 x float> @intrinsic_vfwsub.w_wv_untie_nxv8f32_nxv8f32_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x float> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vfwsub.w_wv_untie_nxv8f32_nxv8f32_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmv2r.v v16, v8
+; CHECK-NEXT:    fsrmi a0, 0
+; CHECK-NEXT:    vfwsub.wv v8, v12, v16
+; CHECK-NEXT:    fsrm a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x float> @llvm.riscv.vfwsub.w.nxv8f32.nxv8bf16(
+    <vscale x 8 x float> poison,
+    <vscale x 8 x float> %1,
+    <vscale x 8 x bfloat> %0,
+    iXLen 0, iXLen %2)
+
+  ret <vscale x 8 x float> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vmfeq-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vmfeq-bf.ll
new file mode 100644
index 0000000..9bd859b
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vmfeq-bf.ll
@@ -0,0 +1,496 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfeq_vv_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmfeq.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfeq_mask_vv_nxv1bf16_nxv1bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_mask_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfeq.vv v0, v8, v9
+; CHECK-NEXT:    vmfeq.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %3,
+    <vscale x 1 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfeq_vv_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmfeq.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfeq_mask_vv_nxv2bf16_nxv2bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_mask_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfeq.vv v0, v8, v9
+; CHECK-NEXT:    vmfeq.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16(
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x bfloat> %3,
+    <vscale x 2 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfeq_vv_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmfeq.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfeq_mask_vv_nxv4bf16_nxv4bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_mask_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfeq.vv v0, v8, v9
+; CHECK-NEXT:    vmfeq.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16(
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x bfloat> %3,
+    <vscale x 4 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfeq_vv_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmfeq.vv v0, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfeq_mask_vv_nxv8bf16_nxv8bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_mask_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v14, v0
+; CHECK-NEXT:    vmfeq.vv v0, v8, v10
+; CHECK-NEXT:    vmfeq.vv v14, v10, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v14
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16(
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x bfloat> %3,
+    <vscale x 8 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfeq_vv_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmfeq.vv v0, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfeq_mask_vv_nxv16bf16_nxv16bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_mask_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v20, v0
+; CHECK-NEXT:    vmfeq.vv v0, v8, v12
+; CHECK-NEXT:    vmfeq.vv v20, v12, v16, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v20
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16(
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x bfloat> %3,
+    <vscale x 16 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfeq_vf_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmfeq.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfeq.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfeq_mask_vf_nxv1bf16_bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_mask_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfeq.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfeq.mask.nxv1bf16.bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfeq_vf_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmfeq.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfeq.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfeq_mask_vf_nxv2bf16_bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_mask_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfeq.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfeq.mask.nxv2bf16.bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfeq_vf_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmfeq.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfeq.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfeq_mask_vf_nxv4bf16_bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_mask_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfeq.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfeq.mask.nxv4bf16.bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfeq_vf_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmfeq.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfeq.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfeq_mask_vf_nxv8bf16_bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_mask_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    vmfeq.vf v11, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfeq.mask.nxv8bf16.bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfeq_vf_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmfeq.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfeq.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfeq_mask_vf_nxv16bf16_bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfeq_mask_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v13, v0
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    vmfeq.vf v13, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v13
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfeq.mask.nxv16bf16.bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vmfge-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vmfge-bf.ll
new file mode 100644
index 0000000..73946dc
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vmfge-bf.ll
@@ -0,0 +1,496 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfge_vv_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmfle.vv v0, v9, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfge_mask_vv_nxv1bf16_nxv1bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_mask_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfle.vv v0, v9, v8
+; CHECK-NEXT:    vmfle.vv v11, v10, v9, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %3,
+    <vscale x 1 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfge_vv_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmfle.vv v0, v9, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfge_mask_vv_nxv2bf16_nxv2bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_mask_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfle.vv v0, v9, v8
+; CHECK-NEXT:    vmfle.vv v11, v10, v9, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16(
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x bfloat> %3,
+    <vscale x 2 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfge_vv_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmfle.vv v0, v9, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfge_mask_vv_nxv4bf16_nxv4bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_mask_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfle.vv v0, v9, v8
+; CHECK-NEXT:    vmfle.vv v11, v10, v9, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16(
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x bfloat> %3,
+    <vscale x 4 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfge_vv_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmfle.vv v0, v10, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfge_mask_vv_nxv8bf16_nxv8bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_mask_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v14, v0
+; CHECK-NEXT:    vmfle.vv v0, v10, v8
+; CHECK-NEXT:    vmfle.vv v14, v12, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v14
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16(
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x bfloat> %3,
+    <vscale x 8 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfge_vv_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmfle.vv v0, v12, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfge_mask_vv_nxv16bf16_nxv16bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_mask_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v20, v0
+; CHECK-NEXT:    vmfle.vv v0, v12, v8
+; CHECK-NEXT:    vmfle.vv v20, v16, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v20
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16(
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x bfloat> %3,
+    <vscale x 16 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfge_vf_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmfge.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfge.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfge_mask_vf_nxv1bf16_bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_mask_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfge.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfge.mask.nxv1bf16.bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfge_vf_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmfge.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfge.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfge_mask_vf_nxv2bf16_bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_mask_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfge.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfge.mask.nxv2bf16.bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfge_vf_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmfge.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfge.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfge_mask_vf_nxv4bf16_bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_mask_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfge.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfge.mask.nxv4bf16.bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfge_vf_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmfge.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfge.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfge_mask_vf_nxv8bf16_bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_mask_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    vmfge.vf v11, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfge.mask.nxv8bf16.bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfge_vf_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmfge.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfge.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfge_mask_vf_nxv16bf16_bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfge_mask_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v13, v0
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    vmfge.vf v13, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v13
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfge.mask.nxv16bf16.bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vmfgt-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vmfgt-bf.ll
new file mode 100644
index 0000000..fac324c
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vmfgt-bf.ll
@@ -0,0 +1,496 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfgt_vv_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmflt.vv v0, v9, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfgt_mask_vv_nxv1bf16_nxv1bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_mask_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmflt.vv v0, v9, v8
+; CHECK-NEXT:    vmflt.vv v11, v10, v9, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %3,
+    <vscale x 1 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfgt_vv_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmflt.vv v0, v9, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfgt_mask_vv_nxv2bf16_nxv2bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_mask_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmflt.vv v0, v9, v8
+; CHECK-NEXT:    vmflt.vv v11, v10, v9, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16(
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x bfloat> %3,
+    <vscale x 2 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfgt_vv_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmflt.vv v0, v9, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfgt_mask_vv_nxv4bf16_nxv4bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_mask_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmflt.vv v0, v9, v8
+; CHECK-NEXT:    vmflt.vv v11, v10, v9, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16(
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x bfloat> %3,
+    <vscale x 4 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfgt_vv_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmflt.vv v0, v10, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfgt_mask_vv_nxv8bf16_nxv8bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_mask_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v14, v0
+; CHECK-NEXT:    vmflt.vv v0, v10, v8
+; CHECK-NEXT:    vmflt.vv v14, v12, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v14
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16(
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x bfloat> %3,
+    <vscale x 8 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfgt_vv_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmflt.vv v0, v12, v8
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfgt_mask_vv_nxv16bf16_nxv16bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_mask_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v20, v0
+; CHECK-NEXT:    vmflt.vv v0, v12, v8
+; CHECK-NEXT:    vmflt.vv v20, v16, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v20
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16(
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x bfloat> %3,
+    <vscale x 16 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfgt_vf_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmfgt.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfgt.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfgt_mask_vf_nxv1bf16_bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_mask_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfgt.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfgt.mask.nxv1bf16.bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfgt_vf_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmfgt.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfgt.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfgt_mask_vf_nxv2bf16_bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_mask_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfgt.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfgt.mask.nxv2bf16.bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfgt_vf_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmfgt.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfgt.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfgt_mask_vf_nxv4bf16_bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_mask_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfgt.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfgt.mask.nxv4bf16.bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfgt_vf_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmfgt.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfgt.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfgt_mask_vf_nxv8bf16_bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_mask_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    vmfgt.vf v11, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfgt.mask.nxv8bf16.bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfgt_vf_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmfgt.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfgt.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfgt_mask_vf_nxv16bf16_bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfgt_mask_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v13, v0
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    vmfgt.vf v13, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v13
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfgt.mask.nxv16bf16.bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vmfle-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vmfle-bf.ll
new file mode 100644
index 0000000..8356b7b
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vmfle-bf.ll
@@ -0,0 +1,496 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfle_vv_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmfle.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfle_mask_vv_nxv1bf16_nxv1bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_mask_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfle.vv v0, v8, v9
+; CHECK-NEXT:    vmfle.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %3,
+    <vscale x 1 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfle_vv_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmfle.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfle_mask_vv_nxv2bf16_nxv2bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_mask_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfle.vv v0, v8, v9
+; CHECK-NEXT:    vmfle.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16(
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x bfloat> %3,
+    <vscale x 2 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfle_vv_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmfle.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfle_mask_vv_nxv4bf16_nxv4bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_mask_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfle.vv v0, v8, v9
+; CHECK-NEXT:    vmfle.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16(
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x bfloat> %3,
+    <vscale x 4 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfle_vv_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmfle.vv v0, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfle_mask_vv_nxv8bf16_nxv8bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_mask_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v14, v0
+; CHECK-NEXT:    vmfle.vv v0, v8, v10
+; CHECK-NEXT:    vmfle.vv v14, v10, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v14
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16(
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x bfloat> %3,
+    <vscale x 8 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfle_vv_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmfle.vv v0, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfle_mask_vv_nxv16bf16_nxv16bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_mask_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v20, v0
+; CHECK-NEXT:    vmfle.vv v0, v8, v12
+; CHECK-NEXT:    vmfle.vv v20, v12, v16, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v20
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16(
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x bfloat> %3,
+    <vscale x 16 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfle_vf_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmfle.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfle.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfle_mask_vf_nxv1bf16_bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_mask_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfle.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfle.mask.nxv1bf16.bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfle_vf_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmfle.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfle.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfle_mask_vf_nxv2bf16_bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_mask_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfle.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfle.mask.nxv2bf16.bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfle_vf_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmfle.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfle.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfle_mask_vf_nxv4bf16_bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_mask_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfle.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfle.mask.nxv4bf16.bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfle_vf_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmfle.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfle.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfle_mask_vf_nxv8bf16_bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_mask_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    vmfle.vf v11, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfle.mask.nxv8bf16.bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfle_vf_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmfle.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfle.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfle_mask_vf_nxv16bf16_bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfle_mask_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v13, v0
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    vmfle.vf v13, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v13
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfle.mask.nxv16bf16.bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vmflt-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vmflt-bf.ll
new file mode 100644
index 0000000..2e1bcc5
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vmflt-bf.ll
@@ -0,0 +1,496 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmflt_vv_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmflt.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmflt_mask_vv_nxv1bf16_nxv1bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_mask_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmflt.vv v0, v8, v9
+; CHECK-NEXT:    vmflt.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %3,
+    <vscale x 1 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmflt_vv_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmflt.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmflt_mask_vv_nxv2bf16_nxv2bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_mask_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmflt.vv v0, v8, v9
+; CHECK-NEXT:    vmflt.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16(
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x bfloat> %3,
+    <vscale x 2 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmflt_vv_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmflt.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmflt_mask_vv_nxv4bf16_nxv4bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_mask_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmflt.vv v0, v8, v9
+; CHECK-NEXT:    vmflt.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16(
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x bfloat> %3,
+    <vscale x 4 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmflt_vv_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmflt.vv v0, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmflt_mask_vv_nxv8bf16_nxv8bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_mask_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v14, v0
+; CHECK-NEXT:    vmflt.vv v0, v8, v10
+; CHECK-NEXT:    vmflt.vv v14, v10, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v14
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16(
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x bfloat> %3,
+    <vscale x 8 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmflt_vv_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmflt.vv v0, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmflt_mask_vv_nxv16bf16_nxv16bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_mask_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v20, v0
+; CHECK-NEXT:    vmflt.vv v0, v8, v12
+; CHECK-NEXT:    vmflt.vv v20, v12, v16, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v20
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16(
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x bfloat> %3,
+    <vscale x 16 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmflt_vf_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmflt.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmflt.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmflt_mask_vf_nxv1bf16_bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_mask_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmflt.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmflt.mask.nxv1bf16.bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmflt_vf_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmflt.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmflt.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmflt_mask_vf_nxv2bf16_bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_mask_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmflt.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmflt.mask.nxv2bf16.bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmflt_vf_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmflt.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmflt.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmflt_mask_vf_nxv4bf16_bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_mask_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmflt.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmflt.mask.nxv4bf16.bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmflt_vf_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmflt.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmflt.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmflt_mask_vf_nxv8bf16_bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_mask_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    vmflt.vf v11, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmflt.mask.nxv8bf16.bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmflt_vf_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmflt.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmflt.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmflt_mask_vf_nxv16bf16_bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmflt_mask_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v13, v0
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    vmflt.vf v13, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v13
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmflt.mask.nxv16bf16.bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vmfne-bf.ll b/llvm/test/CodeGen/RISCV/rvv/vmfne-bf.ll
new file mode 100644
index 0000000..283ffc5
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/vmfne-bf.ll
@@ -0,0 +1,496 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=ilp32d | FileCheck %s
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+experimental-zvfbfa \
+; RUN:   -verify-machineinstrs -target-abi=lp64d | FileCheck %s
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16(
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfne_vv_nxv1bf16_nxv1bf16(<vscale x 1 x bfloat> %0, <vscale x 1 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmfne.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16(
+    <vscale x 1 x bfloat> %0,
+    <vscale x 1 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x bfloat>,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfne_mask_vv_nxv1bf16_nxv1bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, <vscale x 1 x bfloat> %2, <vscale x 1 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_mask_vv_nxv1bf16_nxv1bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfne.vv v0, v8, v9
+; CHECK-NEXT:    vmfne.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16(
+    <vscale x 1 x bfloat> %1,
+    <vscale x 1 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %2,
+    <vscale x 1 x bfloat> %3,
+    <vscale x 1 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16(
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfne_vv_nxv2bf16_nxv2bf16(<vscale x 2 x bfloat> %0, <vscale x 2 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmfne.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16(
+    <vscale x 2 x bfloat> %0,
+    <vscale x 2 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x bfloat>,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfne_mask_vv_nxv2bf16_nxv2bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, <vscale x 2 x bfloat> %2, <vscale x 2 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_mask_vv_nxv2bf16_nxv2bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfne.vv v0, v8, v9
+; CHECK-NEXT:    vmfne.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16(
+    <vscale x 2 x bfloat> %1,
+    <vscale x 2 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %2,
+    <vscale x 2 x bfloat> %3,
+    <vscale x 2 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16(
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfne_vv_nxv4bf16_nxv4bf16(<vscale x 4 x bfloat> %0, <vscale x 4 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmfne.vv v0, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16(
+    <vscale x 4 x bfloat> %0,
+    <vscale x 4 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x bfloat>,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfne_mask_vv_nxv4bf16_nxv4bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, <vscale x 4 x bfloat> %2, <vscale x 4 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_mask_vv_nxv4bf16_nxv4bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmfne.vv v0, v8, v9
+; CHECK-NEXT:    vmfne.vv v11, v9, v10, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16(
+    <vscale x 4 x bfloat> %1,
+    <vscale x 4 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %2,
+    <vscale x 4 x bfloat> %3,
+    <vscale x 4 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16(
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfne_vv_nxv8bf16_nxv8bf16(<vscale x 8 x bfloat> %0, <vscale x 8 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmfne.vv v0, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16(
+    <vscale x 8 x bfloat> %0,
+    <vscale x 8 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x bfloat>,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfne_mask_vv_nxv8bf16_nxv8bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, <vscale x 8 x bfloat> %2, <vscale x 8 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_mask_vv_nxv8bf16_nxv8bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v14, v0
+; CHECK-NEXT:    vmfne.vv v0, v8, v10
+; CHECK-NEXT:    vmfne.vv v14, v10, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v14
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16(
+    <vscale x 8 x bfloat> %1,
+    <vscale x 8 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %2,
+    <vscale x 8 x bfloat> %3,
+    <vscale x 8 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16(
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfne_vv_nxv16bf16_nxv16bf16(<vscale x 16 x bfloat> %0, <vscale x 16 x bfloat> %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmfne.vv v0, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16(
+    <vscale x 16 x bfloat> %0,
+    <vscale x 16 x bfloat> %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x bfloat>,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfne_mask_vv_nxv16bf16_nxv16bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, <vscale x 16 x bfloat> %2, <vscale x 16 x bfloat> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_mask_vv_nxv16bf16_nxv16bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v20, v0
+; CHECK-NEXT:    vmfne.vv v0, v8, v12
+; CHECK-NEXT:    vmfne.vv v20, v12, v16, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v20
+; CHECK-NEXT:    ret
+entry:
+  %mask = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16(
+    <vscale x 16 x bfloat> %1,
+    <vscale x 16 x bfloat> %2,
+    iXLen %4)
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %2,
+    <vscale x 16 x bfloat> %3,
+    <vscale x 16 x i1> %mask,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16.bf16(
+  <vscale x 1 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfne_vf_nxv1bf16_bf16(<vscale x 1 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, ma
+; CHECK-NEXT:    vmfne.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfne.nxv1bf16.bf16(
+    <vscale x 1 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.bf16(
+  <vscale x 1 x i1>,
+  <vscale x 1 x bfloat>,
+  bfloat,
+  <vscale x 1 x i1>,
+  iXLen);
+
+define <vscale x 1 x i1> @intrinsic_vmfne_mask_vf_nxv1bf16_bf16(<vscale x 1 x i1> %0, <vscale x 1 x bfloat> %1, bfloat %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_mask_vf_nxv1bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf4, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfne.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i1> @llvm.riscv.vmfne.mask.nxv1bf16.bf16(
+    <vscale x 1 x i1> %0,
+    <vscale x 1 x bfloat> %1,
+    bfloat %2,
+    <vscale x 1 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 1 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16.bf16(
+  <vscale x 2 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfne_vf_nxv2bf16_bf16(<vscale x 2 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, ma
+; CHECK-NEXT:    vmfne.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfne.nxv2bf16.bf16(
+    <vscale x 2 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.bf16(
+  <vscale x 2 x i1>,
+  <vscale x 2 x bfloat>,
+  bfloat,
+  <vscale x 2 x i1>,
+  iXLen);
+
+define <vscale x 2 x i1> @intrinsic_vmfne_mask_vf_nxv2bf16_bf16(<vscale x 2 x i1> %0, <vscale x 2 x bfloat> %1, bfloat %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_mask_vf_nxv2bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, mf2, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfne.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i1> @llvm.riscv.vmfne.mask.nxv2bf16.bf16(
+    <vscale x 2 x i1> %0,
+    <vscale x 2 x bfloat> %1,
+    bfloat %2,
+    <vscale x 2 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 2 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16.bf16(
+  <vscale x 4 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfne_vf_nxv4bf16_bf16(<vscale x 4 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, ma
+; CHECK-NEXT:    vmfne.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfne.nxv4bf16.bf16(
+    <vscale x 4 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.bf16(
+  <vscale x 4 x i1>,
+  <vscale x 4 x bfloat>,
+  bfloat,
+  <vscale x 4 x i1>,
+  iXLen);
+
+define <vscale x 4 x i1> @intrinsic_vmfne_mask_vf_nxv4bf16_bf16(<vscale x 4 x i1> %0, <vscale x 4 x bfloat> %1, bfloat %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_mask_vf_nxv4bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m1, ta, mu
+; CHECK-NEXT:    vmv1r.v v10, v0
+; CHECK-NEXT:    vmv1r.v v0, v9
+; CHECK-NEXT:    vmfne.vf v10, v8, fa0, v0.t
+; CHECK-NEXT:    vmv.v.v v0, v10
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i1> @llvm.riscv.vmfne.mask.nxv4bf16.bf16(
+    <vscale x 4 x i1> %0,
+    <vscale x 4 x bfloat> %1,
+    bfloat %2,
+    <vscale x 4 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 4 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16.bf16(
+  <vscale x 8 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfne_vf_nxv8bf16_bf16(<vscale x 8 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, ma
+; CHECK-NEXT:    vmfne.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfne.nxv8bf16.bf16(
+    <vscale x 8 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.bf16(
+  <vscale x 8 x i1>,
+  <vscale x 8 x bfloat>,
+  bfloat,
+  <vscale x 8 x i1>,
+  iXLen);
+
+define <vscale x 8 x i1> @intrinsic_vmfne_mask_vf_nxv8bf16_bf16(<vscale x 8 x i1> %0, <vscale x 8 x bfloat> %1, bfloat %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_mask_vf_nxv8bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m2, ta, mu
+; CHECK-NEXT:    vmv1r.v v11, v0
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    vmfne.vf v11, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v11
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i1> @llvm.riscv.vmfne.mask.nxv8bf16.bf16(
+    <vscale x 8 x i1> %0,
+    <vscale x 8 x bfloat> %1,
+    bfloat %2,
+    <vscale x 8 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 8 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16.bf16(
+  <vscale x 16 x bfloat>,
+  bfloat,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfne_vf_nxv16bf16_bf16(<vscale x 16 x bfloat> %0, bfloat %1, iXLen %2) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, ma
+; CHECK-NEXT:    vmfne.vf v0, v8, fa0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfne.nxv16bf16.bf16(
+    <vscale x 16 x bfloat> %0,
+    bfloat %1,
+    iXLen %2)
+
+  ret <vscale x 16 x i1> %a
+}
+
+declare <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.bf16(
+  <vscale x 16 x i1>,
+  <vscale x 16 x bfloat>,
+  bfloat,
+  <vscale x 16 x i1>,
+  iXLen);
+
+define <vscale x 16 x i1> @intrinsic_vmfne_mask_vf_nxv16bf16_bf16(<vscale x 16 x i1> %0, <vscale x 16 x bfloat> %1, bfloat %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
+; CHECK-LABEL: intrinsic_vmfne_mask_vf_nxv16bf16_bf16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, a0, e16alt, m4, ta, mu
+; CHECK-NEXT:    vmv1r.v v13, v0
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    vmfne.vf v13, v8, fa0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v13
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i1> @llvm.riscv.vmfne.mask.nxv16bf16.bf16(
+    <vscale x 16 x i1> %0,
+    <vscale x 16 x bfloat> %1,
+    bfloat %2,
+    <vscale x 16 x i1> %3,
+    iXLen %4)
+
+  ret <vscale x 16 x i1> %a
+}
+
diff --git a/llvm/test/Transforms/InstCombine/scmp.ll b/llvm/test/Transforms/InstCombine/scmp.ll
index 2bf22ae..c0be5b9 100644
--- a/llvm/test/Transforms/InstCombine/scmp.ll
+++ b/llvm/test/Transforms/InstCombine/scmp.ll
@@ -423,6 +423,86 @@ define i8 @scmp_from_select_eq_and_gt_commuted3(i32 %x, i32 %y) {
   ret i8 %r
 }
 
+; Commutative tests for (x != y) ? (x > y ? 1 : -1) : 0
+define i8 @scmp_from_select_ne_and_gt_commuted1(i32 %x, i32 %y) {
+; CHECK-LABEL: define i8 @scmp_from_select_ne_and_gt_commuted1(
+; CHECK-SAME: i32 [[X:%.*]], i32 [[Y:%.*]]) {
+; CHECK-NEXT:    [[R:%.*]] = call i8 @llvm.scmp.i8.i32(i32 [[Y]], i32 [[X]])
+; CHECK-NEXT:    ret i8 [[R]]
+;
+  %ne = icmp ne i32 %x, %y
+  %gt = icmp slt i32 %x, %y
+  %sel1 = select i1 %gt, i8 1, i8 -1
+  %r = select i1 %ne, i8 %sel1, i8 0
+  ret i8 %r
+}
+
+define i8 @scmp_from_select_ne_and_gt_commuted2(i32 %x, i32 %y) {
+; CHECK-LABEL: define i8 @scmp_from_select_ne_and_gt_commuted2(
+; CHECK-SAME: i32 [[X:%.*]], i32 [[Y:%.*]]) {
+; CHECK-NEXT:    [[R:%.*]] = call i8 @llvm.scmp.i8.i32(i32 [[Y]], i32 [[X]])
+; CHECK-NEXT:    ret i8 [[R]]
+;
+  %ne = icmp ne i32 %x, %y
+  %gt = icmp sgt i32 %x, %y
+  %sel1 = select i1 %gt, i8 -1, i8 1
+  %r = select i1 %ne, i8 %sel1, i8 0
+  ret i8 %r
+}
+
+define i8 @scmp_from_select_ne_and_gt_commuted3(i32 %x, i32 %y) {
+; CHECK-LABEL: define i8 @scmp_from_select_ne_and_gt_commuted3(
+; CHECK-SAME: i32 [[X:%.*]], i32 [[Y:%.*]]) {
+; CHECK-NEXT:    [[R:%.*]] = call i8 @llvm.scmp.i8.i32(i32 [[X]], i32 [[Y]])
+; CHECK-NEXT:    ret i8 [[R]]
+;
+  %ne = icmp ne i32 %x, %y
+  %gt = icmp sgt i32 %x, %y
+  %sel1 = select i1 %gt, i8 1, i8 -1
+  %r = select i1 %ne, i8 %sel1, i8 0
+  ret i8 %r
+}
+
+; Commutative tests for x != C ? (x > C - 1 ? 1 : -1) : 0
+define i8 @scmp_from_select_ne_const_and_gt_commuted1(i32 %x) {
+; CHECK-LABEL: define i8 @scmp_from_select_ne_const_and_gt_commuted1(
+; CHECK-SAME: i32 [[X:%.*]]) {
+; CHECK-NEXT:    [[R:%.*]] = call i8 @llvm.scmp.i8.i32(i32 [[X]], i32 5)
+; CHECK-NEXT:    ret i8 [[R]]
+;
+  %ne = icmp ne i32 %x, 5
+  %gt = icmp sgt i32 %x, 4
+  %sel1 = select i1 %gt, i8 1, i8 -1
+  %r = select i1 %ne, i8 %sel1, i8 0
+  ret i8 %r
+}
+
+define i8 @scmp_from_select_ne_const_and_gt_commuted2(i32 %x) {
+; CHECK-LABEL: define i8 @scmp_from_select_ne_const_and_gt_commuted2(
+; CHECK-SAME: i32 [[X:%.*]]) {
+; CHECK-NEXT:    [[R:%.*]] = call i8 @llvm.scmp.i8.i32(i32 [[X]], i32 5)
+; CHECK-NEXT:    ret i8 [[R]]
+;
+  %ne = icmp ne i32 %x, 5
+  %gt = icmp sgt i32 %x, 4
+  %sel1 = select i1 %gt, i8 1, i8 -1
+  %r = select i1 %ne, i8 %sel1, i8 0
+  ret i8 %r
+}
+
+define i8 @scmp_from_select_ne_const_and_gt_commuted3(i32 %x) {
+; CHECK-LABEL: define i8 @scmp_from_select_ne_const_and_gt_commuted3(
+; CHECK-SAME: i32 [[X:%.*]]) {
+; CHECK-NEXT:    [[R:%.*]] = call i8 @llvm.scmp.i8.i32(i32 [[X]], i32 5)
+; CHECK-NEXT:    ret i8 [[R]]
+;
+  %ne = icmp ne i32 %x, 5
+  %gt = icmp sgt i32 %x, 4
+  %sel1 = select i1 %gt, i8 1, i8 -1
+  %r = select i1 %ne, i8 %sel1, i8 0
+  ret i8 %r
+}
+
 define <3 x i2> @scmp_unary_shuffle_ops(<3 x i8> %x, <3 x i8> %y) {
 ; CHECK-LABEL: define <3 x i2> @scmp_unary_shuffle_ops(
 ; CHECK-SAME: <3 x i8> [[X:%.*]], <3 x i8> [[Y:%.*]]) {
@@ -436,6 +516,187 @@ define <3 x i2> @scmp_unary_shuffle_ops(<3 x i8> %x, <3 x i8> %y) {
   ret <3 x i2> %r
 }
 
+define i32 @scmp_sgt_slt(i32 %a) {
+; CHECK-LABEL: define i32 @scmp_sgt_slt(
+; CHECK-SAME: i32 [[A:%.*]]) {
+; CHECK-NEXT:    [[A_LOBIT:%.*]] = ashr i32 [[A]], 31
+; CHECK-NEXT:    [[CMP_INV:%.*]] = icmp slt i32 [[A]], 1
+; CHECK-NEXT:    [[RETVAL_0:%.*]] = select i1 [[CMP_INV]], i32 [[A_LOBIT]], i32 1
+; CHECK-NEXT:    ret i32 [[RETVAL_0]]
+;
+  %cmp = icmp sgt i32 %a, 0
+  %cmp1 = icmp slt i32 %a, 0
+  %. = select i1 %cmp1, i32 -1, i32 0
+  %retval.0 = select i1 %cmp, i32 1, i32 %.
+  ret i32 %retval.0
+}
+
+define i32 @scmp_zero_slt(i32 %a) {
+; CHECK-LABEL: define i32 @scmp_zero_slt(
+; CHECK-SAME: i32 [[A:%.*]]) {
+; CHECK-NEXT:    [[RETVAL_0:%.*]] = call i32 @llvm.scmp.i32.i32(i32 [[A]], i32 0)
+; CHECK-NEXT:    ret i32 [[RETVAL_0]]
+;
+  %cmp = icmp eq i32 %a, 0
+  %cmp1.inv = icmp slt i32 %a, 1
+  %. = select i1 %cmp1.inv, i32 -1, i32 1
+  %retval.0 = select i1 %cmp, i32 0, i32 %.
+  ret i32 %retval.0
+}
+
+define i32 @scmp_zero_sgt(i32 %a) {
+; CHECK-LABEL: define i32 @scmp_zero_sgt(
+; CHECK-SAME: i32 [[A:%.*]]) {
+; CHECK-NEXT:    [[RETVAL_0:%.*]] = call i32 @llvm.scmp.i32.i32(i32 [[A]], i32 0)
+; CHECK-NEXT:    ret i32 [[RETVAL_0]]
+;
+  %cmp = icmp eq i32 %a, 0
+  %cmp1.inv = icmp sgt i32 %a, -1
+  %. = select i1 %cmp1.inv, i32 1, i32 -1
+  %retval.0 = select i1 %cmp, i32 0, i32 %.
+  ret i32 %retval.0
+}
+
+
+define i32 @scmp_zero_sgt_1(i32 %a) {
+; CHECK-LABEL: define i32 @scmp_zero_sgt_1(
+; CHECK-SAME: i32 [[A:%.*]]) {
+; CHECK-NEXT:    [[COND2:%.*]] = call i32 @llvm.scmp.i32.i32(i32 [[A]], i32 0)
+; CHECK-NEXT:    ret i32 [[COND2]]
+;
+  %cmp = icmp eq i32 %a, 0
+  %cmp1 = icmp sgt i32 %a, -1
+  %cond = select i1 %cmp1, i32 1, i32 -1
+  %cond2 = select i1 %cmp, i32 0, i32 %cond
+  ret i32 %cond2
+}
+
+define i32 @scmp_zero_slt_1(i32 %a) {
+; CHECK-LABEL: define i32 @scmp_zero_slt_1(
+; CHECK-SAME: i32 [[A:%.*]]) {
+; CHECK-NEXT:    [[COND2:%.*]] = call i32 @llvm.scmp.i32.i32(i32 [[A]], i32 0)
+; CHECK-NEXT:    ret i32 [[COND2]]
+;
+  %cmp = icmp eq i32 %a, 0
+  %cmp1 = icmp slt i32 %a, 1
+  %cond = select i1 %cmp1, i32 -1, i32 1
+  %cond2 = select i1 %cmp, i32 0, i32 %cond
+  ret i32 %cond2
+}
+
+define i32 @scmp_zero_slt_neg(i32 %a) {
+; CHECK-LABEL: define i32 @scmp_zero_slt_neg(
+; CHECK-SAME: i32 [[A:%.*]]) {
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[A]], 0
+; CHECK-NEXT:    [[CMP1:%.*]] = icmp slt i32 [[A]], -1
+; CHECK-NEXT:    [[COND:%.*]] = select i1 [[CMP1]], i32 -1, i32 1
+; CHECK-NEXT:    [[COND2:%.*]] = select i1 [[CMP]], i32 0, i32 [[COND]]
+; CHECK-NEXT:    ret i32 [[COND2]]
+;
+  %cmp = icmp eq i32 %a, 0
+  %cmp1 = icmp slt i32 %a, -1
+  %cond = select i1 %cmp1, i32 -1, i32 1
+  %cond2 = select i1 %cmp, i32 0, i32 %cond
+  ret i32 %cond2
+}
+
+define i32 @scmp_zero_sgt_neg(i32 %a) {
+; CHECK-LABEL: define i32 @scmp_zero_sgt_neg(
+; CHECK-SAME: i32 [[A:%.*]]) {
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[A]], 0
+; CHECK-NEXT:    [[CMP1:%.*]] = icmp sgt i32 [[A]], 1
+; CHECK-NEXT:    [[COND:%.*]] = select i1 [[CMP1]], i32 1, i32 -1
+; CHECK-NEXT:    [[COND2:%.*]] = select i1 [[CMP]], i32 0, i32 [[COND]]
+; CHECK-NEXT:    ret i32 [[COND2]]
+;
+  %cmp = icmp eq i32 %a, 0
+  %cmp1 = icmp sgt i32 %a, 1
+  %cond = select i1 %cmp1, i32 1, i32 -1
+  %cond2 = select i1 %cmp, i32 0, i32 %cond
+  ret i32 %cond2
+}
+
+define i32 @ucmp_ugt_ult_neg(i32 %a) {
+; CHECK-LABEL: define i32 @ucmp_ugt_ult_neg(
+; CHECK-SAME: i32 [[A:%.*]]) {
+; CHECK-NEXT:    [[CMP_NOT:%.*]] = icmp ne i32 [[A]], 0
+; CHECK-NEXT:    [[RETVAL_0:%.*]] = zext i1 [[CMP_NOT]] to i32
+; CHECK-NEXT:    ret i32 [[RETVAL_0]]
+;
+  %cmp = icmp ugt i32 %a, 0
+  %cmp1 = icmp ult i32 %a, 0
+  %. = select i1 %cmp1, i32 -1, i32 0
+  %retval.0 = select i1 %cmp, i32 1, i32 %.
+  ret i32 %retval.0
+}
+
+define i32 @ucmp_zero_ult_neg(i32 %a) {
+; CHECK-LABEL: define i32 @ucmp_zero_ult_neg(
+; CHECK-SAME: i32 [[A:%.*]]) {
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i32 [[A]], 0
+; CHECK-NEXT:    [[RETVAL_0:%.*]] = zext i1 [[CMP]] to i32
+; CHECK-NEXT:    ret i32 [[RETVAL_0]]
+;
+  %cmp = icmp eq i32 %a, 0
+  %cmp1.inv = icmp ult i32 %a, 1
+  %. = select i1 %cmp1.inv, i32 -1, i32 1
+  %retval.0 = select i1 %cmp, i32 0, i32 %.
+  ret i32 %retval.0
+}
+
+define i32 @ucmp_zero_ugt_neg(i32 %a) {
+; CHECK-LABEL: define i32 @ucmp_zero_ugt_neg(
+; CHECK-SAME: i32 [[A:%.*]]) {
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i32 [[A]], 0
+; CHECK-NEXT:    [[RETVAL_0:%.*]] = sext i1 [[CMP]] to i32
+; CHECK-NEXT:    ret i32 [[RETVAL_0]]
+;
+  %cmp = icmp eq i32 %a, 0
+  %cmp1.inv = icmp ugt i32 %a, -1
+  %. = select i1 %cmp1.inv, i32 1, i32 -1
+  %retval.0 = select i1 %cmp, i32 0, i32 %.
+  ret i32 %retval.0
+}
+
+define i32 @scmp_sgt_slt_ab(i32 %a, i32 %b) {
+; CHECK-LABEL: define i32 @scmp_sgt_slt_ab(
+; CHECK-SAME: i32 [[A:%.*]], i32 [[B:%.*]]) {
+; CHECK-NEXT:    [[RETVAL_0:%.*]] = call i32 @llvm.scmp.i32.i32(i32 [[A]], i32 [[B]])
+; CHECK-NEXT:    ret i32 [[RETVAL_0]]
+;
+  %cmp = icmp sgt i32 %a, %b
+  %cmp1 = icmp slt i32 %a, %b
+  %. = select i1 %cmp1, i32 -1, i32 0
+  %retval.0 = select i1 %cmp, i32 1, i32 %.
+  ret i32 %retval.0
+}
+
+define i32 @scmp_zero_slt_ab(i32 %a, i32 %b) {
+; CHECK-LABEL: define i32 @scmp_zero_slt_ab(
+; CHECK-SAME: i32 [[A:%.*]], i32 [[B:%.*]]) {
+; CHECK-NEXT:    [[RETVAL_0:%.*]] = call i32 @llvm.scmp.i32.i32(i32 [[A]], i32 [[B]])
+; CHECK-NEXT:    ret i32 [[RETVAL_0]]
+;
+  %cmp = icmp eq i32 %a, %b
+  %cmp1.inv = icmp slt i32 %a, %b
+  %. = select i1 %cmp1.inv, i32 -1, i32 1
+  %retval.0 = select i1 %cmp, i32 0, i32 %.
+  ret i32 %retval.0
+}
+
+define i32 @scmp_zero_sgt_ab(i32 %a, i32 %b) {
+; CHECK-LABEL: define i32 @scmp_zero_sgt_ab(
+; CHECK-SAME: i32 [[A:%.*]], i32 [[B:%.*]]) {
+; CHECK-NEXT:    [[RETVAL_0:%.*]] = call i32 @llvm.scmp.i32.i32(i32 [[A]], i32 [[B]])
+; CHECK-NEXT:    ret i32 [[RETVAL_0]]
+;
+  %cmp = icmp eq i32 %a, %b
+  %cmp1.inv = icmp sgt i32 %a, %b
+  %. = select i1 %cmp1.inv, i32 1, i32 -1
+  %retval.0 = select i1 %cmp, i32 0, i32 %.
+  ret i32 %retval.0
+}
+
 ; Negative test: true value of outer select is not zero
 define i8 @scmp_from_select_eq_and_gt_neg1(i32 %x, i32 %y) {
 ; CHECK-LABEL: define i8 @scmp_from_select_eq_and_gt_neg1(
diff --git a/llvm/test/Transforms/SCCP/constant-range-struct.ll b/llvm/test/Transforms/SCCP/constant-range-struct.ll
index 7a399df..0f45b38 100644
--- a/llvm/test/Transforms/SCCP/constant-range-struct.ll
+++ b/llvm/test/Transforms/SCCP/constant-range-struct.ll
@@ -25,7 +25,7 @@ true:
   br label %exit
 
 false:
-  %s.3 = insertvalue {i64, i64} undef, i64 30, 0
+  %s.3 = insertvalue {i64, i64} poison, i64 30, 0
   %s.4 = insertvalue {i64, i64} %s.3, i64 300, 1
   br label %exit
 
@@ -39,14 +39,14 @@ define void @struct1_caller() {
 ; CHECK-NEXT:    [[S:%.*]] = call { i64, i64 } @struct1()
 ; CHECK-NEXT:    [[V1:%.*]] = extractvalue { i64, i64 } [[S]], 0
 ; CHECK-NEXT:    [[V2:%.*]] = extractvalue { i64, i64 } [[S]], 1
-; CHECK-NEXT:    [[T_1:%.*]] = icmp ne i64 [[V1]], 10
-; CHECK-NEXT:    call void @use(i1 [[T_1]])
-; CHECK-NEXT:    [[T_2:%.*]] = icmp ult i64 [[V1]], 100
-; CHECK-NEXT:    call void @use(i1 [[T_2]])
-; CHECK-NEXT:    [[T_3:%.*]] = icmp ne i64 [[V2]], 0
+; CHECK-NEXT:    call void @use(i1 true)
+; CHECK-NEXT:    call void @use(i1 true)
+; CHECK-NEXT:    [[T_3:%.*]] = icmp eq i64 [[V1]], 20
 ; CHECK-NEXT:    call void @use(i1 [[T_3]])
-; CHECK-NEXT:    [[T_4:%.*]] = icmp ult i64 [[V2]], 301
-; CHECK-NEXT:    call void @use(i1 [[T_4]])
+; CHECK-NEXT:    call void @use(i1 true)
+; CHECK-NEXT:    call void @use(i1 true)
+; CHECK-NEXT:    [[T_6:%.*]] = icmp eq i64 [[V2]], 300
+; CHECK-NEXT:    call void @use(i1 [[T_6]])
 ; CHECK-NEXT:    ret void
 ;
   %s = call {i64, i64} @struct1()
@@ -57,10 +57,14 @@ define void @struct1_caller() {
   call void @use(i1 %t.1)
   %t.2 = icmp ult i64 %v1, 100
   call void @use(i1 %t.2)
-  %t.3 = icmp ne i64 %v2, 0
+  %t.3 = icmp eq i64 %v1, 20
   call void @use(i1 %t.3)
-  %t.4 = icmp ult i64 %v2, 301
+  %t.4 = icmp ne i64 %v2, 0
   call void @use(i1 %t.4)
+  %t.5 = icmp ult i64 %v2, 301
+  call void @use(i1 %t.5)
+  %t.6 = icmp eq i64 %v2, 300
+  call void @use(i1 %t.6)
 
   ret void
 }
@@ -76,7 +80,7 @@ define internal {i64, i64} @struct2() {
 ; CHECK:       exit:
 ; CHECK-NEXT:    [[V1:%.*]] = phi i64 [ 20, [[TRUE]] ], [ 30, [[FALSE]] ]
 ; CHECK-NEXT:    [[V2:%.*]] = phi i64 [ 200, [[TRUE]] ], [ 300, [[FALSE]] ]
-; CHECK-NEXT:    [[S_1:%.*]] = insertvalue { i64, i64 } undef, i64 [[V1]], 0
+; CHECK-NEXT:    [[S_1:%.*]] = insertvalue { i64, i64 } poison, i64 [[V1]], 0
 ; CHECK-NEXT:    [[S_2:%.*]] = insertvalue { i64, i64 } [[S_1]], i64 [[V2]], 1
 ; CHECK-NEXT:    ret { i64, i64 } [[S_2]]
 ;
@@ -92,7 +96,7 @@ false:
 exit:
   %v1 = phi i64 [ 20, %true ], [ 30, %false ]
   %v2 = phi i64 [ 200, %true ], [ 300, %false ]
-  %s.1 = insertvalue {i64, i64} undef, i64 %v1, 0
+  %s.1 = insertvalue {i64, i64} poison, i64 %v1, 0
   %s.2 = insertvalue {i64, i64} %s.1, i64 %v2, 1
   ret {i64, i64} %s.2
 }
@@ -153,3 +157,40 @@ define void @struct2_caller() {
 
   ret void
 }
+
+%"phi_type" =  type {i64, i64}
+
+define internal %"phi_type" @test(i32 %input) {
+; CHECK-LABEL: @test(
+; CHECK-NEXT:    br label [[COND_TRUE_I:%.*]]
+; CHECK:       cond.true.i:
+; CHECK-NEXT:    br label [[COND_END_I:%.*]]
+; CHECK:       cond.end.i:
+; CHECK-NEXT:    ret [[PHI_TYPE:%.*]] poison
+;
+  %cmp.cond = icmp eq i32 %input, 1
+  br i1 %cmp.cond, label %cond.true.i, label %cond.false.i
+
+cond.true.i:
+  %r1.tmp = insertvalue %"phi_type" poison, i64 1, 0
+  %r1.tmp.2 = insertvalue %"phi_type" %r1.tmp, i64 2, 1
+  br label %cond.end.i
+
+cond.false.i:
+  %r2.tmp = insertvalue %"phi_type" poison, i64 3, 0
+  %r2.tmp.2 = insertvalue %"phi_type" %r2.tmp, i64 4, 1
+  br label %cond.end.i
+
+cond.end.i:
+  %retval = phi %"phi_type" [ %r1.tmp.2, %cond.true.i ], [ %r2.tmp.2, %cond.false.i ]
+  ret %"phi_type" %retval
+}
+
+define %"phi_type" @test2() {
+; CHECK-LABEL: @test2(
+; CHECK-NEXT:    [[CALL_1:%.*]] = tail call fastcc [[PHI_TYPE:%.*]] @[[TEST:[a-zA-Z0-9_$\"\\.-]*[a-zA-Z_$\"\\.-][a-zA-Z0-9_$\"\\.-]*]](i32 noundef 1)
+; CHECK-NEXT:    ret [[PHI_TYPE]] { i64 1, i64 2 }
+;
+  %call.1 = tail call fastcc noundef %"phi_type" @test(i32 noundef 1)
+  ret %"phi_type" %call.1
+}
diff --git a/llvm/unittests/ADT/BitTest.cpp b/llvm/unittests/ADT/BitTest.cpp
index eaed4e1..5b3df91 100644
--- a/llvm/unittests/ADT/BitTest.cpp
+++ b/llvm/unittests/ADT/BitTest.cpp
@@ -270,6 +270,22 @@ TEST(BitTest, BitWidthConstexpr) {
       llvm::bit_width_constexpr(std::numeric_limits<uint64_t>::max()) == 64);
 }
 
+TEST(BitTest, BitCeilConstexpr) {
+  static_assert(llvm::bit_ceil_constexpr(0u) == 1);
+  static_assert(llvm::bit_ceil_constexpr(1u) == 1);
+  static_assert(llvm::bit_ceil_constexpr(2u) == 2);
+  static_assert(llvm::bit_ceil_constexpr(3u) == 4);
+  static_assert(llvm::bit_ceil_constexpr(4u) == 4);
+  static_assert(llvm::bit_ceil_constexpr(5u) == 8);
+  static_assert(llvm::bit_ceil_constexpr(6u) == 8);
+  static_assert(llvm::bit_ceil_constexpr(7u) == 8);
+  static_assert(llvm::bit_ceil_constexpr(8u) == 8);
+
+  static_assert(llvm::bit_ceil_constexpr(255u) == 256);
+  static_assert(llvm::bit_ceil_constexpr(256u) == 256);
+  static_assert(llvm::bit_ceil_constexpr(257u) == 512);
+}
+
 TEST(BitTest, CountlZero) {
   uint8_t Z8 = 0;
   uint16_t Z16 = 0;
diff --git a/mlir/include/mlir/Dialect/SMT/IR/SMTOps.td b/mlir/include/mlir/Dialect/SMT/IR/SMTOps.td
index 3143ab7..99b22e5 100644
--- a/mlir/include/mlir/Dialect/SMT/IR/SMTOps.td
+++ b/mlir/include/mlir/Dialect/SMT/IR/SMTOps.td
@@ -220,8 +220,6 @@ def YieldOp : SMTOp<"yield", [
   Pure,
   Terminator,
   ReturnLike,
-  ParentOneOf<["smt::SolverOp", "smt::CheckOp",
-               "smt::ForallOp", "smt::ExistsOp"]>,
 ]> {
   let summary = "terminator operation for various regions of SMT operations";
   let arguments = (ins Variadic<AnyType>:$values);
diff --git a/mlir/include/mlir/Dialect/Transform/SMTExtension/SMTExtensionOps.h b/mlir/include/mlir/Dialect/Transform/SMTExtension/SMTExtensionOps.h
index fc69b03..f6353a9 100644
--- a/mlir/include/mlir/Dialect/Transform/SMTExtension/SMTExtensionOps.h
+++ b/mlir/include/mlir/Dialect/Transform/SMTExtension/SMTExtensionOps.h
@@ -10,6 +10,7 @@
 #define MLIR_DIALECT_TRANSFORM_SMTEXTENSION_SMTEXTENSIONOPS_H
 
 #include "mlir/Bytecode/BytecodeOpInterface.h"
+#include "mlir/Dialect/SMT/IR/SMTOps.h"
 #include "mlir/Dialect/Transform/IR/TransformDialect.h"
 #include "mlir/Dialect/Transform/Interfaces/TransformInterfaces.h"
 #include "mlir/IR/OpDefinition.h"
diff --git a/mlir/include/mlir/Dialect/Transform/SMTExtension/SMTExtensionOps.td b/mlir/include/mlir/Dialect/Transform/SMTExtension/SMTExtensionOps.td
index b987cb3..9d9783a 100644
--- a/mlir/include/mlir/Dialect/Transform/SMTExtension/SMTExtensionOps.td
+++ b/mlir/include/mlir/Dialect/Transform/SMTExtension/SMTExtensionOps.td
@@ -16,7 +16,7 @@ include "mlir/Interfaces/SideEffectInterfaces.td"
 def ConstrainParamsOp : Op<Transform_Dialect, "smt.constrain_params", [
   DeclareOpInterfaceMethods<TransformOpInterface>,
   DeclareOpInterfaceMethods<MemoryEffectsOpInterface>,
-  NoTerminator
+  SingleBlockImplicitTerminator<"::mlir::smt::YieldOp">
 ]> {
   let cppNamespace = [{ mlir::transform::smt }];
 
@@ -24,14 +24,20 @@ def ConstrainParamsOp : Op<Transform_Dialect, "smt.constrain_params", [
   let description = [{
     Allows expressing constraints on params using the SMT dialect.
 
-    Each Transform dialect param provided as an operand has a corresponding
+    Each Transform-dialect param provided as an operand has a corresponding
     argument of SMT-type in the region. The SMT-Dialect ops in the region use
-    these arguments as operands.
+    these params-as-SMT-vars as operands, thereby expressing relevant
+    constraints on their allowed values.
+
+    Computations w.r.t. passed-in params can also be expressed through the
+    region's SMT-ops. Namely, the constraints express relationships to other
+    SMT-variables which can then be yielded from the region (with `smt.yield`).
 
     The semantics of this op is that all the ops in the region together express
     a constraint on the params-interpreted-as-smt-vars. The op fails in case the
     expressed constraint is not satisfiable per SMTLIB semantics. Otherwise the
-    op succeeds.
+    op succeeds and any one satisfying assignment is used to map the
+    SMT-variables yielded in the region to `transform.param`s.
 
     ---
 
@@ -42,9 +48,10 @@ def ConstrainParamsOp : Op<Transform_Dialect, "smt.constrain_params", [
   }];
 
   let arguments = (ins Variadic<TransformParamTypeInterface>:$params);
+  let results = (outs Variadic<TransformParamTypeInterface>:$results);
   let regions = (region SizedRegion<1>:$body);
   let assemblyFormat =
-      "`(` $params `)` attr-dict `:` type(operands) $body";
+      "`(` $params `)` attr-dict `:` functional-type(operands, results) $body";
 
   let hasVerifier = 1;
 }
diff --git a/mlir/lib/Dialect/Transform/SMTExtension/SMTExtensionOps.cpp b/mlir/lib/Dialect/Transform/SMTExtension/SMTExtensionOps.cpp
index 8e7af05..abc1316 100644
--- a/mlir/lib/Dialect/Transform/SMTExtension/SMTExtensionOps.cpp
+++ b/mlir/lib/Dialect/Transform/SMTExtension/SMTExtensionOps.cpp
@@ -8,8 +8,8 @@
 
 #include "mlir/Dialect/Transform/SMTExtension/SMTExtensionOps.h"
 #include "mlir/Dialect/SMT/IR/SMTDialect.h"
-#include "mlir/Dialect/Transform/IR/TransformOps.h"
-#include "mlir/Dialect/Transform/SMTExtension/SMTExtension.h"
+#include "mlir/Dialect/SMT/IR/SMTOps.h"
+#include "mlir/Dialect/Transform/IR/TransformTypes.h"
 
 using namespace mlir;
 
@@ -23,6 +23,7 @@ using namespace mlir;
 void transform::smt::ConstrainParamsOp::getEffects(
     SmallVectorImpl<MemoryEffects::EffectInstance> &effects) {
   onlyReadsHandle(getParamsMutable(), effects);
+  producesHandle(getResults(), effects);
 }
 
 DiagnosedSilenceableFailure
@@ -37,19 +38,95 @@ transform::smt::ConstrainParamsOp::apply(transform::TransformRewriter &rewriter,
   //       and allow for users to attach their own implementation, which would,
   //       e.g., translate the ops to SMTLIB and hand that over to the user's
   //       favourite solver. This requires changes to the dialect's verifier.
-  return emitDefiniteFailure() << "op does not have interpreted semantics yet";
+  return emitSilenceableFailure(getLoc())
+         << "op does not have interpreted semantics yet";
 }
 
 LogicalResult transform::smt::ConstrainParamsOp::verify() {
+  auto yieldTerminator =
+      dyn_cast<mlir::smt::YieldOp>(getRegion().front().back());
+  if (!yieldTerminator)
+    return emitOpError() << "expected '"
+                         << mlir::smt::YieldOp::getOperationName()
+                         << "' as terminator";
+
+  auto checkTypes = [](size_t idx, Type smtType, StringRef smtDesc,
+                       Type paramType, StringRef paramDesc,
+                       auto *atOp) -> InFlightDiagnostic {
+    if (!isa<mlir::smt::BoolType, mlir::smt::IntType, mlir::smt::BitVectorType>(
+            smtType))
+      return atOp->emitOpError() << "the type of " << smtDesc << " #" << idx
+                                 << " is expected to be either a !smt.bool, a "
+                                    "!smt.int, or a !smt.bv";
+
+    assert(isa<TransformParamTypeInterface>(paramType) &&
+           "ODS specifies params' type should implement param interface");
+    if (isa<transform::AnyParamType>(paramType))
+      return {}; // No further checks can be done.
+
+    // NB: This cast must succeed as long as the only implementors of
+    //     TransformParamTypeInterface are AnyParamType and ParamType.
+    Type typeWrappedByParam = cast<ParamType>(paramType).getType();
+
+    if (isa<mlir::smt::IntType>(smtType)) {
+      if (!isa<IntegerType>(typeWrappedByParam))
+        return atOp->emitOpError()
+               << "the type of " << smtDesc << " #" << idx
+               << " is !smt.int though the corresponding " << paramDesc
+               << " type (" << paramType << ") is not wrapping an integer type";
+    } else if (isa<mlir::smt::BoolType>(smtType)) {
+      auto wrappedIntType = dyn_cast<IntegerType>(typeWrappedByParam);
+      if (!wrappedIntType || wrappedIntType.getWidth() != 1)
+        return atOp->emitOpError()
+               << "the type of " << smtDesc << " #" << idx
+               << " is !smt.bool though the corresponding " << paramDesc
+               << " type (" << paramType << ") is not wrapping i1";
+    } else if (auto bvSmtType = dyn_cast<mlir::smt::BitVectorType>(smtType)) {
+      auto wrappedIntType = dyn_cast<IntegerType>(typeWrappedByParam);
+      if (!wrappedIntType || wrappedIntType.getWidth() != bvSmtType.getWidth())
+        return atOp->emitOpError()
+               << "the type of " << smtDesc << " #" << idx << " is " << smtType
+               << " though the corresponding " << paramDesc << " type ("
+               << paramType
+               << ") is not wrapping an integer type of the same bitwidth";
+    }
+
+    return {};
+  };
+
   if (getOperands().size() != getBody().getNumArguments())
     return emitOpError(
         "must have the same number of block arguments as operands");
 
+  for (auto [idx, operandType, blockArgType] :
+       llvm::enumerate(getOperandTypes(), getBody().getArgumentTypes())) {
+    InFlightDiagnostic typeCheckResult =
+        checkTypes(idx, blockArgType, "block arg", operandType, "operand",
+                   /*atOp=*/this);
+    if (LogicalResult(typeCheckResult).failed())
+      return typeCheckResult;
+  }
+
   for (auto &op : getBody().getOps()) {
     if (!isa<mlir::smt::SMTDialect>(op.getDialect()))
       return emitOpError(
           "ops contained in region should belong to SMT-dialect");
   }
 
+  if (yieldTerminator->getNumOperands() != getNumResults())
+    return yieldTerminator.emitOpError()
+           << "expected terminator to have as many operands as the parent op "
+              "has results";
+
+  for (auto [idx, termOperandType, resultType] : llvm::enumerate(
+           yieldTerminator->getOperands().getType(), getResultTypes())) {
+    InFlightDiagnostic typeCheckResult =
+        checkTypes(idx, termOperandType, "terminator operand",
+                   cast<transform::ParamType>(resultType), "result",
+                   /*atOp=*/&yieldTerminator);
+    if (LogicalResult(typeCheckResult).failed())
+      return typeCheckResult;
+  }
+
   return success();
 }
diff --git a/mlir/python/mlir/dialects/transform/smt.py b/mlir/python/mlir/dialects/transform/smt.py
index 1f0b7f0..af88fff 100644
--- a/mlir/python/mlir/dialects/transform/smt.py
+++ b/mlir/python/mlir/dialects/transform/smt.py
@@ -19,6 +19,7 @@ except ImportError as e:
 class ConstrainParamsOp(ConstrainParamsOp):
     def __init__(
         self,
+        results: Sequence[Type],
         params: Sequence[transform.AnyParamType],
         arg_types: Sequence[Type],
         loc=None,
@@ -27,6 +28,7 @@ class ConstrainParamsOp(ConstrainParamsOp):
         if len(params) != len(arg_types):
             raise ValueError(f"{params=} not same length as {arg_types=}")
         super().__init__(
+            results,
             params,
             loc=loc,
             ip=ip,
@@ -36,3 +38,13 @@ class ConstrainParamsOp(ConstrainParamsOp):
     @property
     def body(self) -> Block:
         return self.regions[0].blocks[0]
+
+
+def constrain_params(
+    results: Sequence[Type],
+    params: Sequence[transform.AnyParamType],
+    arg_types: Sequence[Type],
+    loc=None,
+    ip=None,
+):
+    return ConstrainParamsOp(results, params, arg_types, loc=loc, ip=ip)
diff --git a/mlir/test/Dialect/Transform/test-smt-extension-invalid.mlir b/mlir/test/Dialect/Transform/test-smt-extension-invalid.mlir
index 314b8d4..d91d69a 100644
--- a/mlir/test/Dialect/Transform/test-smt-extension-invalid.mlir
+++ b/mlir/test/Dialect/Transform/test-smt-extension-invalid.mlir
@@ -1,11 +1,40 @@
 // RUN: mlir-opt %s --transform-interpreter --split-input-file --verify-diagnostics
 
+// CHECK-LABEL: @incorrect terminator
+module attributes {transform.with_named_sequence} {
+  transform.named_sequence @operands_not_one_to_one_with_vars(%arg0: !transform.any_op {transform.readonly}) {
+    %param_as_param = transform.param.constant 42 -> !transform.param<i64>
+    // expected-error@below {{op expected 'smt.yield' as terminator}}
+    transform.smt.constrain_params(%param_as_param) : (!transform.param<i64>) -> () {
+      ^bb0(%param_as_smt_var: !smt.int):
+      transform.yield
+    }
+    transform.yield
+  }
+}
+
+// -----
+
+// CHECK-LABEL: @operands_not_one_to_one_with_vars
+module attributes {transform.with_named_sequence} {
+  transform.named_sequence @operands_not_one_to_one_with_vars(%arg0: !transform.any_op {transform.readonly}) {
+    %param_as_param = transform.param.constant 42 -> !transform.param<i64>
+    // expected-error@below {{must have the same number of block arguments as operands}}
+    transform.smt.constrain_params(%param_as_param) : (!transform.param<i64>) -> () {
+      ^bb0(%param_as_smt_var: !smt.int, %param_as_another_smt_var: !smt.int):
+    }
+    transform.yield
+  }
+}
+
+// -----
+
 // CHECK-LABEL: @constraint_not_using_smt_ops
 module attributes {transform.with_named_sequence} {
   transform.named_sequence @constraint_not_using_smt_ops(%arg0: !transform.any_op {transform.readonly}) {
     %param_as_param = transform.param.constant 42 -> !transform.param<i64>
     // expected-error@below {{ops contained in region should belong to SMT-dialect}}
-    transform.smt.constrain_params(%param_as_param) : !transform.param<i64> {
+    transform.smt.constrain_params(%param_as_param) : (!transform.param<i64>) -> () {
       ^bb0(%param_as_smt_var: !smt.int):
       %c4 = arith.constant 4 : i32
       // This is the kind of thing one might think works:
@@ -17,13 +46,90 @@ module attributes {transform.with_named_sequence} {
 
 // -----
 
-// CHECK-LABEL: @operands_not_one_to_one_with_vars
+// CHECK-LABEL: @results_not_one_to_one_with_vars
 module attributes {transform.with_named_sequence} {
-  transform.named_sequence @operands_not_one_to_one_with_vars(%arg0: !transform.any_op {transform.readonly}) {
+  transform.named_sequence @results_not_one_to_one_with_vars(%arg0: !transform.any_op {transform.readonly}) {
     %param_as_param = transform.param.constant 42 -> !transform.param<i64>
-    // expected-error@below {{must have the same number of block arguments as operands}}
-    transform.smt.constrain_params(%param_as_param) : !transform.param<i64> {
+    transform.smt.constrain_params(%param_as_param, %param_as_param) : (!transform.param<i64>, !transform.param<i64>) -> () {
       ^bb0(%param_as_smt_var: !smt.int, %param_as_another_smt_var: !smt.int):
+      // expected-error@below {{expected terminator to have as many operands as the parent op has results}}
+      smt.yield %param_as_smt_var : !smt.int
+    }
+    transform.yield
+  }
+}
+
+// -----
+
+// CHECK-LABEL: @non_smt_type_block_args
+module attributes {transform.with_named_sequence} {
+  transform.named_sequence @non_smt_type_block_args(%arg0: !transform.any_op {transform.readonly}) {
+    %param_as_param = transform.param.constant 42 -> !transform.param<i8>
+    // expected-error@below {{the type of block arg #0 is expected to be either a !smt.bool, a !smt.int, or a !smt.bv}}
+    transform.smt.constrain_params(%param_as_param) : (!transform.param<i8>) -> (!transform.param<i8>) {
+      ^bb0(%param_as_smt_var: !transform.param<i8>):
+      smt.yield %param_as_smt_var : !transform.param<i8>
+    }
+    transform.yield
+  }
+}
+
+
+// -----
+
+// CHECK-LABEL: @mismatched_arg_type_bool
+module attributes {transform.with_named_sequence} {
+  transform.named_sequence @mismatched_arg_type_bool(%arg0: !transform.any_op {transform.readonly}) {
+    %param_as_param = transform.param.constant 42 -> !transform.param<i64>
+    // expected-error@below {{the type of block arg #0 is !smt.bool though the corresponding operand type ('!transform.param<i64>') is not wrapping i1}}
+    transform.smt.constrain_params(%param_as_param) : (!transform.param<i64>) -> (!transform.param<i64>) {
+      ^bb0(%param_as_smt_var: !smt.bool):
+      smt.yield %param_as_smt_var : !smt.bool
+    }
+    transform.yield
+  }
+}
+
+// -----
+
+// CHECK-LABEL: @mismatched_arg_type_bitvector
+module attributes {transform.with_named_sequence} {
+  transform.named_sequence @mismatched_arg_type_bitvector(%arg0: !transform.any_op {transform.readonly}) {
+    %param_as_param = transform.param.constant 42 -> !transform.param<i64>
+    // expected-error@below {{the type of block arg #0 is '!smt.bv<8>' though the corresponding operand type ('!transform.param<i64>') is not wrapping an integer type of the same bitwidth}}
+    transform.smt.constrain_params(%param_as_param) : (!transform.param<i64>) -> (!transform.param<i64>) {
+      ^bb0(%param_as_smt_var: !smt.bv<8>):
+      smt.yield %param_as_smt_var : !smt.bv<8>
+    }
+    transform.yield
+  }
+}
+
+// -----
+
+// CHECK-LABEL: @mismatched_result_type_bool
+module attributes {transform.with_named_sequence} {
+  transform.named_sequence @mismatched_result_type_bool(%arg0: !transform.any_op {transform.readonly}) {
+    %param_as_param = transform.param.constant 1 -> !transform.param<i1>
+    transform.smt.constrain_params(%param_as_param) : (!transform.param<i1>) -> (!transform.param<i64>) {
+      ^bb0(%param_as_smt_var: !smt.bool):
+      // expected-error@below {{the type of terminator operand #0 is !smt.bool though the corresponding result type ('!transform.param<i64>') is not wrapping i1}}
+      smt.yield %param_as_smt_var : !smt.bool
+    }
+    transform.yield
+  }
+}
+
+// -----
+
+// CHECK-LABEL: @mismatched_result_type_bitvector
+module attributes {transform.with_named_sequence} {
+  transform.named_sequence @mismatched_result_type_bitvector(%arg0: !transform.any_op {transform.readonly}) {
+    %param_as_param = transform.param.constant 42 -> !transform.param<i8>
+    transform.smt.constrain_params(%param_as_param) : (!transform.param<i8>) -> (!transform.param<i64>) {
+      ^bb0(%param_as_smt_var: !smt.bv<8>):
+      // expected-error@below {{the type of terminator operand #0 is '!smt.bv<8>' though the corresponding result type ('!transform.param<i64>') is not wrapping an integer type of the same bitwidth}}
+      smt.yield %param_as_smt_var : !smt.bv<8>
     }
     transform.yield
   }
diff --git a/mlir/test/Dialect/Transform/test-smt-extension.mlir b/mlir/test/Dialect/Transform/test-smt-extension.mlir
index 29d1517..6cc41dd 100644
--- a/mlir/test/Dialect/Transform/test-smt-extension.mlir
+++ b/mlir/test/Dialect/Transform/test-smt-extension.mlir
@@ -7,7 +7,7 @@ module attributes {transform.with_named_sequence} {
     %param_as_param = transform.param.constant 42 -> !transform.param<i64>
 
     // CHECK: transform.smt.constrain_params(%[[PARAM_AS_PARAM]])
-    transform.smt.constrain_params(%param_as_param) : !transform.param<i64> {
+    transform.smt.constrain_params(%param_as_param) : (!transform.param<i64>) -> () {
       // CHECK: ^bb{{.*}}(%[[PARAM_AS_SMT_SYMB:.*]]: !smt.int):
       ^bb0(%param_as_smt_var: !smt.int):
       // CHECK: %[[C0:.*]] = smt.int.constant 0
@@ -31,18 +31,20 @@ module attributes {transform.with_named_sequence} {
 
 // -----
 
-// CHECK-LABEL: @schedule_with_constraint_on_multiple_params
+// CHECK-LABEL: @schedule_with_constraint_on_multiple_params_returning_computed_value
 module attributes {transform.with_named_sequence} {
-  transform.named_sequence @schedule_with_constraint_on_multiple_params(%arg0: !transform.any_op {transform.readonly}) {
+  transform.named_sequence @schedule_with_constraint_on_multiple_params_returning_computed_value(%arg0: !transform.any_op {transform.readonly}) {
     // CHECK: %[[PARAM_A:.*]] = transform.param.constant
     %param_a = transform.param.constant 4 -> !transform.param<i64>
     // CHECK: %[[PARAM_B:.*]] = transform.param.constant
-    %param_b = transform.param.constant 16 -> !transform.param<i64>
+    %param_b = transform.param.constant 32 -> !transform.param<i64>
 
     // CHECK: transform.smt.constrain_params(%[[PARAM_A]], %[[PARAM_B]])
-    transform.smt.constrain_params(%param_a, %param_b) : !transform.param<i64>, !transform.param<i64> {
+    %divisor = transform.smt.constrain_params(%param_a, %param_b) : (!transform.param<i64>, !transform.param<i64>) -> (!transform.param<i64>) {
       // CHECK: ^bb{{.*}}(%[[VAR_A:.*]]: !smt.int, %[[VAR_B:.*]]: !smt.int):
       ^bb0(%var_a: !smt.int, %var_b: !smt.int):
+      // CHECK: %[[DIV:.*]] = smt.int.div %[[VAR_B]], %[[VAR_A]]
+      %divisor = smt.int.div %var_b, %var_a
       // CHECK: %[[C0:.*]] = smt.int.constant 0
       %c0 = smt.int.constant 0
       // CHECK: %[[REMAINDER:.*]] = smt.int.mod %[[VAR_B]], %[[VAR_A]]
@@ -51,8 +53,11 @@ module attributes {transform.with_named_sequence} {
       %eq = smt.eq %remainder, %c0 : !smt.int
       // CHECK: smt.assert %[[EQ]]
       smt.assert %eq
+      // CHECK: smt.yield %[[DIV]]
+      smt.yield %divisor : !smt.int
     }
-    // NB: from here can rely on that %param_a is a divisor of %param_b
+    // NB: from here can rely on that %param_a is a divisor of %param_b and
+    //     that the relevant factor, 8, got associated to %divisor.
     transform.yield
   }
 }
@@ -63,10 +68,10 @@ module attributes {transform.with_named_sequence} {
 module attributes {transform.with_named_sequence} {
   transform.named_sequence @schedule_with_param_as_a_bool(%arg0: !transform.any_op {transform.readonly}) {
     // CHECK: %[[PARAM_AS_PARAM:.*]] = transform.param.constant
-    %param_as_param = transform.param.constant true -> !transform.any_param
+    %param_as_param = transform.param.constant true -> !transform.param<i1>
 
     // CHECK: transform.smt.constrain_params(%[[PARAM_AS_PARAM]])
-    transform.smt.constrain_params(%param_as_param) : !transform.any_param {
+    transform.smt.constrain_params(%param_as_param) : (!transform.param<i1>) -> () {
       // CHECK: ^bb{{.*}}(%[[PARAM_AS_SMT_VAR:.*]]: !smt.bool):
       ^bb0(%param_as_smt_var: !smt.bool):
       // CHECK: %[[C0:.*]] = smt.int.constant 0
diff --git a/mlir/test/python/dialects/transform_smt_ext.py b/mlir/test/python/dialects/transform_smt_ext.py
index 3692fd9..e28c56f 100644
--- a/mlir/test/python/dialects/transform_smt_ext.py
+++ b/mlir/test/python/dialects/transform_smt_ext.py
@@ -25,26 +25,44 @@ def run(f):
 # CHECK-LABEL: TEST: testConstrainParamsOp
 @run
 def testConstrainParamsOp(target):
-    dummy_value = ir.IntegerAttr.get(ir.IntegerType.get_signless(32), 42)
+    c42_attr = ir.IntegerAttr.get(ir.IntegerType.get_signless(32), 42)
     # CHECK: %[[PARAM_AS_PARAM:.*]] = transform.param.constant
-    symbolic_value = transform.ParamConstantOp(
-        transform.AnyParamType.get(), dummy_value
+    symbolic_value_as_param = transform.ParamConstantOp(
+        transform.AnyParamType.get(), c42_attr
     )
     # CHECK: transform.smt.constrain_params(%[[PARAM_AS_PARAM]])
     constrain_params = transform_smt.ConstrainParamsOp(
-        [symbolic_value], [smt.IntType.get()]
+        [], [symbolic_value_as_param], [smt.IntType.get()]
     )
     # CHECK-NEXT: ^bb{{.*}}(%[[PARAM_AS_SMT_SYMB:.*]]: !smt.int):
     with ir.InsertionPoint(constrain_params.body):
+        symbolic_value_as_smt_var = constrain_params.body.arguments[0]
         # CHECK: %[[C0:.*]] = smt.int.constant 0
         c0 = smt.IntConstantOp(ir.IntegerAttr.get(ir.IntegerType.get_signless(32), 0))
         # CHECK: %[[C43:.*]] = smt.int.constant 43
         c43 = smt.IntConstantOp(ir.IntegerAttr.get(ir.IntegerType.get_signless(32), 43))
         # CHECK: %[[LB:.*]] = smt.int.cmp le %[[C0]], %[[PARAM_AS_SMT_SYMB]]
-        lb = smt.IntCmpOp(smt.IntPredicate.le, c0, constrain_params.body.arguments[0])
+        lb = smt.IntCmpOp(smt.IntPredicate.le, c0, symbolic_value_as_smt_var)
         # CHECK: %[[UB:.*]] = smt.int.cmp le %[[PARAM_AS_SMT_SYMB]], %[[C43]]
-        ub = smt.IntCmpOp(smt.IntPredicate.le, constrain_params.body.arguments[0], c43)
+        ub = smt.IntCmpOp(smt.IntPredicate.le, symbolic_value_as_smt_var, c43)
         # CHECK: %[[BOUNDED:.*]] = smt.and %[[LB]], %[[UB]]
         bounded = smt.AndOp([lb, ub])
         # CHECK: smt.assert %[[BOUNDED:.*]]
         smt.AssertOp(bounded)
+        smt.YieldOp([])
+
+    # CHECK: transform.smt.constrain_params(%[[PARAM_AS_PARAM]])
+    compute_with_params = transform_smt.ConstrainParamsOp(
+        [transform.ParamType.get(ir.IntegerType.get_signless(32))],
+        [symbolic_value_as_param],
+        [smt.IntType.get()],
+    )
+    # CHECK-NEXT: ^bb{{.*}}(%[[SMT_SYMB:.*]]: !smt.int):
+    with ir.InsertionPoint(compute_with_params.body):
+        symbolic_value_as_smt_var = compute_with_params.body.arguments[0]
+        # CHECK: %[[TWICE:.*]] = smt.int.add %[[SMT_SYMB]], %[[SMT_SYMB]]
+        twice_symb = smt.IntAddOp(
+            [symbolic_value_as_smt_var, symbolic_value_as_smt_var]
+        )
+        # CHECK: smt.yield %[[TWICE]]
+        smt.YieldOp([twice_symb])