[RISCV] Add codegen support for Zvfbfmin (#87911)

This patch adds basic codegen support for Zvfbfmin extension.

19 files changed, 1032 insertions, 67 deletions

diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index 19ef1f2..2818e19 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -1087,6 +1087,23 @@ RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
       }
     }
 
+    // TODO: Could we merge some code with zvfhmin?
+    if (Subtarget.hasVInstructionsBF16()) {
+      for (MVT VT : BF16VecVTs) {
+        if (!isTypeLegal(VT))
+          continue;
+        setOperationAction({ISD::FP_ROUND, ISD::FP_EXTEND}, VT, Custom);
+        setOperationAction({ISD::VP_FP_ROUND, ISD::VP_FP_EXTEND}, VT, Custom);
+        setOperationAction({ISD::STRICT_FP_ROUND, ISD::STRICT_FP_EXTEND}, VT,
+                           Custom);
+        setOperationAction({ISD::CONCAT_VECTORS, ISD::INSERT_SUBVECTOR,
+                            ISD::EXTRACT_SUBVECTOR},
+                           VT, Custom);
+        setOperationAction({ISD::LOAD, ISD::STORE}, VT, Custom);
+        // TODO: Promote to fp32.
+      }
+    }
+
     if (Subtarget.hasVInstructionsF32()) {
       for (MVT VT : F32VecVTs) {
         if (!isTypeLegal(VT))
@@ -1302,6 +1319,19 @@ RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
           continue;
         }
 
+        if (VT.getVectorElementType() == MVT::bf16) {
+          setOperationAction({ISD::FP_ROUND, ISD::FP_EXTEND}, VT, Custom);
+          setOperationAction({ISD::VP_FP_ROUND, ISD::VP_FP_EXTEND}, VT, Custom);
+          setOperationAction({ISD::STRICT_FP_ROUND, ISD::STRICT_FP_EXTEND}, VT,
+                             Custom);
+          setOperationAction({ISD::CONCAT_VECTORS, ISD::INSERT_SUBVECTOR,
+                              ISD::EXTRACT_SUBVECTOR},
+                             VT, Custom);
+          setOperationAction({ISD::LOAD, ISD::STORE}, VT, Custom);
+          // TODO: Promote to fp32.
+          continue;
+        }
+
         // We use EXTRACT_SUBVECTOR as a "cast" from scalable to fixed.
         setOperationAction({ISD::INSERT_SUBVECTOR, ISD::EXTRACT_SUBVECTOR}, VT,
                            Custom);
@@ -2561,6 +2591,10 @@ static bool useRVVForFixedLengthVectorVT(MVT VT,
     if (!Subtarget.hasVInstructionsF16Minimal())
       return false;
     break;
+  case MVT::bf16:
+    if (!Subtarget.hasVInstructionsBF16())
+      return false;
+    break;
   case MVT::f32:
     if (!Subtarget.hasVInstructionsF32())
       return false;
@@ -2612,6 +2646,7 @@ static MVT getContainerForFixedLengthVector(const TargetLowering &TLI, MVT VT,
   case MVT::i16:
   case MVT::i32:
   case MVT::i64:
+  case MVT::bf16:
   case MVT::f16:
   case MVT::f32:
   case MVT::f64: {
@@ -8101,8 +8136,10 @@ RISCVTargetLowering::lowerStrictFPExtendOrRoundLike(SDValue Op,
 
   // RVV can only widen/truncate fp to types double/half the size as the source.
   if ((VT.getVectorElementType() == MVT::f64 &&
-       SrcVT.getVectorElementType() == MVT::f16) ||
-      (VT.getVectorElementType() == MVT::f16 &&
+       (SrcVT.getVectorElementType() == MVT::f16 ||
+        SrcVT.getVectorElementType() == MVT::bf16)) ||
+      ((VT.getVectorElementType() == MVT::f16 ||
+        VT.getVectorElementType() == MVT::bf16) &&
        SrcVT.getVectorElementType() == MVT::f64)) {
     // For double rounding, the intermediate rounding should be round-to-odd.
     unsigned InterConvOpc = Op.getOpcode() == ISD::STRICT_FP_EXTEND
@@ -8146,9 +8183,12 @@ RISCVTargetLowering::lowerVectorFPExtendOrRoundLike(SDValue Op,
   SDValue Src = Op.getOperand(0);
   MVT SrcVT = Src.getSimpleValueType();
 
-  bool IsDirectExtend = IsExtend && (VT.getVectorElementType() != MVT::f64 ||
-                                     SrcVT.getVectorElementType() != MVT::f16);
-  bool IsDirectTrunc = !IsExtend && (VT.getVectorElementType() != MVT::f16 ||
+  bool IsDirectExtend =
+      IsExtend && (VT.getVectorElementType() != MVT::f64 ||
+                   (SrcVT.getVectorElementType() != MVT::f16 &&
+                    SrcVT.getVectorElementType() != MVT::bf16));
+  bool IsDirectTrunc = !IsExtend && ((VT.getVectorElementType() != MVT::f16 &&
+                                      VT.getVectorElementType() != MVT::bf16) ||
                                      SrcVT.getVectorElementType() != MVT::f64);
 
   bool IsDirectConv = IsDirectExtend || IsDirectTrunc;
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
index 22e5488..4adc26f 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
@@ -355,24 +355,24 @@ defset list<VTypeInfo> AllVectors = {
                                  V_M8, f64, FPR64>;
     }
   }
-}
 
-defset list<VTypeInfo> AllBFloatVectors = {
-  defset list<VTypeInfo> NoGroupBFloatVectors = {
-    defset list<VTypeInfo> FractionalGroupBFloatVectors = {
-      def VBF16MF4: VTypeInfo<vbfloat16mf4_t, vbool64_t, 16, V_MF4, bf16, FPR16>;
-      def VBF16MF2: VTypeInfo<vbfloat16mf2_t, vbool32_t, 16, V_MF2, bf16, FPR16>;
+  defset list<VTypeInfo> AllBFloatVectors = {
+    defset list<VTypeInfo> NoGroupBFloatVectors = {
+      defset list<VTypeInfo> FractionalGroupBFloatVectors = {
+        def VBF16MF4: VTypeInfo<vbfloat16mf4_t, vbool64_t, 16, V_MF4, bf16, FPR16>;
+        def VBF16MF2: VTypeInfo<vbfloat16mf2_t, vbool32_t, 16, V_MF2, bf16, FPR16>;
+      }
+      def VBF16M1:  VTypeInfo<vbfloat16m1_t, vbool16_t, 16, V_M1, bf16, FPR16>;
+    }
+  
+    defset list<GroupVTypeInfo> GroupBFloatVectors = {
+      def VBF16M2: GroupVTypeInfo<vbfloat16m2_t, vbfloat16m1_t, vbool8_t, 16,
+                                  V_M2, bf16, FPR16>;
+      def VBF16M4: GroupVTypeInfo<vbfloat16m4_t, vbfloat16m1_t, vbool4_t, 16,
+                                  V_M4, bf16, FPR16>;
+      def VBF16M8: GroupVTypeInfo<vbfloat16m8_t, vbfloat16m1_t, vbool2_t, 16,
+                                  V_M8, bf16, FPR16>;
     }
-    def VBF16M1:  VTypeInfo<vbfloat16m1_t, vbool16_t, 16, V_M1, bf16, FPR16>;
-  }
-
-  defset list<GroupVTypeInfo> GroupBFloatVectors = {
-    def VBF16M2: GroupVTypeInfo<vbfloat16m2_t, vbfloat16m1_t, vbool8_t, 16,
-                                V_M2, bf16, FPR16>;
-    def VBF16M4: GroupVTypeInfo<vbfloat16m4_t, vbfloat16m1_t, vbool4_t, 16,
-                                V_M4, bf16, FPR16>;
-    def VBF16M8: GroupVTypeInfo<vbfloat16m8_t, vbfloat16m1_t, vbool2_t, 16,
-                                V_M8, bf16, FPR16>;
   }
 }
 
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td
index b4af83a..714f8cf 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td
@@ -1495,6 +1495,20 @@ foreach fvtiToFWti = AllWidenableFloatVectors in {
                 fvti.AVL, fvti.Log2SEW, TA_MA)>;
 }
 
+foreach fvtiToFWti = AllWidenableBFloatToFloatVectors in {
+  defvar fvti = fvtiToFWti.Vti;
+  defvar fwti = fvtiToFWti.Wti;
+  let Predicates = [HasVInstructionsBF16] in
+  def : Pat<(fvti.Vector (fpround (fwti.Vector fwti.RegClass:$rs1))),
+            (!cast<Instruction>("PseudoVFNCVTBF16_F_F_W_"#fvti.LMul.MX#"_E"#fvti.SEW)
+                (fvti.Vector (IMPLICIT_DEF)),
+                fwti.RegClass:$rs1,
+                // Value to indicate no rounding mode change in
+                // RISCVInsertReadWriteCSR
+                FRM_DYN,
+                fvti.AVL, fvti.Log2SEW, TA_MA)>;
+}
+
 //===----------------------------------------------------------------------===//
 // Vector Splats
 //===----------------------------------------------------------------------===//
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td
index 6c6ecb6..e10b8bf 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td
@@ -2670,6 +2670,20 @@ foreach fvtiToFWti = AllWidenableFloatVectors in {
                 GPR:$vl, fvti.Log2SEW, TA_MA)>;
 }
 
+foreach fvtiToFWti = AllWidenableBFloatToFloatVectors in {
+  defvar fvti = fvtiToFWti.Vti;
+  defvar fwti = fvtiToFWti.Wti;
+  let Predicates = [HasVInstructionsBF16] in
+  def : Pat<(fwti.Vector (any_riscv_fpextend_vl
+                             (fvti.Vector fvti.RegClass:$rs1),
+                             (fvti.Mask V0),
+                             VLOpFrag)),
+            (!cast<Instruction>("PseudoVFWCVTBF16_F_F_V_"#fvti.LMul.MX#"_E"#fvti.SEW#"_MASK")
+                (fwti.Vector (IMPLICIT_DEF)), fvti.RegClass:$rs1,
+                (fvti.Mask V0),
+                GPR:$vl, fvti.Log2SEW, TA_MA)>;
+}
+
 // 13.19 Narrowing Floating-Point/Integer Type-Convert Instructions
 defm : VPatNConvertFP2IVL_W_RM<riscv_vfcvt_xu_f_vl, "PseudoVFNCVT_XU_F_W">;
 defm : VPatNConvertFP2IVL_W_RM<riscv_vfcvt_x_f_vl, "PseudoVFNCVT_X_F_W">;
@@ -2714,6 +2728,22 @@ foreach fvtiToFWti = AllWidenableFloatVectors in {
   }
 }
 
+foreach fvtiToFWti = AllWidenableBFloatToFloatVectors in {
+  defvar fvti = fvtiToFWti.Vti;
+  defvar fwti = fvtiToFWti.Wti;
+  let Predicates = [HasVInstructionsBF16] in
+    def : Pat<(fvti.Vector (any_riscv_fpround_vl
+                               (fwti.Vector fwti.RegClass:$rs1),
+                               (fwti.Mask V0), VLOpFrag)),
+              (!cast<Instruction>("PseudoVFNCVTBF16_F_F_W_"#fvti.LMul.MX#"_E"#fvti.SEW#"_MASK")
+                  (fvti.Vector (IMPLICIT_DEF)), fwti.RegClass:$rs1,
+                  (fwti.Mask V0),
+                  // Value to indicate no rounding mode change in
+                  // RISCVInsertReadWriteCSR
+                  FRM_DYN,
+                  GPR:$vl, fvti.Log2SEW, TA_MA)>;
+}
+
 // 14. Vector Reduction Operations
 
 // 14.1. Vector Single-Width Integer Reduction Instructions
diff --git a/llvm/test/CodeGen/RISCV/rvv/extract-subvector.ll b/llvm/test/CodeGen/RISCV/rvv/extract-subvector.ll
index e15e645..4f1fcfb 100644
--- a/llvm/test/CodeGen/RISCV/rvv/extract-subvector.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/extract-subvector.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple riscv32 -mattr=+m,+d,+zfh,+zvfh,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple riscv64 -mattr=+m,+d,+zfh,+zvfh,+v -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple riscv32 -mattr=+m,+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple riscv64 -mattr=+m,+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
 
 define <vscale x 4 x i32> @extract_nxv8i32_nxv4i32_0(<vscale x 8 x i32> %vec) {
 ; CHECK-LABEL: extract_nxv8i32_nxv4i32_0:
@@ -481,6 +481,60 @@ define <vscale x 6 x half> @extract_nxv6f16_nxv12f16_6(<vscale x 12 x half> %in)
   ret <vscale x 6 x half> %res
 }
 
+define <vscale x 2 x bfloat> @extract_nxv2bf16_nxv16bf16_0(<vscale x 16 x bfloat> %vec) {
+; CHECK-LABEL: extract_nxv2bf16_nxv16bf16_0:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ret
+  %c = call <vscale x 2 x bfloat> @llvm.vector.extract.nxv2bf16.nxv16bf16(<vscale x 16 x bfloat> %vec, i64 0)
+  ret <vscale x 2 x bfloat> %c
+}
+
+define <vscale x 2 x bfloat> @extract_nxv2bf16_nxv16bf16_2(<vscale x 16 x bfloat> %vec) {
+; CHECK-LABEL: extract_nxv2bf16_nxv16bf16_2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    srli a0, a0, 2
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vslidedown.vx v8, v8, a0
+; CHECK-NEXT:    ret
+  %c = call <vscale x 2 x bfloat> @llvm.vector.extract.nxv2bf16.nxv16bf16(<vscale x 16 x bfloat> %vec, i64 2)
+  ret <vscale x 2 x bfloat> %c
+}
+
+define <vscale x 2 x bfloat> @extract_nxv2bf16_nxv16bf16_4(<vscale x 16 x bfloat> %vec) {
+; CHECK-LABEL: extract_nxv2bf16_nxv16bf16_4:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %c = call <vscale x 2 x bfloat> @llvm.vector.extract.nxv2bf16.nxv16bf16(<vscale x 16 x bfloat> %vec, i64 4)
+  ret <vscale x 2 x bfloat> %c
+}
+
+define <vscale x 6 x bfloat> @extract_nxv6bf16_nxv12bf16_0(<vscale x 12 x bfloat> %in) {
+; CHECK-LABEL: extract_nxv6bf16_nxv12bf16_0:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ret
+  %res = call <vscale x 6 x bfloat> @llvm.vector.extract.nxv6bf16.nxv12bf16(<vscale x 12 x bfloat> %in, i64 0)
+  ret <vscale x 6 x bfloat> %res
+}
+
+define <vscale x 6 x bfloat> @extract_nxv6bf16_nxv12bf16_6(<vscale x 12 x bfloat> %in) {
+; CHECK-LABEL: extract_nxv6bf16_nxv12bf16_6:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    srli a0, a0, 2
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vslidedown.vx v13, v10, a0
+; CHECK-NEXT:    vslidedown.vx v12, v9, a0
+; CHECK-NEXT:    add a1, a0, a0
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vslideup.vx v12, v10, a0
+; CHECK-NEXT:    vmv2r.v v8, v12
+; CHECK-NEXT:    ret
+  %res = call <vscale x 6 x bfloat> @llvm.vector.extract.nxv6bf16.nxv12bf16(<vscale x 12 x bfloat> %in, i64 6)
+  ret <vscale x 6 x bfloat> %res
+}
+
 declare <vscale x 6 x half> @llvm.vector.extract.nxv6f16.nxv12f16(<vscale x 12 x half>, i64)
 
 declare <vscale x 1 x i8> @llvm.vector.extract.nxv1i8.nxv4i8(<vscale x 4 x i8> %vec, i64 %idx)
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fpext-vp.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fpext-vp.ll
index 51ac27a..48cc3f1 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fpext-vp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fpext-vp.ll
@@ -1,8 +1,8 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
 
 declare <2 x float> @llvm.vp.fpext.v2f32.v2f16(<2 x half>, <2 x i1>, i32)
 
@@ -120,3 +120,53 @@ define <32 x double> @vfpext_v32f32_v32f64(<32 x float> %a, <32 x i1> %m, i32 ze
   %v = call <32 x double> @llvm.vp.fpext.v32f64.v32f32(<32 x float> %a, <32 x i1> %m, i32 %vl)
   ret <32 x double> %v
 }
+
+declare <2 x float> @llvm.vp.fpext.v2f32.v2bf16(<2 x bfloat>, <2 x i1>, i32)
+
+define <2 x float> @vfpext_v2bf16_v2f32(<2 x bfloat> %a, <2 x i1> %m, i32 zeroext %vl) {
+; CHECK-LABEL: vfpext_v2bf16_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <2 x float> @llvm.vp.fpext.v2f32.v2bf16(<2 x bfloat> %a, <2 x i1> %m, i32 %vl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfpext_v2bf16_v2f32_unmasked(<2 x bfloat> %a, i32 zeroext %vl) {
+; CHECK-LABEL: vfpext_v2bf16_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <2 x float> @llvm.vp.fpext.v2f32.v2bf16(<2 x bfloat> %a, <2 x i1> shufflevector (<2 x i1> insertelement (<2 x i1> undef, i1 true, i32 0), <2 x i1> undef, <2 x i32> zeroinitializer), i32 %vl)
+  ret <2 x float> %v
+}
+
+declare <2 x double> @llvm.vp.fpext.v2f64.v2bf16(<2 x bfloat>, <2 x i1>, i32)
+
+define <2 x double> @vfpext_v2bf16_v2f64(<2 x bfloat> %a, <2 x i1> %m, i32 zeroext %vl) {
+; CHECK-LABEL: vfpext_v2bf16_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8, v0.t
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x double> @llvm.vp.fpext.v2f64.v2bf16(<2 x bfloat> %a, <2 x i1> %m, i32 %vl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfpext_v2bf16_v2f64_unmasked(<2 x bfloat> %a, i32 zeroext %vl) {
+; CHECK-LABEL: vfpext_v2bf16_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <2 x double> @llvm.vp.fpext.v2f64.v2bf16(<2 x bfloat> %a, <2 x i1> shufflevector (<2 x i1> insertelement (<2 x i1> undef, i1 true, i32 0), <2 x i1> undef, <2 x i32> zeroinitializer), i32 %vl)
+  ret <2 x double> %v
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fptrunc-vp.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fptrunc-vp.ll
index de11f9e..d890bf5 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fptrunc-vp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fptrunc-vp.ll
@@ -1,8 +1,8 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
 
 
 declare <2 x half> @llvm.vp.fptrunc.v2f16.v2f32(<2 x float>, <2 x i1>, i32)
@@ -122,3 +122,53 @@ define <32 x float> @vfptrunc_v32f32_v32f64(<32 x double> %a, <32 x i1> %m, i32
   %v = call <32 x float> @llvm.vp.fptrunc.v32f64.v32f32(<32 x double> %a, <32 x i1> %m, i32 %vl)
   ret <32 x float> %v
 }
+
+declare <2 x bfloat> @llvm.vp.fptrunc.v2bf16.v2f32(<2 x float>, <2 x i1>, i32)
+
+define <2 x bfloat> @vfptrunc_v2bf16_v2f32(<2 x float> %a, <2 x i1> %m, i32 zeroext %vl) {
+; CHECK-LABEL: vfptrunc_v2bf16_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v9, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <2 x bfloat> @llvm.vp.fptrunc.v2bf16.v2f32(<2 x float> %a, <2 x i1> %m, i32 %vl)
+  ret <2 x bfloat> %v
+}
+
+define <2 x bfloat> @vfptrunc_v2bf16_v2f32_unmasked(<2 x float> %a, i32 zeroext %vl) {
+; CHECK-LABEL: vfptrunc_v2bf16_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <2 x bfloat> @llvm.vp.fptrunc.v2bf16.v2f32(<2 x float> %a, <2 x i1> shufflevector (<2 x i1> insertelement (<2 x i1> undef, i1 true, i32 0), <2 x i1> undef, <2 x i32> zeroinitializer), i32 %vl)
+  ret <2 x bfloat> %v
+}
+
+declare <2 x bfloat> @llvm.vp.fptrunc.v2bf16.v2f64(<2 x double>, <2 x i1>, i32)
+
+define <2 x bfloat> @vfptrunc_v2bf16_v2f64(<2 x double> %a, <2 x i1> %m, i32 zeroext %vl) {
+; CHECK-LABEL: vfptrunc_v2bf16_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v9, v8, v0.t
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x bfloat> @llvm.vp.fptrunc.v2bf16.v2f64(<2 x double> %a, <2 x i1> %m, i32 %vl)
+  ret <2 x bfloat> %v
+}
+
+define <2 x bfloat> @vfptrunc_v2bf16_v2f64_unmasked(<2 x double> %a, i32 zeroext %vl) {
+; CHECK-LABEL: vfptrunc_v2bf16_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v9, v8
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v9
+; CHECK-NEXT:    ret
+  %v = call <2 x bfloat> @llvm.vp.fptrunc.v2bf16.v2f64(<2 x double> %a, <2 x i1> shufflevector (<2 x i1> insertelement (<2 x i1> undef, i1 true, i32 0), <2 x i1> undef, <2 x i32> zeroinitializer), i32 %vl)
+  ret <2 x bfloat> %v
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-load-store.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-load-store.ll
index 38aee56..fbe8bcb 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-load-store.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-load-store.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
-; RUN: llc -mtriple=riscv32 -mattr=+v,+zfh,+zvfh -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV32 %s
-; RUN: llc -mtriple=riscv64 -mattr=+v,+zfh,+zvfh -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV64 %s
+; RUN: llc -mtriple=riscv32 -mattr=+v,+zfh,+zvfh,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV32 %s
+; RUN: llc -mtriple=riscv64 -mattr=+v,+zfh,+zvfh,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV64 %s
 
 define void @v2i8(ptr %p, ptr %q) {
 ; CHECK-LABEL: v2i8:
@@ -301,3 +301,15 @@ define void @v2i8_volatile_store(ptr %p, ptr %q) {
   store volatile <2 x i8> %v, ptr %q
   ret void
 }
+
+define void @v8bf16(ptr %p, ptr %q) {
+; CHECK-LABEL: v8bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
+; CHECK-NEXT:    vle16.v v8, (a0)
+; CHECK-NEXT:    vse16.v v8, (a1)
+; CHECK-NEXT:    ret
+  %v = load <8 x bfloat>, ptr %p
+  store <8 x bfloat> %v, ptr %q
+  ret void
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-load.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-load.ll
index 791e6eb..d80d75d 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-load.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-load.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
-; RUN: llc -mtriple=riscv32 -mattr=+v,+zfh,+zvfh -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV32 %s
-; RUN: llc -mtriple=riscv64 -mattr=+v,+zfh,+zvfh -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV64 %s
+; RUN: llc -mtriple=riscv32 -mattr=+v,+zfh,+zvfh,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV32 %s
+; RUN: llc -mtriple=riscv64 -mattr=+v,+zfh,+zvfh,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV64 %s
 
 define <5 x i8> @load_v5i8(ptr %p) {
 ; CHECK-LABEL: load_v5i8:
@@ -181,3 +181,13 @@ define <16 x i64> @exact_vlen_i64_m8(ptr %p) vscale_range(2,2) {
   %v = load <16 x i64>, ptr %p
   ret <16 x i64> %v
 }
+
+define <8 x bfloat> @load_v8bf16(ptr %p) {
+; CHECK-LABEL: load_v8bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
+; CHECK-NEXT:    vle16.v v8, (a0)
+; CHECK-NEXT:    ret
+  %x = load <8 x bfloat>, ptr %p
+  ret <8 x bfloat> %x
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-store.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-store.ll
index b747d73..6317a49 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-store.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-store.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
-; RUN: llc -mtriple=riscv32 -mattr=+v,+zfh,+zvfh -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV32 %s
-; RUN: llc -mtriple=riscv64 -mattr=+v,+zfh,+zvfh -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV64 %s
+; RUN: llc -mtriple=riscv32 -mattr=+v,+zfh,+zvfh,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV32 %s
+; RUN: llc -mtriple=riscv64 -mattr=+v,+zfh,+zvfh,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck -check-prefixes=CHECK,RV64 %s
 
 define void @store_v5i8(ptr %p, <5 x i8> %v) {
 ; CHECK-LABEL: store_v5i8:
@@ -294,6 +294,16 @@ define void @exact_vlen_i64_m8(ptr %p) vscale_range(2,2) {
   ret void
 }
 
+define void @store_v8bf16(ptr %p, <8 x bfloat> %v) {
+; CHECK-LABEL: store_v8bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
+; CHECK-NEXT:    vse16.v v8, (a0)
+; CHECK-NEXT:    ret
+  store <8 x bfloat> %v, ptr %p
+  ret void
+}
+
 ;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
 ; RV32: {{.*}}
 ; RV64: {{.*}}
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfpext-constrained-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfpext-constrained-sdnode.ll
index b0e6a6a..5d90762 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfpext-constrained-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfpext-constrained-sdnode.ll
@@ -1,7 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v -target-abi=ilp32d \
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v -target-abi=lp64d \
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 
 declare <2 x float> @llvm.experimental.constrained.fpext.v2f32.v2f16(<2 x half>, metadata)
@@ -114,3 +114,78 @@ define <8 x double> @vfpext_v8f32_v8f64(<8 x float> %va) strictfp {
   %evec = call <8 x double> @llvm.experimental.constrained.fpext.v8f64.v8f32(<8 x float> %va, metadata !"fpexcept.strict")
   ret <8 x double> %evec
 }
+
+declare <2 x float> @llvm.experimental.constrained.fpext.v2f32.v2bf16(<2 x bfloat>, metadata)
+define <2 x float> @vfpext_v2bf16_v2f32(<2 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_v2bf16_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <2 x float> @llvm.experimental.constrained.fpext.v2f32.v2bf16(<2 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <2 x float> %evec
+}
+
+declare <2 x double> @llvm.experimental.constrained.fpext.v2f64.v2bf16(<2 x bfloat>, metadata)
+define <2 x double> @vfpext_v2bf16_v2f64(<2 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_v2bf16_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <2 x double> @llvm.experimental.constrained.fpext.v2f64.v2bf16(<2 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <2 x double> %evec
+}
+
+declare <4 x float> @llvm.experimental.constrained.fpext.v4f32.v4bf16(<4 x bfloat>, metadata)
+define <4 x float> @vfpext_v4bf16_v4f32(<4 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_v4bf16_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <4 x float> @llvm.experimental.constrained.fpext.v4f32.v4bf16(<4 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <4 x float> %evec
+}
+
+declare <4 x double> @llvm.experimental.constrained.fpext.v4f64.v4bf16(<4 x bfloat>, metadata)
+define <4 x double> @vfpext_v4bf16_v4f64(<4 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_v4bf16_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v10, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v10
+; CHECK-NEXT:    ret
+  %evec = call <4 x double> @llvm.experimental.constrained.fpext.v4f64.v4bf16(<4 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <4 x double> %evec
+}
+
+declare <8 x float> @llvm.experimental.constrained.fpext.v8f32.v8bf16(<8 x bfloat>, metadata)
+define <8 x float> @vfpext_v8bf16_v8f32(<8 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_v8bf16_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v10, v8
+; CHECK-NEXT:    vmv2r.v v8, v10
+; CHECK-NEXT:    ret
+  %evec = call <8 x float> @llvm.experimental.constrained.fpext.v8f32.v8bf16(<8 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <8 x float> %evec
+}
+
+declare <8 x double> @llvm.experimental.constrained.fpext.v8f64.v8bf16(<8 x bfloat>, metadata)
+define <8 x double> @vfpext_v8bf16_v8f64(<8 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_v8bf16_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v12, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v12
+; CHECK-NEXT:    ret
+  %evec = call <8 x double> @llvm.experimental.constrained.fpext.v8f64.v8bf16(<8 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <8 x double> %evec
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfptrunc-constrained-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfptrunc-constrained-sdnode.ll
index fd53113..5781223 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfptrunc-constrained-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfptrunc-constrained-sdnode.ll
@@ -1,11 +1,11 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v -target-abi=ilp32d \
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v -target-abi=lp64d \
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v -target-abi=ilp32d \
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v -target-abi=lp64d \
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 
 declare <2 x float> @llvm.experimental.constrained.fptrunc.v2f32.v2f64(<2 x double>, metadata, metadata)
@@ -118,3 +118,78 @@ define <8 x half> @vfptrunc_v8f32_v8f16(<8 x float> %va) strictfp {
   %evec = call <8 x half> @llvm.experimental.constrained.fptrunc.v8f16.v8f32(<8 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
   ret <8 x half> %evec
 }
+
+declare <2 x bfloat> @llvm.experimental.constrained.fptrunc.v2bf16.v2f64(<2 x double>, metadata, metadata)
+define <2 x bfloat> @vfptrunc_v2f64_v2bf16(<2 x double> %va) strictfp {
+; CHECK-LABEL: vfptrunc_v2f64_v2bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v9, v8
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <2 x bfloat> @llvm.experimental.constrained.fptrunc.v2bf16.v2f64(<2 x double> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <2 x bfloat> %evec
+}
+
+declare <2 x bfloat> @llvm.experimental.constrained.fptrunc.v2bf16.v2f32(<2 x float>, metadata, metadata)
+define <2 x bfloat> @vfptrunc_v2f32_v2bf16(<2 x float> %va) strictfp {
+; CHECK-LABEL: vfptrunc_v2f32_v2bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <2 x bfloat> @llvm.experimental.constrained.fptrunc.v2bf16.v2f32(<2 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <2 x bfloat> %evec
+}
+
+declare <4 x bfloat> @llvm.experimental.constrained.fptrunc.v4bf16.v4f64(<4 x double>, metadata, metadata)
+define <4 x bfloat> @vfptrunc_v4f64_v4bf16(<4 x double> %va) strictfp {
+; CHECK-LABEL: vfptrunc_v4f64_v4bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v10, v8
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v10
+; CHECK-NEXT:    ret
+  %evec = call <4 x bfloat> @llvm.experimental.constrained.fptrunc.v4bf16.v4f64(<4 x double> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <4 x bfloat> %evec
+}
+
+declare <4 x bfloat> @llvm.experimental.constrained.fptrunc.v4bf16.v4f32(<4 x float>, metadata, metadata)
+define <4 x bfloat> @vfptrunc_v4f32_v4bf16(<4 x float> %va) strictfp {
+; CHECK-LABEL: vfptrunc_v4f32_v4bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <4 x bfloat> @llvm.experimental.constrained.fptrunc.v4bf16.v4f32(<4 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <4 x bfloat> %evec
+}
+
+declare <8 x bfloat> @llvm.experimental.constrained.fptrunc.v8bf16.v8f64(<8 x double>, metadata, metadata)
+define <8 x bfloat> @vfptrunc_v8f64_v8bf16(<8 x double> %va) strictfp {
+; CHECK-LABEL: vfptrunc_v8f64_v8bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v12, v8
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v12
+; CHECK-NEXT:    ret
+  %evec = call <8 x bfloat> @llvm.experimental.constrained.fptrunc.v8bf16.v8f64(<8 x double> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <8 x bfloat> %evec
+}
+
+declare <8 x bfloat> @llvm.experimental.constrained.fptrunc.v8bf16.v8f32(<8 x float>, metadata, metadata)
+define <8 x bfloat> @vfptrunc_v8f32_v8bf16(<8 x float> %va) strictfp {
+; CHECK-LABEL: vfptrunc_v8f32_v8bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v10, v8
+; CHECK-NEXT:    vmv.v.v v8, v10
+; CHECK-NEXT:    ret
+  %evec = call <8 x bfloat> @llvm.experimental.constrained.fptrunc.v8bf16.v8f32(<8 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <8 x bfloat> %evec
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/insert-subvector.ll b/llvm/test/CodeGen/RISCV/rvv/insert-subvector.ll
index b1589658..0cd4f42 100644
--- a/llvm/test/CodeGen/RISCV/rvv/insert-subvector.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/insert-subvector.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple riscv32 -mattr=+m,+d,+zfh,+zvfh,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple riscv64 -mattr=+m,+d,+zfh,+zvfh,+v -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple riscv32 -mattr=+m,+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple riscv64 -mattr=+m,+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
 
 define <vscale x 8 x i32> @insert_nxv8i32_nxv4i32_0(<vscale x 8 x i32> %vec, <vscale x 4 x i32> %subvec) {
 ; CHECK-LABEL: insert_nxv8i32_nxv4i32_0:
@@ -531,6 +531,65 @@ define <vscale x 8 x i32> @insert_insert_combine2(<vscale x 2 x i32> %subvec) {
   ret <vscale x 8 x i32> %outer
 }
 
+define <vscale x 32 x bfloat> @insert_nxv32bf16_nxv2bf16_0(<vscale x 32 x bfloat> %vec, <vscale x 2 x bfloat> %subvec) {
+; CHECK-LABEL: insert_nxv32bf16_nxv2bf16_0:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    srli a0, a0, 2
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    vmv.v.v v8, v16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x bfloat> @llvm.vector.insert.nxv2bf16.nxv32bf16(<vscale x 32 x bfloat> %vec, <vscale x 2 x bfloat> %subvec, i64 0)
+  ret <vscale x 32 x bfloat> %v
+}
+
+define <vscale x 32 x bfloat> @insert_nxv32bf16_nxv2bf16_2(<vscale x 32 x bfloat> %vec, <vscale x 2 x bfloat> %subvec) {
+; CHECK-LABEL: insert_nxv32bf16_nxv2bf16_2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    srli a0, a0, 2
+; CHECK-NEXT:    add a1, a0, a0
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vslideup.vx v8, v16, a0
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x bfloat> @llvm.vector.insert.nxv2bf16.nxv32bf16(<vscale x 32 x bfloat> %vec, <vscale x 2 x bfloat> %subvec, i64 2)
+  ret <vscale x 32 x bfloat> %v
+}
+
+define <vscale x 32 x bfloat> @insert_nxv32bf16_nxv2bf16_26(<vscale x 32 x bfloat> %vec, <vscale x 2 x bfloat> %subvec) {
+; CHECK-LABEL: insert_nxv32bf16_nxv2bf16_26:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    srli a0, a0, 2
+; CHECK-NEXT:    add a1, a0, a0
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vslideup.vx v14, v16, a0
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x bfloat> @llvm.vector.insert.nxv2bf16.nxv32bf16(<vscale x 32 x bfloat> %vec, <vscale x 2 x bfloat> %subvec, i64 26)
+  ret <vscale x 32 x bfloat> %v
+}
+
+define <vscale x 32 x bfloat> @insert_nxv32bf16_undef_nxv1bf16_0(<vscale x 1 x bfloat> %subvec) {
+; CHECK-LABEL: insert_nxv32bf16_undef_nxv1bf16_0:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x bfloat> @llvm.vector.insert.nxv1bf16.nxv32bf16(<vscale x 32 x bfloat> undef, <vscale x 1 x bfloat> %subvec, i64 0)
+  ret <vscale x 32 x bfloat> %v
+}
+
+define <vscale x 32 x bfloat> @insert_nxv32bf16_undef_nxv1bf16_26(<vscale x 1 x bfloat> %subvec) {
+; CHECK-LABEL: insert_nxv32bf16_undef_nxv1bf16_26:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    csrr a0, vlenb
+; CHECK-NEXT:    srli a1, a0, 3
+; CHECK-NEXT:    srli a0, a0, 2
+; CHECK-NEXT:    add a1, a0, a1
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vslideup.vx v14, v8, a0
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x bfloat> @llvm.vector.insert.nxv1bf16.nxv32bf16(<vscale x 32 x bfloat> undef, <vscale x 1 x bfloat> %subvec, i64 26)
+  ret <vscale x 32 x bfloat> %v
+}
 
 attributes #0 = { vscale_range(2,1024) }
 
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfpext-constrained-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/vfpext-constrained-sdnode.ll
index 5de3097..8b49b72 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vfpext-constrained-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfpext-constrained-sdnode.ll
@@ -1,7 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v -target-abi=ilp32d \
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v -target-abi=lp64d \
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 
 declare <vscale x 1 x float> @llvm.experimental.constrained.fpext.nxv1f32.nxv1f16(<vscale x 1 x half>, metadata)
@@ -151,3 +151,103 @@ define <vscale x 8 x double> @vfpext_nxv8f32_nxv8f64(<vscale x 8 x float> %va) s
   %evec = call <vscale x 8 x double> @llvm.experimental.constrained.fpext.nxv8f64.nxv8f32(<vscale x 8 x float> %va, metadata !"fpexcept.strict")
   ret <vscale x 8 x double> %evec
 }
+
+declare <vscale x 1 x float> @llvm.experimental.constrained.fpext.nxv1f32.nxv1bf16(<vscale x 1 x bfloat>, metadata)
+define <vscale x 1 x float> @vfpext_nxv1bf16_nxv1f32(<vscale x 1 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_nxv1bf16_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 1 x float> @llvm.experimental.constrained.fpext.nxv1f32.nxv1bf16(<vscale x 1 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <vscale x 1 x float> %evec
+}
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fpext.nxv1f64.nxv1bf16(<vscale x 1 x bfloat>, metadata)
+define <vscale x 1 x double> @vfpext_nxv1bf16_nxv1f64(<vscale x 1 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_nxv1bf16_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 1 x double> @llvm.experimental.constrained.fpext.nxv1f64.nxv1bf16(<vscale x 1 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <vscale x 1 x double> %evec
+}
+
+declare <vscale x 2 x float> @llvm.experimental.constrained.fpext.nxv2f32.nxv2bf16(<vscale x 2 x bfloat>, metadata)
+define <vscale x 2 x float> @vfpext_nxv2bf16_nxv2f32(<vscale x 2 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_nxv2bf16_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 2 x float> @llvm.experimental.constrained.fpext.nxv2f32.nxv2bf16(<vscale x 2 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <vscale x 2 x float> %evec
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fpext.nxv2f64.nxv2bf16(<vscale x 2 x bfloat>, metadata)
+define <vscale x 2 x double> @vfpext_nxv2bf16_nxv2f64(<vscale x 2 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_nxv2bf16_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v10, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v10
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 2 x double> @llvm.experimental.constrained.fpext.nxv2f64.nxv2bf16(<vscale x 2 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <vscale x 2 x double> %evec
+}
+
+declare <vscale x 4 x float> @llvm.experimental.constrained.fpext.nxv4f32.nxv4bf16(<vscale x 4 x bfloat>, metadata)
+define <vscale x 4 x float> @vfpext_nxv4bf16_nxv4f32(<vscale x 4 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_nxv4bf16_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v10, v8
+; CHECK-NEXT:    vmv2r.v v8, v10
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 4 x float> @llvm.experimental.constrained.fpext.nxv4f32.nxv4bf16(<vscale x 4 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <vscale x 4 x float> %evec
+}
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fpext.nxv4f64.nxv4bf16(<vscale x 4 x bfloat>, metadata)
+define <vscale x 4 x double> @vfpext_nxv4bf16_nxv4f64(<vscale x 4 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_nxv4bf16_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v12, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v12
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 4 x double> @llvm.experimental.constrained.fpext.nxv4f64.nxv4bf16(<vscale x 4 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <vscale x 4 x double> %evec
+}
+
+declare <vscale x 8 x float> @llvm.experimental.constrained.fpext.nxv8f32.nxv8bf16(<vscale x 8 x bfloat>, metadata)
+define <vscale x 8 x float> @vfpext_nxv8bf16_nxv8f32(<vscale x 8 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_nxv8bf16_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v12, v8
+; CHECK-NEXT:    vmv4r.v v8, v12
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 8 x float> @llvm.experimental.constrained.fpext.nxv8f32.nxv8bf16(<vscale x 8 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <vscale x 8 x float> %evec
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fpext.nxv8f64.nxv8bf16(<vscale x 8 x bfloat>, metadata)
+define <vscale x 8 x double> @vfpext_nxv8bf16_nxv8f64(<vscale x 8 x bfloat> %va) strictfp {
+; CHECK-LABEL: vfpext_nxv8bf16_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v16, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v16
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 8 x double> @llvm.experimental.constrained.fpext.nxv8f64.nxv8bf16(<vscale x 8 x bfloat> %va, metadata !"fpexcept.strict")
+  ret <vscale x 8 x double> %evec
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfpext-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/vfpext-sdnode.ll
index d805a10..b002b8e 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vfpext-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfpext-sdnode.ll
@@ -1,11 +1,11 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v -target-abi=ilp32d \
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v -target-abi=lp64d \
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v -target-abi=ilp32d \
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v -target-abi=lp64d \
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 
 define <vscale x 1 x float> @vfpext_nxv1f16_nxv1f32(<vscale x 1 x half> %va) {
@@ -167,3 +167,115 @@ define <vscale x 8 x double> @vfpext_nxv8f32_nxv8f64(<vscale x 8 x float> %va) {
   %evec = fpext <vscale x 8 x float> %va to <vscale x 8 x double>
   ret <vscale x 8 x double> %evec
 }
+
+define <vscale x 1 x float> @vfpext_nxv1bf16_nxv1f32(<vscale x 1 x bfloat> %va) {
+;
+; CHECK-LABEL: vfpext_nxv1bf16_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = fpext <vscale x 1 x bfloat> %va to <vscale x 1 x float>
+  ret <vscale x 1 x float> %evec
+}
+
+define <vscale x 1 x double> @vfpext_nxv1bf16_nxv1f64(<vscale x 1 x bfloat> %va) {
+;
+; CHECK-LABEL: vfpext_nxv1bf16_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = fpext <vscale x 1 x bfloat> %va to <vscale x 1 x double>
+  ret <vscale x 1 x double> %evec
+}
+
+define <vscale x 2 x float> @vfpext_nxv2bf16_nxv2f32(<vscale x 2 x bfloat> %va) {
+;
+; CHECK-LABEL: vfpext_nxv2bf16_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = fpext <vscale x 2 x bfloat> %va to <vscale x 2 x float>
+  ret <vscale x 2 x float> %evec
+}
+
+define <vscale x 2 x double> @vfpext_nxv2bf16_nxv2f64(<vscale x 2 x bfloat> %va) {
+;
+; CHECK-LABEL: vfpext_nxv2bf16_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v10, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v10
+; CHECK-NEXT:    ret
+  %evec = fpext <vscale x 2 x bfloat> %va to <vscale x 2 x double>
+  ret <vscale x 2 x double> %evec
+}
+
+define <vscale x 4 x float> @vfpext_nxv4bf16_nxv4f32(<vscale x 4 x bfloat> %va) {
+;
+; CHECK-LABEL: vfpext_nxv4bf16_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v10, v8
+; CHECK-NEXT:    vmv2r.v v8, v10
+; CHECK-NEXT:    ret
+  %evec = fpext <vscale x 4 x bfloat> %va to <vscale x 4 x float>
+  ret <vscale x 4 x float> %evec
+}
+
+define <vscale x 4 x double> @vfpext_nxv4bf16_nxv4f64(<vscale x 4 x bfloat> %va) {
+;
+; CHECK-LABEL: vfpext_nxv4bf16_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v12, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v12
+; CHECK-NEXT:    ret
+  %evec = fpext <vscale x 4 x bfloat> %va to <vscale x 4 x double>
+  ret <vscale x 4 x double> %evec
+}
+
+define <vscale x 8 x float> @vfpext_nxv8bf16_nxv8f32(<vscale x 8 x bfloat> %va) {
+;
+; CHECK-LABEL: vfpext_nxv8bf16_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v12, v8
+; CHECK-NEXT:    vmv4r.v v8, v12
+; CHECK-NEXT:    ret
+  %evec = fpext <vscale x 8 x bfloat> %va to <vscale x 8 x float>
+  ret <vscale x 8 x float> %evec
+}
+
+define <vscale x 8 x double> @vfpext_nxv8bf16_nxv8f64(<vscale x 8 x bfloat> %va) {
+;
+; CHECK-LABEL: vfpext_nxv8bf16_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v16, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v16
+; CHECK-NEXT:    ret
+  %evec = fpext <vscale x 8 x bfloat> %va to <vscale x 8 x double>
+  ret <vscale x 8 x double> %evec
+}
+
+define <vscale x 16 x float> @vfpext_nxv16bf16_nxv16f32(<vscale x 16 x bfloat> %va) {
+;
+; CHECK-LABEL: vfpext_nxv16bf16_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v16, v8
+; CHECK-NEXT:    vmv8r.v v8, v16
+; CHECK-NEXT:    ret
+  %evec = fpext <vscale x 16 x bfloat> %va to <vscale x 16 x float>
+  ret <vscale x 16 x float> %evec
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfpext-vp.ll b/llvm/test/CodeGen/RISCV/rvv/vfpext-vp.ll
index 5cfa989..aaaf4ad 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vfpext-vp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfpext-vp.ll
@@ -1,8 +1,8 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
 
 declare <vscale x 2 x float> @llvm.vp.fpext.nxv2f32.nxv2f16(<vscale x 2 x half>, <vscale x 2 x i1>, i32)
 
@@ -120,3 +120,54 @@ define <vscale x 32 x float> @vfpext_nxv32f16_nxv32f32(<vscale x 32 x half> %a,
   %v = call <vscale x 32 x float> @llvm.vp.fpext.nxv32f32.nxv32f16(<vscale x 32 x half> %a, <vscale x 32 x i1> %m, i32 %vl)
   ret <vscale x 32 x float> %v
 }
+
+declare <vscale x 2 x float> @llvm.vp.fpext.nxv2f32.nxv2bf16(<vscale x 2 x bfloat>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x float> @vfpext_nxv2bf16_nxv2f32(<vscale x 2 x bfloat> %a, <vscale x 2 x i1> %m, i32 zeroext %vl) {
+; CHECK-LABEL: vfpext_nxv2bf16_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x float> @llvm.vp.fpext.nxv2f32.nxv2bf16(<vscale x 2 x bfloat> %a, <vscale x 2 x i1> %m, i32 %vl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfpext_nxv2bf16_nxv2f32_unmasked(<vscale x 2 x bfloat> %a, i32 zeroext %vl) {
+; CHECK-LABEL: vfpext_nxv2bf16_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x float> @llvm.vp.fpext.nxv2f32.nxv2bf16(<vscale x 2 x bfloat> %a, <vscale x 2 x i1> splat (i1 true), i32 %vl)
+  ret <vscale x 2 x float> %v
+}
+
+declare <vscale x 2 x double> @llvm.vp.fpext.nxv2f64.nxv2bf16(<vscale x 2 x bfloat>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x double> @vfpext_nxv2bf16_nxv2f64(<vscale x 2 x bfloat> %a, <vscale x 2 x i1> %m, i32 zeroext %vl) {
+; CHECK-LABEL: vfpext_nxv2bf16_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v10, v8, v0.t
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x double> @llvm.vp.fpext.nxv2f64.nxv2bf16(<vscale x 2 x bfloat> %a, <vscale x 2 x i1> %m, i32 %vl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfpext_nxv2bf16_nxv2f64_unmasked(<vscale x 2 x bfloat> %a, i32 zeroext %vl) {
+; CHECK-LABEL: vfpext_nxv2bf16_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwcvtbf16.f.f.v v10, v8
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v10
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x double> @llvm.vp.fpext.nxv2f64.nxv2bf16(<vscale x 2 x bfloat> %a, <vscale x 2 x i1> splat (i1 true), i32 %vl)
+  ret <vscale x 2 x double> %v
+}
+
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfptrunc-constrained-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/vfptrunc-constrained-sdnode.ll
index 4404a27..4341f45 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vfptrunc-constrained-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfptrunc-constrained-sdnode.ll
@@ -1,11 +1,11 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v -target-abi=ilp32d \
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v -target-abi=lp64d \
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v -target-abi=ilp32d \
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v -target-abi=lp64d \
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 
 declare <vscale x 1 x float> @llvm.experimental.constrained.fptrunc.nxv1f32.nxv1f64(<vscale x 1 x double>, metadata, metadata)
@@ -155,3 +155,103 @@ define <vscale x 8 x half> @vfptrunc_nxv8f32_nxv8f16(<vscale x 8 x float> %va) s
   %evec = call <vscale x 8 x half> @llvm.experimental.constrained.fptrunc.nxv8f16.nxv8f32(<vscale x 8 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
   ret <vscale x 8 x half> %evec
 }
+
+declare <vscale x 1 x bfloat> @llvm.experimental.constrained.fptrunc.nxv1bf16.nxv1f64(<vscale x 1 x double>, metadata, metadata)
+define <vscale x 1 x bfloat> @vfptrunc_nxv1f64_nxv1bf16(<vscale x 1 x double> %va) strictfp {
+; CHECK-LABEL: vfptrunc_nxv1f64_nxv1bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v9, v8
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 1 x bfloat> @llvm.experimental.constrained.fptrunc.nxv1bf16.nxv1f64(<vscale x 1 x double> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <vscale x 1 x bfloat> %evec
+}
+
+declare <vscale x 1 x bfloat> @llvm.experimental.constrained.fptrunc.nxv1bf16.nxv1f32(<vscale x 1 x float>, metadata, metadata)
+define <vscale x 1 x bfloat> @vfptrunc_nxv1f32_nxv1bf16(<vscale x 1 x float> %va) strictfp {
+; CHECK-LABEL: vfptrunc_nxv1f32_nxv1bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 1 x bfloat> @llvm.experimental.constrained.fptrunc.nxv1bf16.nxv1f32(<vscale x 1 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <vscale x 1 x bfloat> %evec
+}
+
+declare <vscale x 2 x bfloat> @llvm.experimental.constrained.fptrunc.nxv2bf16.nxv2f64(<vscale x 2 x double>, metadata, metadata)
+define <vscale x 2 x bfloat> @vfptrunc_nxv2f64_nxv2bf16(<vscale x 2 x double> %va) strictfp {
+; CHECK-LABEL: vfptrunc_nxv2f64_nxv2bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v10, v8
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v10
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 2 x bfloat> @llvm.experimental.constrained.fptrunc.nxv2bf16.nxv2f64(<vscale x 2 x double> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <vscale x 2 x bfloat> %evec
+}
+
+declare <vscale x 2 x bfloat> @llvm.experimental.constrained.fptrunc.nxv2bf16.nxv2f32(<vscale x 2 x float>, metadata, metadata)
+define <vscale x 2 x bfloat> @vfptrunc_nxv2f32_nxv2bf16(<vscale x 2 x float> %va) strictfp {
+; CHECK-LABEL: vfptrunc_nxv2f32_nxv2bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 2 x bfloat> @llvm.experimental.constrained.fptrunc.nxv2bf16.nxv2f32(<vscale x 2 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <vscale x 2 x bfloat> %evec
+}
+
+declare <vscale x 4 x bfloat> @llvm.experimental.constrained.fptrunc.nxv4bf16.nxv4f64(<vscale x 4 x double>, metadata, metadata)
+define <vscale x 4 x bfloat> @vfptrunc_nxv4f64_nxv4bf16(<vscale x 4 x double> %va) strictfp {
+; CHECK-LABEL: vfptrunc_nxv4f64_nxv4bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v12, v8
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v12
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 4 x bfloat> @llvm.experimental.constrained.fptrunc.nxv4bf16.nxv4f64(<vscale x 4 x double> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <vscale x 4 x bfloat> %evec
+}
+
+declare <vscale x 4 x bfloat> @llvm.experimental.constrained.fptrunc.nxv4bf16.nxv4f32(<vscale x 4 x float>, metadata, metadata)
+define <vscale x 4 x bfloat> @vfptrunc_nxv4f32_nxv4bf16(<vscale x 4 x float> %va) strictfp {
+; CHECK-LABEL: vfptrunc_nxv4f32_nxv4bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v10, v8
+; CHECK-NEXT:    vmv.v.v v8, v10
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 4 x bfloat> @llvm.experimental.constrained.fptrunc.nxv4bf16.nxv4f32(<vscale x 4 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <vscale x 4 x bfloat> %evec
+}
+
+declare <vscale x 8 x bfloat> @llvm.experimental.constrained.fptrunc.nxv8bf16.nxv8f64(<vscale x 8 x double>, metadata, metadata)
+define <vscale x 8 x bfloat> @vfptrunc_nxv8f64_nxv8bf16(<vscale x 8 x double> %va) strictfp {
+; CHECK-LABEL: vfptrunc_nxv8f64_nxv8bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v16, v8
+; CHECK-NEXT:    vsetvli zero, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v16
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 8 x bfloat> @llvm.experimental.constrained.fptrunc.nxv8bf16.nxv8f64(<vscale x 8 x double> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <vscale x 8 x bfloat> %evec
+}
+
+declare <vscale x 8 x bfloat> @llvm.experimental.constrained.fptrunc.nxv8bf16.nxv8f32(<vscale x 8 x float>, metadata, metadata)
+define <vscale x 8 x bfloat> @vfptrunc_nxv8f32_nxv8bf16(<vscale x 8 x float> %va) strictfp {
+; CHECK-LABEL: vfptrunc_nxv8f32_nxv8bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v12, v8
+; CHECK-NEXT:    vmv.v.v v8, v12
+; CHECK-NEXT:    ret
+  %evec = call <vscale x 8 x bfloat> @llvm.experimental.constrained.fptrunc.nxv8bf16.nxv8f32(<vscale x 8 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.strict")
+  ret <vscale x 8 x bfloat> %evec
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfptrunc-sdnode.ll b/llvm/test/CodeGen/RISCV/rvv/vfptrunc-sdnode.ll
index d715b46..9148a79 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vfptrunc-sdnode.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfptrunc-sdnode.ll
@@ -1,11 +1,11 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v -target-abi=ilp32d \
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v -target-abi=lp64d \
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+experimental-zvfbfmin -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v -target-abi=ilp32d \
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v -target-abi=lp64d \
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v,+experimental-zvfbfmin -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 
 define <vscale x 1 x half> @vfptrunc_nxv1f32_nxv1f16(<vscale x 1 x float> %va) {
@@ -167,3 +167,76 @@ define <vscale x 8 x float> @vfptrunc_nxv8f64_nxv8f32(<vscale x 8 x double> %va)
   %evec = fptrunc <vscale x 8 x double> %va to <vscale x 8 x float>
   ret <vscale x 8 x float> %evec
 }
+
+define <vscale x 1 x bfloat> @vfptrunc_nxv1f32_nxv1bf16(<vscale x 1 x float> %va) {
+;
+; CHECK-LABEL: vfptrunc_nxv1f32_nxv1bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = fptrunc <vscale x 1 x float> %va to <vscale x 1 x bfloat>
+  ret <vscale x 1 x bfloat> %evec
+}
+
+define <vscale x 2 x bfloat> @vfptrunc_nxv2f32_nxv2bf16(<vscale x 2 x float> %va) {
+;
+; CHECK-LABEL: vfptrunc_nxv2f32_nxv2bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %evec = fptrunc <vscale x 2 x float> %va to <vscale x 2 x bfloat>
+  ret <vscale x 2 x bfloat> %evec
+}
+
+define <vscale x 4 x bfloat> @vfptrunc_nxv4f32_nxv4bf16(<vscale x 4 x float> %va) {
+;
+; CHECK-LABEL: vfptrunc_nxv4f32_nxv4bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v10, v8
+; CHECK-NEXT:    vmv.v.v v8, v10
+; CHECK-NEXT:    ret
+  %evec = fptrunc <vscale x 4 x float> %va to <vscale x 4 x bfloat>
+  ret <vscale x 4 x bfloat> %evec
+}
+
+define <vscale x 8 x bfloat> @vfptrunc_nxv8f32_nxv8bf16(<vscale x 8 x float> %va) {
+;
+; CHECK-LABEL: vfptrunc_nxv8f32_nxv8bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v12, v8
+; CHECK-NEXT:    vmv.v.v v8, v12
+; CHECK-NEXT:    ret
+  %evec = fptrunc <vscale x 8 x float> %va to <vscale x 8 x bfloat>
+  ret <vscale x 8 x bfloat> %evec
+}
+
+define <vscale x 16 x bfloat> @vfptrunc_nxv16f32_nxv16bf16(<vscale x 16 x float> %va) {
+;
+; CHECK-LABEL: vfptrunc_nxv16f32_nxv16bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v16, v8
+; CHECK-NEXT:    vmv.v.v v8, v16
+; CHECK-NEXT:    ret
+  %evec = fptrunc <vscale x 16 x float> %va to <vscale x 16 x bfloat>
+  ret <vscale x 16 x bfloat> %evec
+}
+
+define <vscale x 1 x bfloat> @vfptrunc_nxv1f64_nxv1bf16(<vscale x 1 x double> %va) {
+;
+; CHECK-LABEL: vfptrunc_nxv1f64_nxv1bf16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v9, v8
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v9
+; CHECK-NEXT:    ret
+  %evec = fptrunc <vscale x 1 x double> %va to <vscale x 1 x bfloat>
+  ret <vscale x 1 x bfloat> %evec
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfptrunc-vp.ll b/llvm/test/CodeGen/RISCV/rvv/vfptrunc-vp.ll
index dd122f1..0c3abe3 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vfptrunc-vp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfptrunc-vp.ll
@@ -1,8 +1,8 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+m -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+m -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v,+m -verify-machineinstrs < %s | FileCheck %s
-; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v,+m -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+m,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+m,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfhmin,+v,+m,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfhmin,+v,+m,+experimental-zvfbfmin -verify-machineinstrs < %s | FileCheck %s
 
 declare <vscale x 2 x half> @llvm.vp.fptrunc.nxv2f16.nxv2f32(<vscale x 2 x float>, <vscale x 2 x i1>, i32)
 
@@ -218,3 +218,53 @@ define <vscale x 32 x float> @vfptrunc_nxv32f32_nxv32f64(<vscale x 32 x double>
   %v = call <vscale x 32 x float> @llvm.vp.fptrunc.nxv32f64.nxv32f32(<vscale x 32 x double> %a, <vscale x 32 x i1> %m, i32 %vl)
   ret <vscale x 32 x float> %v
 }
+
+declare <vscale x 2 x bfloat> @llvm.vp.fptrunc.nxv2bf16.nxv2f32(<vscale x 2 x float>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x bfloat> @vfptrunc_nxv2bf16_nxv2f32(<vscale x 2 x float> %a, <vscale x 2 x i1> %m, i32 zeroext %vl) {
+; CHECK-LABEL: vfptrunc_nxv2bf16_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v9, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x bfloat> @llvm.vp.fptrunc.nxv2bf16.nxv2f32(<vscale x 2 x float> %a, <vscale x 2 x i1> %m, i32 %vl)
+  ret <vscale x 2 x bfloat> %v
+}
+
+define <vscale x 2 x bfloat> @vfptrunc_nxv2bf16_nxv2f32_unmasked(<vscale x 2 x float> %a, i32 zeroext %vl) {
+; CHECK-LABEL: vfptrunc_nxv2bf16_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v9, v8
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x bfloat> @llvm.vp.fptrunc.nxv2bf16.nxv2f32(<vscale x 2 x float> %a, <vscale x 2 x i1> splat (i1 true), i32 %vl)
+  ret <vscale x 2 x bfloat> %v
+}
+
+declare <vscale x 2 x bfloat> @llvm.vp.fptrunc.nxv2bf16.nxv2f64(<vscale x 2 x double>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x bfloat> @vfptrunc_nxv2bf16_nxv2f64(<vscale x 2 x double> %a, <vscale x 2 x i1> %m, i32 zeroext %vl) {
+; CHECK-LABEL: vfptrunc_nxv2bf16_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v10, v8, v0.t
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x bfloat> @llvm.vp.fptrunc.nxv2bf16.nxv2f64(<vscale x 2 x double> %a, <vscale x 2 x i1> %m, i32 %vl)
+  ret <vscale x 2 x bfloat> %v
+}
+
+define <vscale x 2 x bfloat> @vfptrunc_nxv2bf16_nxv2f64_unmasked(<vscale x 2 x double> %a, i32 zeroext %vl) {
+; CHECK-LABEL: vfptrunc_nxv2bf16_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    vfncvt.rod.f.f.w v10, v8
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfncvtbf16.f.f.w v8, v10
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x bfloat> @llvm.vp.fptrunc.nxv2bf16.nxv2f64(<vscale x 2 x double> %a, <vscale x 2 x i1> splat (i1 true), i32 %vl)
+  ret <vscale x 2 x bfloat> %v
+}