diff options
Diffstat (limited to 'llvm/test/Transforms/SLPVectorizer/X86/minbitwidth-icmp-to-trunc.ll')
-rw-r--r-- | llvm/test/Transforms/SLPVectorizer/X86/minbitwidth-icmp-to-trunc.ll | 75 |
1 files changed, 75 insertions, 0 deletions
diff --git a/llvm/test/Transforms/SLPVectorizer/X86/minbitwidth-icmp-to-trunc.ll b/llvm/test/Transforms/SLPVectorizer/X86/minbitwidth-icmp-to-trunc.ll new file mode 100644 index 0000000..fc28d7a --- /dev/null +++ b/llvm/test/Transforms/SLPVectorizer/X86/minbitwidth-icmp-to-trunc.ll @@ -0,0 +1,75 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4 +; RUN: opt -S --passes=slp-vectorizer -mtriple=x86_64-unknown-linux -mcpu=cascadelake < %s | FileCheck %s + +define i1 @test(ptr noalias %0, i64 %1, ptr noalias %p, ptr %p1) { +; CHECK-LABEL: define i1 @test( +; CHECK-SAME: ptr noalias [[TMP0:%.*]], i64 [[TMP1:%.*]], ptr noalias [[P:%.*]], ptr [[P1:%.*]]) #[[ATTR0:[0-9]+]] { +; CHECK-NEXT: newFuncRoot: +; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16 +; CHECK-NEXT: [[BF_LOAD_I1336:%.*]] = load i24, ptr [[TMP2]], align 16 +; CHECK-NEXT: [[AND_I_I_I1342:%.*]] = and i64 [[TMP1]], -16 +; CHECK-NEXT: [[TMP3:%.*]] = inttoptr i64 [[AND_I_I_I1342]] to ptr +; CHECK-NEXT: store ptr [[TMP3]], ptr [[P]], align 8 +; CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 16 +; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i64 16 +; CHECK-NEXT: [[BF_LOAD_I1345:%.*]] = load i24, ptr [[TMP5]], align 16 +; CHECK-NEXT: [[TMP6:%.*]] = insertelement <2 x i24> poison, i24 [[BF_LOAD_I1336]], i32 0 +; CHECK-NEXT: [[TMP7:%.*]] = insertelement <2 x i24> [[TMP6]], i24 [[BF_LOAD_I1345]], i32 1 +; CHECK-NEXT: [[TMP8:%.*]] = and <2 x i24> [[TMP7]], <i24 255, i24 255> +; CHECK-NEXT: [[TMP9:%.*]] = icmp eq <2 x i24> [[TMP8]], <i24 24, i24 24> +; CHECK-NEXT: [[TMP10:%.*]] = select <2 x i1> [[TMP9]], <2 x i24> <i24 23, i24 23>, <2 x i24> [[TMP8]] +; CHECK-NEXT: [[TMP23:%.*]] = trunc <2 x i24> [[TMP10]] to <2 x i8> +; CHECK-NEXT: [[TMP11:%.*]] = zext <2 x i8> [[TMP23]] to <2 x i32> +; CHECK-NEXT: [[TMP12:%.*]] = and <2 x i32> [[TMP11]], <i32 254, i32 254> +; CHECK-NEXT: [[TMP13:%.*]] = icmp eq <2 x i32> [[TMP12]], <i32 4, i32 4> +; CHECK-NEXT: [[TMP25:%.*]] = select <2 x i1> [[TMP13]], <2 x i8> <i8 2, i8 2>, <2 x i8> [[TMP23]] +; CHECK-NEXT: [[TMP14:%.*]] = zext <2 x i8> [[TMP25]] to <2 x i32> +; CHECK-NEXT: [[TMP15:%.*]] = icmp eq <2 x i32> [[TMP14]], <i32 32, i32 32> +; CHECK-NEXT: [[TMP18:%.*]] = select <2 x i1> [[TMP15]], <2 x i8> <i8 31, i8 31>, <2 x i8> [[TMP25]] +; CHECK-NEXT: [[TMP16:%.*]] = zext <2 x i8> [[TMP18]] to <2 x i32> +; CHECK-NEXT: [[TMP17:%.*]] = icmp eq <2 x i32> [[TMP16]], <i32 54, i32 54> +; CHECK-NEXT: [[TMP21:%.*]] = select <2 x i1> [[TMP17]], <2 x i8> <i8 53, i8 53>, <2 x i8> [[TMP18]] +; CHECK-NEXT: [[TMP22:%.*]] = extractelement <2 x i8> [[TMP21]], i32 0 +; CHECK-NEXT: [[TMP19:%.*]] = zext i8 [[TMP22]] to i32 +; CHECK-NEXT: store i32 [[TMP19]], ptr [[P1]], align 4 +; CHECK-NEXT: [[TMP24:%.*]] = extractelement <2 x i8> [[TMP21]], i32 1 +; CHECK-NEXT: [[TMP20:%.*]] = zext i8 [[TMP24]] to i32 +; CHECK-NEXT: [[CMP210_NOT:%.*]] = icmp eq i32 [[TMP19]], [[TMP20]] +; CHECK-NEXT: ret i1 [[CMP210_NOT]] +; +newFuncRoot: + %2 = getelementptr inbounds i8, ptr %0, i64 16 + %bf.load.i1336 = load i24, ptr %2, align 16 + %bf.clear.i1337 = and i24 %bf.load.i1336, 255 + %and.i.i.i1342 = and i64 %1, -16 + %3 = inttoptr i64 %and.i.i.i1342 to ptr + store ptr %3, ptr %p, align 8 + %4 = load ptr, ptr %3, align 16 + %5 = getelementptr inbounds i8, ptr %4, i64 16 + %bf.load.i1345 = load i24, ptr %5, align 16 + %bf.clear.i1346 = and i24 %bf.load.i1345, 255 + %cmp182 = icmp eq i24 %bf.clear.i1337, 24 + %narrow = select i1 %cmp182, i24 23, i24 %bf.clear.i1337 + %s = zext nneg i24 %narrow to i32 + %cmp185 = icmp eq i24 %bf.clear.i1346, 24 + %narrow1790 = select i1 %cmp185, i24 23, i24 %bf.clear.i1346 + %s1139 = zext nneg i24 %narrow1790 to i32 + %6 = and i32 %s, 254 + %or.cond1132 = icmp eq i32 %6, 4 + %s1142 = select i1 %or.cond1132, i32 2, i32 %s + %7 = and i32 %s1139, 254 + %or.cond1133 = icmp eq i32 %7, 4 + %s1140 = select i1 %or.cond1133, i32 2, i32 %s1139 + %cmp198 = icmp eq i32 %s1142, 32 + %s1134 = select i1 %cmp198, i32 31, i32 %s1142 + %cmp201 = icmp eq i32 %s1140, 32 + %s1143 = select i1 %cmp201, i32 31, i32 %s1140 + %cmp204 = icmp eq i32 %s1134, 54 + %s1135 = select i1 %cmp204, i32 53, i32 %s1134 + store i32 %s1135, ptr %p1, align 4 + %cmp207 = icmp eq i32 %s1143, 54 + %s1141 = select i1 %cmp207, i32 53, i32 %s1143 + %cmp210.not = icmp eq i32 %s1135, %s1141 + ret i1 %cmp210.not +} + |