| ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py |
| ; RUN: opt -slp-vectorizer -S -mtriple=x86_64-unknown-linux-gnu < %s | FileCheck %s |
| |
| define void @foo(i8* %c, float* %d) { |
| ; CHECK-LABEL: @foo( |
| ; CHECK-NEXT: entry: |
| ; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, i8* [[C:%.*]], i64 4 |
| ; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, i8* [[C]], i64 1 |
| ; CHECK-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i8, i8* [[C]], i64 2 |
| ; CHECK-NEXT: [[ARRAYIDX17:%.*]] = getelementptr inbounds i8, i8* [[C]], i64 3 |
| ; CHECK-NEXT: [[TMP0:%.*]] = bitcast i8* [[ARRAYIDX4]] to <4 x i8>* |
| ; CHECK-NEXT: [[TMP1:%.*]] = load <4 x i8>, <4 x i8>* [[TMP0]], align 1 |
| ; CHECK-NEXT: [[TMP2:%.*]] = zext <4 x i8> [[TMP1]] to <4 x i32> |
| ; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw <4 x i32> [[TMP2]], <i32 2, i32 2, i32 2, i32 3> |
| ; CHECK-NEXT: [[TMP4:%.*]] = and <4 x i32> [[TMP2]], <i32 2, i32 2, i32 2, i32 3> |
| ; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <4 x i32> [[TMP3]], <4 x i32> [[TMP4]], <4 x i32> <i32 1, i32 2, i32 7, i32 0> |
| ; CHECK-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds float, float* [[D:%.*]], i64 -1 |
| ; CHECK-NEXT: [[ADD_PTR37:%.*]] = getelementptr inbounds float, float* [[D]], i64 -2 |
| ; CHECK-NEXT: [[ADD_PTR45:%.*]] = getelementptr inbounds float, float* [[D]], i64 -3 |
| ; CHECK-NEXT: [[TMP6:%.*]] = add nsw <4 x i32> poison, [[TMP5]] |
| ; CHECK-NEXT: [[TMP7:%.*]] = sitofp <4 x i32> [[TMP6]] to <4 x float> |
| ; CHECK-NEXT: [[TMP8:%.*]] = fdiv <4 x float> [[TMP7]], poison |
| ; CHECK-NEXT: [[ADD_PTR53:%.*]] = getelementptr inbounds float, float* [[D]], i64 -4 |
| ; CHECK-NEXT: [[TMP9:%.*]] = bitcast float* [[ADD_PTR53]] to <4 x float>* |
| ; CHECK-NEXT: store <4 x float> [[TMP8]], <4 x float>* [[TMP9]], align 4 |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| %arrayidx1 = getelementptr inbounds i8, i8* %c, i64 4 |
| %0 = load i8, i8* %arrayidx1, align 1 |
| %conv2 = zext i8 %0 to i32 |
| %and = and i32 %conv2, 3 |
| %arrayidx4 = getelementptr inbounds i8, i8* %c, i64 1 |
| %1 = load i8, i8* %arrayidx4, align 1 |
| %conv5 = zext i8 %1 to i32 |
| %shl6 = shl nuw nsw i32 %conv5, 2 |
| %arrayidx12 = getelementptr inbounds i8, i8* %c, i64 2 |
| %2 = load i8, i8* %arrayidx12, align 1 |
| %conv13 = zext i8 %2 to i32 |
| %shl14 = shl nuw nsw i32 %conv13, 2 |
| %arrayidx17 = getelementptr inbounds i8, i8* %c, i64 3 |
| %3 = load i8, i8* %arrayidx17, align 1 |
| %conv18 = zext i8 %3 to i32 |
| %shl19 = shl nuw nsw i32 %conv18, 2 |
| %sub = add nsw i32 undef, %shl6 |
| %conv27 = sitofp i32 %sub to float |
| %div = fdiv float %conv27, undef |
| %add.ptr = getelementptr inbounds float, float* %d, i64 -1 |
| store float %div, float* %add.ptr, align 4 |
| %sub32 = add nsw i32 undef, %and |
| %conv33 = sitofp i32 %sub32 to float |
| %div36 = fdiv float %conv33, undef |
| %add.ptr37 = getelementptr inbounds float, float* %d, i64 -2 |
| store float %div36, float* %add.ptr37, align 4 |
| %sub40 = add nsw i32 undef, %shl19 |
| %conv41 = sitofp i32 %sub40 to float |
| %div44 = fdiv float %conv41, undef |
| %add.ptr45 = getelementptr inbounds float, float* %d, i64 -3 |
| store float %div44, float* %add.ptr45, align 4 |
| %sub48 = add nsw i32 undef, %shl14 |
| %conv49 = sitofp i32 %sub48 to float |
| %div52 = fdiv float %conv49, undef |
| %add.ptr53 = getelementptr inbounds float, float* %d, i64 -4 |
| store float %div52, float* %add.ptr53, align 4 |
| ret void |
| } |