1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5 2; RUN: %if x86-registered-target %{ opt -S --passes=slp-vectorizer -mtriple=x86_64-unknown-linux < %s | FileCheck %s %} 3; RUN: %if aarch64-registered-target %{ opt -S --passes=slp-vectorizer -mtriple=aarch64-unknown-linux < %s | FileCheck %s %} 4 5@a = external global [32000 x float], align 64 6 7define void @test() { 8; CHECK-LABEL: define void @test() { 9; CHECK-NEXT: [[ENTRY:.*]]: 10; CHECK-NEXT: br label %[[FOR_BODY:.*]] 11; CHECK: [[FOR_BODY]]: 12; CHECK-NEXT: [[TMP0:%.*]] = phi float [ 0.000000e+00, %[[ENTRY]] ], [ [[TMP16:%.*]], %[[FOR_BODY]] ] 13; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ] 14; CHECK-NEXT: [[TMP1:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 1 15; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [32000 x float], ptr @a, i64 0, i64 [[TMP1]] 16; CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [32000 x float], ptr @a, i64 0, i64 [[INDVARS_IV]] 17; CHECK-NEXT: [[TMP5:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 4 18; CHECK-NEXT: [[ARRAYIDX31:%.*]] = getelementptr inbounds [32000 x float], ptr @a, i64 0, i64 [[TMP5]] 19; CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[ARRAYIDX31]], align 4 20; CHECK-NEXT: [[TMP14:%.*]] = load <4 x float>, ptr [[ARRAYIDX]], align 4 21; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <4 x float> [[TMP14]], <4 x float> poison, <4 x i32> <i32 poison, i32 0, i32 1, i32 2> 22; CHECK-NEXT: [[TMP11:%.*]] = insertelement <4 x float> [[TMP4]], float [[TMP0]], i32 0 23; CHECK-NEXT: [[TMP15:%.*]] = fmul fast <4 x float> [[TMP11]], [[TMP14]] 24; CHECK-NEXT: store <4 x float> [[TMP15]], ptr [[ARRAYIDX6]], align 4 25; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 5 26; CHECK-NEXT: [[ARRAYIDX41:%.*]] = getelementptr inbounds [32000 x float], ptr @a, i64 0, i64 [[INDVARS_IV_NEXT]] 27; CHECK-NEXT: [[TMP16]] = load float, ptr [[ARRAYIDX41]], align 4 28; CHECK-NEXT: [[MUL45:%.*]] = fmul fast float [[TMP16]], [[TMP6]] 29; CHECK-NEXT: store float [[MUL45]], ptr [[ARRAYIDX31]], align 4 30; CHECK-NEXT: [[CMP2:%.*]] = icmp ult i64 [[INDVARS_IV]], 31990 31; CHECK-NEXT: br i1 [[CMP2]], label %[[FOR_BODY]], label %[[EXIT:.*]] 32; CHECK: [[EXIT]]: 33; CHECK-NEXT: ret void 34; 35entry: 36 br label %for.body 37 38for.body: 39 %0 = phi float [ 0.000000e+00, %entry ], [ %9, %for.body ] 40 %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ] 41 %1 = add nuw nsw i64 %indvars.iv, 1 42 %arrayidx = getelementptr inbounds [32000 x float], ptr @a, i64 0, i64 %1 43 %2 = load float, ptr %arrayidx, align 4 44 %arrayidx6 = getelementptr inbounds [32000 x float], ptr @a, i64 0, i64 %indvars.iv 45 %mul = fmul fast float %0, %2 46 store float %mul, ptr %arrayidx6, align 4 47 %3 = add nuw nsw i64 %indvars.iv, 2 48 %arrayidx11 = getelementptr inbounds [32000 x float], ptr @a, i64 0, i64 %3 49 %4 = load float, ptr %arrayidx11, align 4 50 %mul15 = fmul fast float %4, %2 51 store float %mul15, ptr %arrayidx, align 4 52 %5 = add nuw nsw i64 %indvars.iv, 3 53 %arrayidx21 = getelementptr inbounds [32000 x float], ptr @a, i64 0, i64 %5 54 %6 = load float, ptr %arrayidx21, align 4 55 %mul25 = fmul fast float %6, %4 56 store float %mul25, ptr %arrayidx11, align 4 57 %7 = add nuw nsw i64 %indvars.iv, 4 58 %arrayidx31 = getelementptr inbounds [32000 x float], ptr @a, i64 0, i64 %7 59 %8 = load float, ptr %arrayidx31, align 4 60 %mul35 = fmul fast float %8, %6 61 store float %mul35, ptr %arrayidx21, align 4 62 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 5 63 %arrayidx41 = getelementptr inbounds [32000 x float], ptr @a, i64 0, i64 %indvars.iv.next 64 %9 = load float, ptr %arrayidx41, align 4 65 %mul45 = fmul fast float %9, %8 66 store float %mul45, ptr %arrayidx31, align 4 67 %cmp2 = icmp ult i64 %indvars.iv, 31990 68 br i1 %cmp2, label %for.body, label %exit 69 70exit: 71 ret void 72} 73