1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2 ; RUN: opt -slp-vectorizer -instcombine -S -mtriple=x86_64-- -mcpu=corei7 < %s | FileCheck %s
4 define void @test1(float %a, float %b, float %c, float %d, i32* nocapture %p) {
7 ; CHECK-NEXT: [[TMP0:%.*]] = insertelement <4 x float> poison, float [[A:%.*]], i32 0
8 ; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x float> [[TMP0]], float [[B:%.*]], i32 1
9 ; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x float> [[TMP1]], float [[C:%.*]], i32 2
10 ; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x float> [[TMP2]], float [[D:%.*]], i32 3
11 ; CHECK-NEXT: [[TMP4:%.*]] = fptosi <4 x float> [[TMP3]] to <4 x i32>
12 ; CHECK-NEXT: [[TMP5:%.*]] = bitcast i32* [[P:%.*]] to <4 x i32>*
13 ; CHECK-NEXT: store <4 x i32> [[TMP4]], <4 x i32>* [[TMP5]], align 4, !tbaa [[TBAA0:![0-9]+]]
14 ; CHECK-NEXT: ret void
17 %conv = fptosi float %a to i32
18 %conv1 = fptosi float %b to i32
19 %conv3 = fptosi float %c to i32
20 %conv5 = fptosi float %d to i32
21 %incdec.ptr = getelementptr inbounds i32, i32* %p, i64 1
22 store i32 %conv, i32* %p, align 4, !tbaa !2
23 %incdec.ptr8 = getelementptr inbounds i32, i32* %p, i64 2
24 store i32 %conv1, i32* %incdec.ptr, align 4, !tbaa !2
25 %incdec.ptr10 = getelementptr inbounds i32, i32* %p, i64 3
26 store i32 %conv3, i32* %incdec.ptr8, align 4, !tbaa !2
27 store i32 %conv5, i32* %incdec.ptr10, align 4, !tbaa !2
31 define void @test1_vec(float %a, float %b, float %c, float %d, <4 x i32>* nocapture %p) {
32 ; CHECK-LABEL: @test1_vec(
34 ; CHECK-NEXT: [[TMP0:%.*]] = insertelement <4 x float> poison, float [[A:%.*]], i32 0
35 ; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x float> [[TMP0]], float [[B:%.*]], i32 1
36 ; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x float> [[TMP1]], float [[C:%.*]], i32 2
37 ; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x float> [[TMP2]], float [[D:%.*]], i32 3
38 ; CHECK-NEXT: [[TMP4:%.*]] = fptosi <4 x float> [[TMP3]] to <4 x i32>
39 ; CHECK-NEXT: store <4 x i32> [[TMP4]], <4 x i32>* [[P:%.*]], align 16, !tbaa [[TBAA0]]
40 ; CHECK-NEXT: ret void
43 %conv = fptosi float %a to i32
44 %vecinit = insertelement <4 x i32> undef, i32 %conv, i32 0
45 %conv1 = fptosi float %b to i32
46 %vecinit2 = insertelement <4 x i32> %vecinit, i32 %conv1, i32 1
47 %conv3 = fptosi float %c to i32
48 %vecinit4 = insertelement <4 x i32> %vecinit2, i32 %conv3, i32 2
49 %conv5 = fptosi float %d to i32
50 %vecinit6 = insertelement <4 x i32> %vecinit4, i32 %conv5, i32 3
51 store <4 x i32> %vecinit6, <4 x i32>* %p, align 16, !tbaa !2
55 define void @test2(i32 %a, i32 %b, i32 %c, i32 %d, i32* nocapture %p) {
56 ; CHECK-LABEL: @test2(
58 ; CHECK-NEXT: [[TMP0:%.*]] = insertelement <4 x i32> poison, i32 [[A:%.*]], i32 0
59 ; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i32> [[TMP0]], i32 [[B:%.*]], i32 1
60 ; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i32> [[TMP1]], i32 [[C:%.*]], i32 2
61 ; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x i32> [[TMP2]], i32 [[D:%.*]], i32 3
62 ; CHECK-NEXT: [[TMP4:%.*]] = add nsw <4 x i32> [[TMP3]], <i32 1, i32 1, i32 1, i32 1>
63 ; CHECK-NEXT: [[TMP5:%.*]] = bitcast i32* [[P:%.*]] to <4 x i32>*
64 ; CHECK-NEXT: store <4 x i32> [[TMP4]], <4 x i32>* [[TMP5]], align 4, !tbaa [[TBAA0]]
65 ; CHECK-NEXT: ret void
68 %add = add nsw i32 %a, 1
69 %add1 = add nsw i32 %b, 1
70 %add3 = add nsw i32 %c, 1
71 %add5 = add nsw i32 %d, 1
72 %incdec.ptr = getelementptr inbounds i32, i32* %p, i64 1
73 store i32 %add, i32* %p, align 4, !tbaa !2
74 %incdec.ptr8 = getelementptr inbounds i32, i32* %p, i64 2
75 store i32 %add1, i32* %incdec.ptr, align 4, !tbaa !2
76 %incdec.ptr10 = getelementptr inbounds i32, i32* %p, i64 3
77 store i32 %add3, i32* %incdec.ptr8, align 4, !tbaa !2
78 store i32 %add5, i32* %incdec.ptr10, align 4, !tbaa !2
82 define void @test2_vec(i32 %0, i32 %1, i32 %2, i32 %3, <4 x i32>* nocapture %4) {
83 ; CHECK-LABEL: @test2_vec(
84 ; CHECK-NEXT: [[TMP6:%.*]] = insertelement <4 x i32> poison, i32 [[TMP0:%.*]], i32 0
85 ; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> [[TMP6]], i32 [[TMP1:%.*]], i32 1
86 ; CHECK-NEXT: [[TMP8:%.*]] = insertelement <4 x i32> [[TMP7]], i32 [[TMP2:%.*]], i32 2
87 ; CHECK-NEXT: [[TMP9:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP3:%.*]], i32 3
88 ; CHECK-NEXT: [[TMP10:%.*]] = add nsw <4 x i32> [[TMP9]], <i32 1, i32 1, i32 1, i32 1>
89 ; CHECK-NEXT: store <4 x i32> [[TMP10]], <4 x i32>* [[TMP4:%.*]], align 16, !tbaa [[TBAA0]]
90 ; CHECK-NEXT: ret void
92 %6 = add nsw i32 %0, 1
93 %7 = insertelement <4 x i32> undef, i32 %6, i32 0
94 %8 = add nsw i32 %1, 1
95 %9 = insertelement <4 x i32> %7, i32 %8, i32 1
96 %10 = add nsw i32 %2, 1
97 %11 = insertelement <4 x i32> %9, i32 %10, i32 2
98 %12 = add nsw i32 %3, 1
99 %13 = insertelement <4 x i32> %11, i32 %12, i32 3
100 store <4 x i32> %13, <4 x i32>* %4, align 16, !tbaa !2
104 !2 = !{!3, !3, i64 0}
105 !3 = !{!"int", !4, i64 0}
106 !4 = !{!"omnipotent char", !5, i64 0}
107 !5 = !{!"Simple C++ TBAA"}