xref: /llvm-project/llvm/test/Transforms/SLPVectorizer/insert-shuffle.ll (revision 15ee17c3ce34623261788d7de3c1bdf5860be34e)
1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: %if x86-registered-target %{ opt -passes=slp-vectorizer -S -o - -mtriple=x86_64-unknown-linux-gnu < %s | FileCheck %s %}
3; RUN: %if aarch64-registered-target %{ opt -passes=slp-vectorizer -S -o - -mtriple=aarch64-unknown-linux-gnu < %s | FileCheck %s %}
4
5%struct.sw = type { float, float, float, float }
6
7define { <2 x float>, <2 x float> } @foo(ptr %v) {
8; CHECK-LABEL: @foo(
9; CHECK-NEXT:  entry:
10; CHECK-NEXT:    [[TMP0:%.*]] = load float, ptr undef, align 4
11; CHECK-NEXT:    [[TMP1:%.*]] = load float, ptr undef, align 4
12; CHECK-NEXT:    [[TMP3:%.*]] = load <2 x float>, ptr [[V:%.*]], align 16
13; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <2 x float> [[TMP3]], <2 x float> poison, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
14; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <4 x float> <float undef, float poison, float poison, float undef>, float [[TMP0]], i32 1
15; CHECK-NEXT:    [[TMP5:%.*]] = insertelement <4 x float> [[TMP4]], float [[TMP1]], i32 2
16; CHECK-NEXT:    [[TMP6:%.*]] = fmul <4 x float> [[SHUFFLE]], [[TMP5]]
17; CHECK-NEXT:    [[TMP7:%.*]] = fadd <4 x float> [[TMP6]], undef
18; CHECK-NEXT:    [[TMP8:%.*]] = fadd <4 x float> [[TMP7]], undef
19; CHECK-NEXT:    [[TMP9:%.*]] = fadd <4 x float> [[TMP8]], undef
20; CHECK-NEXT:    [[TMP10:%.*]] = shufflevector <4 x float> [[TMP9]], <4 x float> poison, <2 x i32> <i32 1, i32 0>
21; CHECK-NEXT:    [[TMP11:%.*]] = shufflevector <4 x float> [[TMP9]], <4 x float> poison, <2 x i32> <i32 2, i32 3>
22; CHECK-NEXT:    [[INS1:%.*]] = insertvalue { <2 x float>, <2 x float> } undef, <2 x float> [[TMP10]], 0
23; CHECK-NEXT:    [[INS2:%.*]] = insertvalue { <2 x float>, <2 x float> } [[INS1]], <2 x float> [[TMP11]], 1
24; CHECK-NEXT:    ret { <2 x float>, <2 x float> } [[INS2]]
25;
26entry:
27  %0 = load float, ptr undef, align 4
28  %1 = load float, ptr %v, align 16
29  %y = getelementptr inbounds %struct.sw, ptr %v, i64 0, i32 1
30  %2 = load float, ptr %y, align 4
31  %mul3 = fmul float %0, %2
32  %add = fadd float undef, %mul3
33  %add6 = fadd float %add, undef
34  %add9 = fadd float %add6, undef
35  %mul12 = fmul float %1, undef
36  %add16 = fadd float %mul12, undef
37  %add20 = fadd float undef, %add16
38  %add24 = fadd float undef, %add20
39  %3 = load float, ptr undef, align 4
40  %mul27 = fmul float %1, %3
41  %add31 = fadd float %mul27, undef
42  %add35 = fadd float undef, %add31
43  %add39 = fadd float undef, %add35
44  %mul45 = fmul float %2, undef
45  %add46 = fadd float undef, %mul45
46  %add50 = fadd float undef, %add46
47  %add54 = fadd float undef, %add50
48  %vec1 = insertelement <2 x float> undef, float %add9, i32 0
49  %vec2 = insertelement <2 x float> %vec1, float %add24, i32 1
50  %vec3 = insertelement <2 x float> undef, float %add39, i32 0
51  %vec4 = insertelement <2 x float> %vec3, float %add54, i32 1
52  %ins1 = insertvalue { <2 x float>, <2 x float> } undef, <2 x float> %vec2, 0
53  %ins2 = insertvalue { <2 x float>, <2 x float> } %ins1, <2 x float> %vec4, 1
54  ret { <2 x float>, <2 x float> } %ins2
55}
56