xref: /llvm-project/llvm/test/Transforms/SLPVectorizer/X86/broadcast_long.ll (revision 8b56da5e9f3ba737a5ff4bf5dee654416849042f)
1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: opt -passes=slp-vectorizer -S -mtriple=x86_64-unknown-linux -mcpu=corei7-avx -pass-remarks-output=%t < %s | FileCheck %s
3; RUN: FileCheck %s --input-file=%t --check-prefix=YAML
4
5; YAML-LABEL: --- !Passed
6; YAML-NEXT: Pass:            slp-vectorizer
7; YAML-NEXT: Name:            StoresVectorized
8; YAML-NEXT: Function:        bcast_long
9; YAML-NEXT: Args:
10; YAML-NEXT:  - String:          'Stores SLP vectorized with cost '
11; YAML-NEXT:  - Cost:            '-4'
12; YAML-NEXT:  - String:          ' and with tree size '
13; YAML-NEXT:  - TreeSize:        '2'
14
15define void @bcast_long(ptr %A, ptr %S) {
16; CHECK-LABEL: @bcast_long(
17; CHECK-NEXT:  entry:
18; CHECK-NEXT:    [[A0:%.*]] = load i32, ptr [[A:%.*]], align 8
19; CHECK-NEXT:    [[TMP0:%.*]] = insertelement <8 x i32> poison, i32 [[A0]], i32 0
20; CHECK-NEXT:    [[SHUFFLE:%.*]] = shufflevector <8 x i32> [[TMP0]], <8 x i32> poison, <8 x i32> <i32 0, i32 0, i32 poison, i32 0, i32 0, i32 0, i32 0, i32 0>
21; CHECK-NEXT:    [[TMP1:%.*]] = freeze <8 x i32> [[SHUFFLE]]
22; CHECK-NEXT:    store <8 x i32> [[TMP1]], ptr [[S:%.*]], align 8
23; CHECK-NEXT:    ret void
24;
25entry:
26  %A0 = load i32, ptr %A, align 8
27
28  %idxS1 = getelementptr inbounds i32, ptr %S, i64 1
29  %idxS2 = getelementptr inbounds i32, ptr %S, i64 2
30  %idxS3 = getelementptr inbounds i32, ptr %S, i64 3
31  %idxS4 = getelementptr inbounds i32, ptr %S, i64 4
32  %idxS5 = getelementptr inbounds i32, ptr %S, i64 5
33  %idxS6 = getelementptr inbounds i32, ptr %S, i64 6
34  %idxS7 = getelementptr inbounds i32, ptr %S, i64 7
35
36  store i32 %A0, ptr %S, align 8
37  store i32 %A0, ptr %idxS1, align 8
38  store i32 undef, ptr %idxS2, align 8
39  store i32 %A0, ptr %idxS3, align 8
40  store i32 %A0, ptr %idxS4, align 8
41  store i32 %A0, ptr %idxS5, align 8
42  store i32 %A0, ptr %idxS6, align 8
43  store i32 %A0, ptr %idxS7, align 8
44  ret void
45}
46