1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: opt < %s -S -mtriple=x86_64-unknown -mattr=+avx -slp-vectorizer | FileCheck %s
3
4
5
6define i32 @jumbled-load(i32* noalias nocapture %in, i32* noalias nocapture %inn, i32* noalias nocapture %out) {
7; CHECK-LABEL: @jumbled-load(
8; CHECK-NEXT:    [[IN_ADDR:%.*]] = getelementptr inbounds i32, i32* [[IN:%.*]], i64 0
9; CHECK-NEXT:    [[GEP_1:%.*]] = getelementptr inbounds i32, i32* [[IN_ADDR]], i64 1
10; CHECK-NEXT:    [[GEP_2:%.*]] = getelementptr inbounds i32, i32* [[IN_ADDR]], i64 2
11; CHECK-NEXT:    [[GEP_3:%.*]] = getelementptr inbounds i32, i32* [[IN_ADDR]], i64 3
12; CHECK-NEXT:    [[TMP1:%.*]] = bitcast i32* [[IN_ADDR]] to <4 x i32>*
13; CHECK-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* [[TMP1]], align 4
14; CHECK-NEXT:    [[INN_ADDR:%.*]] = getelementptr inbounds i32, i32* [[INN:%.*]], i64 0
15; CHECK-NEXT:    [[GEP_4:%.*]] = getelementptr inbounds i32, i32* [[INN_ADDR]], i64 1
16; CHECK-NEXT:    [[GEP_5:%.*]] = getelementptr inbounds i32, i32* [[INN_ADDR]], i64 2
17; CHECK-NEXT:    [[GEP_6:%.*]] = getelementptr inbounds i32, i32* [[INN_ADDR]], i64 3
18; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[INN_ADDR]] to <4 x i32>*
19; CHECK-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* [[TMP3]], align 4
20; CHECK-NEXT:    [[TMP5:%.*]] = mul <4 x i32> [[TMP2]], [[TMP4]]
21; CHECK-NEXT:    [[GEP_7:%.*]] = getelementptr inbounds i32, i32* [[OUT:%.*]], i64 0
22; CHECK-NEXT:    [[GEP_8:%.*]] = getelementptr inbounds i32, i32* [[OUT]], i64 1
23; CHECK-NEXT:    [[GEP_9:%.*]] = getelementptr inbounds i32, i32* [[OUT]], i64 2
24; CHECK-NEXT:    [[GEP_10:%.*]] = getelementptr inbounds i32, i32* [[OUT]], i64 3
25; CHECK-NEXT:    [[REORDER_SHUFFLE:%.*]] = shufflevector <4 x i32> [[TMP5]], <4 x i32> undef, <4 x i32> <i32 1, i32 3, i32 0, i32 2>
26; CHECK-NEXT:    [[TMP6:%.*]] = bitcast i32* [[GEP_7]] to <4 x i32>*
27; CHECK-NEXT:    store <4 x i32> [[REORDER_SHUFFLE]], <4 x i32>* [[TMP6]], align 4
28; CHECK-NEXT:    ret i32 undef
29;
30  %in.addr = getelementptr inbounds i32, i32* %in, i64 0
31  %load.1 = load i32, i32* %in.addr, align 4
32  %gep.1 = getelementptr inbounds i32, i32* %in.addr, i64 1
33  %load.2 = load i32, i32* %gep.1, align 4
34  %gep.2 = getelementptr inbounds i32, i32* %in.addr, i64 2
35  %load.3 = load i32, i32* %gep.2, align 4
36  %gep.3 = getelementptr inbounds i32, i32* %in.addr, i64 3
37  %load.4 = load i32, i32* %gep.3, align 4
38  %inn.addr = getelementptr inbounds i32, i32* %inn, i64 0
39  %load.5 = load i32, i32* %inn.addr, align 4
40  %gep.4 = getelementptr inbounds i32, i32* %inn.addr, i64 1
41  %load.6 = load i32, i32* %gep.4, align 4
42  %gep.5 = getelementptr inbounds i32, i32* %inn.addr, i64 2
43  %load.7 = load i32, i32* %gep.5, align 4
44  %gep.6 = getelementptr inbounds i32, i32* %inn.addr, i64 3
45  %load.8 = load i32, i32* %gep.6, align 4
46  %mul.1 = mul i32 %load.1, %load.5
47  %mul.2 = mul i32 %load.2, %load.6
48  %mul.3 = mul i32 %load.3, %load.7
49  %mul.4 = mul i32 %load.4, %load.8
50  %gep.7 = getelementptr inbounds i32, i32* %out, i64 0
51  %gep.8 = getelementptr inbounds i32, i32* %out, i64 1
52  %gep.9 = getelementptr inbounds i32, i32* %out, i64 2
53  %gep.10 = getelementptr inbounds i32, i32* %out, i64 3
54  store i32 %mul.1, i32* %gep.9, align 4
55  store i32 %mul.2, i32* %gep.7, align 4
56  store i32 %mul.3, i32* %gep.10, align 4
57  store i32 %mul.4, i32* %gep.8, align 4
58
59  ret i32 undef
60}
61