| blob | 9c7e8f66c6c6c230d8e97581b0c5d40c3647e362 |
1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -passes=slp-vectorizer -mattr=+sse2 -S | FileCheck %s --check-prefix=SSE
3 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -passes=slp-vectorizer -mattr=+avx -S | FileCheck %s --check-prefix=AVX
4 ; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -passes=slp-vectorizer -mattr=+avx2 -S | FileCheck %s --check-prefix=AVX
6 %class.1 = type { %class.2 }
7 %class.2 = type { %"class.3" }
8 %"class.3" = type { %"struct.1", i64 }
9 %"struct.1" = type { [8 x i64] }
11 $_ZN1C10SwitchModeEv = comdat any
13 ; Function Attrs: uwtable
14 define void @_ZN1C10SwitchModeEv() local_unnamed_addr #0 comdat align 2 {
15 ; SSE-LABEL: @_ZN1C10SwitchModeEv(
16 ; SSE-NEXT: for.body.lr.ph.i:
17 ; SSE-NEXT: [[OR_1:%.*]] = or i64 undef, 1
18 ; SSE-NEXT: store i64 [[OR_1]], ptr undef, align 8
19 ; SSE-NEXT: [[FOO_3:%.*]] = load i64, ptr undef, align 8
20 ; SSE-NEXT: [[FOO_2:%.*]] = getelementptr inbounds [[CLASS_1:%.*]], ptr undef, i64 0, i32 0, i32 0, i32 0, i32 0, i64 1
21 ; SSE-NEXT: [[FOO_4:%.*]] = load i64, ptr [[FOO_2]], align 8
22 ; SSE-NEXT: [[BAR5:%.*]] = load i64, ptr undef, align 8
23 ; SSE-NEXT: [[AND_2:%.*]] = and i64 [[OR_1]], [[FOO_3]]
24 ; SSE-NEXT: [[AND_1:%.*]] = and i64 [[BAR5]], [[FOO_4]]
25 ; SSE-NEXT: store i64 [[AND_2]], ptr undef, align 8
26 ; SSE-NEXT: [[BAR4:%.*]] = getelementptr inbounds [[CLASS_2:%.*]], ptr undef, i64 0, i32 0, i32 0, i32 0, i64 1
27 ; SSE-NEXT: store i64 [[AND_1]], ptr [[BAR4]], align 8
28 ; SSE-NEXT: ret void
29 ;
30 ; AVX-LABEL: @_ZN1C10SwitchModeEv(
31 ; AVX-NEXT: for.body.lr.ph.i:
32 ; AVX-NEXT: [[OR_1:%.*]] = or i64 undef, 1
33 ; AVX-NEXT: store i64 [[OR_1]], ptr undef, align 8
34 ; AVX-NEXT: [[BAR5:%.*]] = load i64, ptr undef, align 8
35 ; AVX-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr undef, align 8
36 ; AVX-NEXT: [[TMP1:%.*]] = insertelement <2 x i64> poison, i64 [[OR_1]], i32 0
37 ; AVX-NEXT: [[TMP2:%.*]] = insertelement <2 x i64> [[TMP1]], i64 [[BAR5]], i32 1
38 ; AVX-NEXT: [[TMP3:%.*]] = and <2 x i64> [[TMP2]], [[TMP0]]
39 ; AVX-NEXT: store <2 x i64> [[TMP3]], ptr undef, align 8
40 ; AVX-NEXT: ret void
41 ;
42 for.body.lr.ph.i:
43 %or.1 = or i64 undef, 1
44 store i64 %or.1, ptr undef, align 8
45 %foo.3 = load i64, ptr undef, align 8
46 %foo.2 = getelementptr inbounds %class.1, ptr undef, i64 0, i32 0, i32 0, i32 0, i32 0, i64 1
47 %foo.4 = load i64, ptr %foo.2, align 8
48 %bar5 = load i64, ptr undef, align 8
49 %and.2 = and i64 %or.1, %foo.3
50 %and.1 = and i64 %bar5, %foo.4
51 store i64 %and.2, ptr undef, align 8
52 %bar4 = getelementptr inbounds %class.2, ptr undef, i64 0, i32 0, i32 0, i32 0, i64 1
53 store i64 %and.1, ptr %bar4, align 8
54 ret void
55 }
57 ; Function Attrs: norecurse nounwind uwtable
58 define void @pr35497() local_unnamed_addr #0 {
59 ; SSE-LABEL: @pr35497(
60 ; SSE-NEXT: entry:
61 ; SSE-NEXT: [[TMP0:%.*]] = load i64, ptr undef, align 1
62 ; SSE-NEXT: [[ADD:%.*]] = add i64 undef, undef
63 ; SSE-NEXT: store i64 [[ADD]], ptr undef, align 1
64 ; SSE-NEXT: [[ARRAYIDX2_2:%.*]] = getelementptr inbounds [0 x i64], ptr undef, i64 0, i64 4
65 ; SSE-NEXT: [[TMP1:%.*]] = insertelement <2 x i64> <i64 poison, i64 undef>, i64 [[TMP0]], i32 0
66 ; SSE-NEXT: [[TMP2:%.*]] = shl <2 x i64> [[TMP1]], <i64 2, i64 2>
67 ; SSE-NEXT: [[TMP3:%.*]] = and <2 x i64> [[TMP2]], <i64 20, i64 20>
68 ; SSE-NEXT: [[TMP4:%.*]] = shufflevector <2 x i64> [[TMP3]], <2 x i64> poison, <2 x i32> <i32 1, i32 0>
69 ; SSE-NEXT: [[TMP5:%.*]] = add nuw nsw <2 x i64> [[TMP4]], zeroinitializer
70 ; SSE-NEXT: store <2 x i64> [[TMP5]], ptr undef, align 1
71 ; SSE-NEXT: [[TMP6:%.*]] = insertelement <2 x i64> [[TMP5]], i64 [[ADD]], i32 0
72 ; SSE-NEXT: [[TMP7:%.*]] = shl <2 x i64> [[TMP6]], <i64 2, i64 2>
73 ; SSE-NEXT: [[TMP8:%.*]] = and <2 x i64> [[TMP7]], <i64 20, i64 20>
74 ; SSE-NEXT: [[TMP9:%.*]] = shufflevector <2 x i64> [[TMP8]], <2 x i64> poison, <2 x i32> <i32 1, i32 0>
75 ; SSE-NEXT: [[TMP10:%.*]] = lshr <2 x i64> [[TMP5]], <i64 6, i64 6>
76 ; SSE-NEXT: [[TMP11:%.*]] = add nuw nsw <2 x i64> [[TMP9]], [[TMP10]]
77 ; SSE-NEXT: store <2 x i64> [[TMP11]], ptr [[ARRAYIDX2_2]], align 1
78 ; SSE-NEXT: ret void
79 ;
80 ; AVX-LABEL: @pr35497(
81 ; AVX-NEXT: entry:
82 ; AVX-NEXT: [[TMP0:%.*]] = load i64, ptr undef, align 1
83 ; AVX-NEXT: [[ADD:%.*]] = add i64 undef, undef
84 ; AVX-NEXT: store i64 [[ADD]], ptr undef, align 1
85 ; AVX-NEXT: [[ARRAYIDX2_2:%.*]] = getelementptr inbounds [0 x i64], ptr undef, i64 0, i64 4
86 ; AVX-NEXT: [[TMP1:%.*]] = insertelement <2 x i64> <i64 undef, i64 poison>, i64 [[TMP0]], i32 1
87 ; AVX-NEXT: [[TMP2:%.*]] = shl <2 x i64> [[TMP1]], <i64 2, i64 2>
88 ; AVX-NEXT: [[TMP3:%.*]] = and <2 x i64> [[TMP2]], <i64 20, i64 20>
89 ; AVX-NEXT: [[TMP4:%.*]] = add nuw nsw <2 x i64> [[TMP3]], zeroinitializer
90 ; AVX-NEXT: store <2 x i64> [[TMP4]], ptr undef, align 1
91 ; AVX-NEXT: [[TMP5:%.*]] = insertelement <2 x i64> [[TMP4]], i64 [[ADD]], i32 0
92 ; AVX-NEXT: [[TMP6:%.*]] = shl <2 x i64> [[TMP5]], <i64 2, i64 2>
93 ; AVX-NEXT: [[TMP7:%.*]] = and <2 x i64> [[TMP6]], <i64 20, i64 20>
94 ; AVX-NEXT: [[TMP8:%.*]] = shufflevector <2 x i64> [[TMP7]], <2 x i64> poison, <2 x i32> <i32 1, i32 0>
95 ; AVX-NEXT: [[TMP9:%.*]] = lshr <2 x i64> [[TMP4]], <i64 6, i64 6>
96 ; AVX-NEXT: [[TMP10:%.*]] = add nuw nsw <2 x i64> [[TMP8]], [[TMP9]]
97 ; AVX-NEXT: store <2 x i64> [[TMP10]], ptr [[ARRAYIDX2_2]], align 1
98 ; AVX-NEXT: ret void
99 ;
100 entry:
101 %0 = load i64, ptr undef, align 1
102 %and = shl i64 %0, 2
103 %shl = and i64 %and, 20
104 %add = add i64 undef, undef
105 store i64 %add, ptr undef, align 1
106 %arrayidx2.1 = getelementptr inbounds [0 x i64], ptr undef, i64 0, i64 5
107 %and.1 = shl i64 undef, 2
108 %shl.1 = and i64 %and.1, 20
109 %shr.1 = lshr i64 undef, 6
110 %add.1 = add nuw nsw i64 %shl, %shr.1
111 %arrayidx2.2 = getelementptr inbounds [0 x i64], ptr undef, i64 0, i64 4
112 %shr.2 = lshr i64 undef, 6
113 %add.2 = add nuw nsw i64 %shl.1, %shr.2
114 %and.4 = shl i64 %add, 2
115 %shl.4 = and i64 %and.4, 20
116 %arrayidx2.5 = getelementptr inbounds [0 x i64], ptr undef, i64 0, i64 1
117 store i64 %add.1, ptr %arrayidx2.5, align 1
118 %and.5 = shl nuw nsw i64 %add.1, 2
119 %shl.5 = and i64 %and.5, 20
120 %shr.5 = lshr i64 %add.1, 6
121 %add.5 = add nuw nsw i64 %shl.4, %shr.5
122 store i64 %add.5, ptr %arrayidx2.1, align 1
123 store i64 %add.2, ptr undef, align 1
124 %shr.6 = lshr i64 %add.2, 6
125 %add.6 = add nuw nsw i64 %shl.5, %shr.6
126 store i64 %add.6, ptr %arrayidx2.2, align 1
127 ret void
128 }