Debian Codegraph @

llvm/test/Transforms/VectorCombine/AArch64/shuffle-of-intrinscis.ll

raw
1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2 ; RUN: opt -mtriple=aarch64 -passes=vector-combine -S %s | FileCheck %s
3
4 define <8 x i32> @test1(<4 x i32> %0, <4 x i32> %1) {
5 ; CHECK-LABEL: @test1(
6 ; CHECK-NEXT: entry:
7 ; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <4 x i32> [[TMP0:%.*]], <4 x i32> [[TMP1:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
8 ; CHECK-NEXT: [[TMP3:%.*]] = call <8 x i32> @llvm.abs.v8i32(<8 x i32> [[TMP2]], i1 false)
9 ; CHECK-NEXT: ret <8 x i32> [[TMP3]]
10 ;
11 entry:
12 %2 = call <4 x i32> @llvm.abs.v4i32(<4 x i32> %0, i1 false)
13 %3 = call <4 x i32> @llvm.abs.v4i32(<4 x i32> %1, i1 false)
14 %4 = shufflevector <4 x i32> %2, <4 x i32> %3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
15 ret <8 x i32> %4
16 }
17
18 define <8 x i32> @test2(<4 x i32> %0, <4 x i32> %1) {
19 ; CHECK-LABEL: @test2(
20 ; CHECK-NEXT: entry:
21 ; CHECK-NEXT: [[TMP2:%.*]] = call <4 x i32> @llvm.abs.v4i32(<4 x i32> [[TMP0:%.*]], i1 true)
22 ; CHECK-NEXT: [[TMP3:%.*]] = call <4 x i32> @llvm.abs.v4i32(<4 x i32> [[TMP1:%.*]], i1 false)
23 ; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <4 x i32> [[TMP2]], <4 x i32> [[TMP3]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
24 ; CHECK-NEXT: ret <8 x i32> [[TMP4]]
25 ;
26 entry:
27 %2 = call <4 x i32> @llvm.abs.v4i32(<4 x i32> %0, i1 true)
28 %3 = call <4 x i32> @llvm.abs.v4i32(<4 x i32> %1, i1 false)
29 %4 = shufflevector <4 x i32> %2, <4 x i32> %3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
30 ret <8 x i32> %4
31 }
32
33 define <8 x i32> @test3(<4 x i32> %0, <4 x i32> %1, <4 x i32> %2, <4 x i32> %3) {
34 ; CHECK-LABEL: @test3(
35 ; CHECK-NEXT: entry:
36 ; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <4 x i32> [[TMP0:%.*]], <4 x i32> [[TMP2:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
37 ; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <4 x i32> [[TMP1:%.*]], <4 x i32> [[TMP3:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
38 ; CHECK-NEXT: [[TMP6:%.*]] = call <8 x i32> @llvm.smax.v8i32(<8 x i32> [[TMP4]], <8 x i32> [[TMP5]])
39 ; CHECK-NEXT: ret <8 x i32> [[TMP6]]
40 ;
41 entry:
42 %4 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> %0, <4 x i32> %1)
43 %5 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> %2, <4 x i32> %3)
44 %6 = shufflevector <4 x i32> %4, <4 x i32> %5, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
45 ret <8 x i32> %6
46 }
47
48 define <8 x i1> @test4(<4 x float> %0, <4 x float> %1) {
49 ; CHECK-LABEL: @test4(
50 ; CHECK-NEXT: entry:
51 ; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <4 x float> [[TMP0:%.*]], <4 x float> [[TMP1:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
52 ; CHECK-NEXT: [[TMP3:%.*]] = call <8 x i1> @llvm.is.fpclass.v8f32(<8 x float> [[TMP2]], i32 0)
53 ; CHECK-NEXT: ret <8 x i1> [[TMP3]]
54 ;
55 entry:
56 %2 = call <4 x i1> @llvm.is.fpclass.v4f32(<4 x float> %0, i32 0)
57 %3 = call <4 x i1> @llvm.is.fpclass.v4f32(<4 x float> %1, i32 0)
58 %4 = shufflevector <4 x i1> %2, <4 x i1> %3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
59 ret <8 x i1> %4
60 }
61
62 define <2 x i1> @test4b(<4 x float> %0, <4 x float> %1) {
63 ; CHECK-LABEL: @test4b(
64 ; CHECK-NEXT: entry:
65 ; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <4 x float> [[TMP0:%.*]], <4 x float> [[TMP1:%.*]], <2 x i32> <i32 0, i32 4>
66 ; CHECK-NEXT: [[TMP3:%.*]] = call <2 x i1> @llvm.is.fpclass.v2f32(<2 x float> [[TMP2]], i32 0)
67 ; CHECK-NEXT: ret <2 x i1> [[TMP3]]
68 ;
69 entry:
70 %2 = call <4 x i1> @llvm.is.fpclass.v4f32(<4 x float> %0, i32 0)
71 %3 = call <4 x i1> @llvm.is.fpclass.v4f32(<4 x float> %1, i32 0)
72 %4 = shufflevector <4 x i1> %2, <4 x i1> %3, <2 x i32> <i32 0, i32 4>
73 ret <2 x i1> %4
74 }
75
76 define <8 x float> @test5(<4 x float> %0, i32 %1, <4 x float> %2, <4 x i32> %3) {
77 ; CHECK-LABEL: @test5(
78 ; CHECK-NEXT: entry:
79 ; CHECK-NEXT: [[TMP4:%.*]] = call <4 x float> @llvm.powi.v4f32.i32(<4 x float> [[TMP0:%.*]], i32 [[TMP1:%.*]])
80 ; CHECK-NEXT: [[TMP5:%.*]] = call <4 x float> @llvm.powi.v4f32.v4i32(<4 x float> [[TMP2:%.*]], <4 x i32> [[TMP3:%.*]])
81 ; CHECK-NEXT: [[TMP6:%.*]] = shufflevector <4 x float> [[TMP4]], <4 x float> [[TMP5]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
82 ; CHECK-NEXT: ret <8 x float> [[TMP6]]
83 ;
84 entry:
85 %4 = call <4 x float> @llvm.powi.v4f32.i32(<4 x float> %0, i32 %1)
86 %5 = call <4 x float> @llvm.powi.v4f32.v4i32(<4 x float> %2, <4 x i32> %3)
87 %6 = shufflevector <4 x float> %4, <4 x float> %5, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
88 ret <8 x float> %6
89 }
90
91 define <2 x float> @test6(<4 x float> %a1, <4 x float> %b1, <4 x float> %c1, <4 x float> %a2, <4 x float> %b2, <4 x float> %c2) {
92 ; CHECK-LABEL: @test6(
93 ; CHECK-NEXT: [[F1:%.*]] = call <4 x float> @llvm.fma.v4f32(<4 x float> [[A1:%.*]], <4 x float> [[B1:%.*]], <4 x float> [[C1:%.*]])
94 ; CHECK-NEXT: [[F2:%.*]] = call <4 x float> @llvm.fma.v4f32(<4 x float> [[A2:%.*]], <4 x float> [[B2:%.*]], <4 x float> [[C2:%.*]])
95 ; CHECK-NEXT: [[S:%.*]] = shufflevector <4 x float> [[F1]], <4 x float> [[F2]], <2 x i32> <i32 0, i32 4>
96 ; CHECK-NEXT: ret <2 x float> [[S]]
97 ;
98 %f1 = call <4 x float> @llvm.fma.v4f32(<4 x float> %a1, <4 x float> %b1, <4 x float> %c1)
99 %f2 = call <4 x float> @llvm.fma.v4f32(<4 x float> %a2, <4 x float> %b2, <4 x float> %c2)
100 %s = shufflevector <4 x float> %f1, <4 x float> %f2, <2 x i32> <i32 0, i32 4>
101 ret <2 x float> %s
102 }
103
104
105 declare <4 x i32> @llvm.abs.v4i32(<4 x i32>, i1)
106 declare <4 x i32> @llvm.smax.v4i32(<4 x i32>, <4 x i32>)
107 declare <4 x i1> @llvm.is.fpclass.v4f32(<4 x float>, i32)
108 declare <4 x float> @llvm.powi.v4f32.i32(<4 x float>, i32)
109 declare <4 x float> @llvm.powi.v4f32.v4i32(<4 x float>, <4 x i32>)