[llvm-project.git] / llvm / test / Transforms / SLPVectorizer / AArch64 / shuffle-vectors-mask-size.ll
| blob | 4f8818237462280fa0d31d7347178298c21abac7 |
1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3
2 ; RUN: opt -S -passes=slp-vectorizer -mtriple=aarch64-unknown-linux-gnu < %s | FileCheck %s
4 define void @p(double %0) {
5 ; CHECK-LABEL: define void @p(
6 ; CHECK-SAME: double [[TMP0:%.*]]) {
7 ; CHECK-NEXT: entry:
8 ; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x double> <double 0.000000e+00, double 0.000000e+00, double 0.000000e+00, double poison>, double [[TMP0]], i32 3
9 ; CHECK-NEXT: [[TMP2:%.*]] = fmul <4 x double> [[TMP1]], zeroinitializer
10 ; CHECK-NEXT: [[TMP3:%.*]] = fadd <4 x double> zeroinitializer, [[TMP2]]
11 ; CHECK-NEXT: [[TMP4:%.*]] = fadd <4 x double> [[TMP3]], zeroinitializer
12 ; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <4 x double> [[TMP2]], <4 x double> [[TMP3]], <2 x i32> <i32 1, i32 7>
13 ; CHECK-NEXT: [[TMP6:%.*]] = fadd <2 x double> zeroinitializer, [[TMP5]]
14 ; CHECK-NEXT: [[TMP7:%.*]] = fmul <2 x double> [[TMP6]], zeroinitializer
15 ; CHECK-NEXT: [[TMP8:%.*]] = fmul <4 x double> [[TMP4]], zeroinitializer
16 ; CHECK-NEXT: [[TMP10:%.*]] = call <4 x double> @llvm.vector.insert.v4f64.v2f64(<4 x double> <double 0.000000e+00, double 0.000000e+00, double poison, double poison>, <2 x double> [[TMP7]], i64 2)
17 ; CHECK-NEXT: [[TMP11:%.*]] = fadd <4 x double> [[TMP8]], [[TMP10]]
18 ; CHECK-NEXT: [[TMP12:%.*]] = fadd <4 x double> [[TMP11]], zeroinitializer
19 ; CHECK-NEXT: [[TMP13:%.*]] = fptosi <4 x double> [[TMP12]] to <4 x i32>
20 ; CHECK-NEXT: store <4 x i32> [[TMP13]], ptr null, align 4
21 ; CHECK-NEXT: ret void
22 ;
23 entry:
24 %mul16.1.i = fmul double %0, 0.000000e+00
25 %add21.1.i = fadd double %mul16.1.i, 0.000000e+00
26 %add21.2.i = fadd double %add21.1.i, 0.000000e+00
27 %mul16.150.i = fmul double 0.000000e+00, 0.000000e+00
28 %add21.1.1.i = fadd double 0.000000e+00, %mul16.150.i
29 %add21.2.1.i = fadd double %add21.1.1.i, 0.000000e+00
30 %mul16.252.i = fmul double 0.000000e+00, 0.000000e+00
31 %add21.1.2.i = fadd double 0.000000e+00, %mul16.252.i
32 %add21.2.2.i = fadd double %add21.1.2.i, 0.000000e+00
33 %add21.2.165.i = fadd double %add21.1.i, 0.000000e+00
34 %mul16.150.1.i = fmul double 0.000000e+00, 0.000000e+00
35 %add21.1.1.1.i = fadd double %mul16.150.1.i, 0.000000e+00
36 %add21.2.1.1.i = fadd double %add21.1.1.1.i, 0.000000e+00
37 %add21.2.2.1.i = fadd double 0.000000e+00, %mul16.150.1.i
38 %mul16.1.1.i36 = fmul double %add21.2.1.1.i, 0.000000e+00
39 %add21.1.1.i37 = fadd double 0.000000e+00, %mul16.1.1.i36
40 %add21.2.1.i40 = fadd double %add21.1.1.i37, 0.000000e+00
41 %mul16.252.i43 = fmul double %add21.2.2.i, 0.000000e+00
42 %mul16.1.2.i45 = fmul double %add21.2.2.1.i, 0.000000e+00
43 %add21.1.2.i46 = fadd double %mul16.252.i43, %mul16.1.2.i45
44 %add21.2.2.i49 = fadd double %add21.1.2.i46, 0.000000e+00
45 %mul16.157.i51 = fmul double %add21.2.i, 0.000000e+00
46 %mul16.1.160.i52 = fmul double %add21.2.165.i, 0.000000e+00
47 %add21.1.161.i53 = fadd double %mul16.157.i51, %mul16.1.160.i52
48 %add21.2.165.i56 = fadd double %add21.1.161.i53, 0.000000e+00
49 %mul16.150.1.i58 = fmul double %add21.2.1.i, 0.000000e+00
50 %add21.1.1.1.i60 = fadd double %mul16.150.1.i58, 0.000000e+00
51 %add21.2.1.1.i62 = fadd double %add21.1.1.1.i60, 0.000000e+00
52 %conv14.1 = fptosi double %add21.2.1.i40 to i32
53 %arrayidx16.1 = getelementptr i32, ptr null, i64 1
54 store i32 %conv14.1, ptr %arrayidx16.1, align 4
55 %conv14.2 = fptosi double %add21.2.2.i49 to i32
56 %arrayidx16.2 = getelementptr i32, ptr null, i64 2
57 store i32 %conv14.2, ptr %arrayidx16.2, align 4
58 %conv14.3 = fptosi double %add21.2.165.i56 to i32
59 %arrayidx16.3 = getelementptr i32, ptr null, i64 3
60 store i32 %conv14.3, ptr %arrayidx16.3, align 4
61 %conv14.4 = fptosi double %add21.2.1.1.i62 to i32
62 store i32 %conv14.4, ptr null, align 4
63 ret void
64 }