xref: /llvm-project/llvm/test/CodeGen/X86/vec-copysign-avx512.ll (revision 9be6e7b0f2496efa76229f70654ebe6494dc2cf9)
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc < %s -mtriple=x86_64-apple-macosx10.10.0 -mattr=+avx512vl | FileCheck %s
3; RUN: llc < %s -mtriple=x86_64-apple-macosx10.10.0 -mattr=+avx512vl,+avx512dq | FileCheck %s
4
5define <4 x float> @v4f32(<4 x float> %a, <4 x float> %b) nounwind {
6; CHECK-LABEL: v4f32:
7; CHECK:       ## %bb.0:
8; CHECK-NEXT:    vpternlogd $228, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm1, %xmm0
9; CHECK-NEXT:    retq
10  %tmp = tail call <4 x float> @llvm.copysign.v4f32( <4 x float> %a, <4 x float> %b )
11  ret <4 x float> %tmp
12}
13
14define <8 x float> @v8f32(<8 x float> %a, <8 x float> %b) nounwind {
15; CHECK-LABEL: v8f32:
16; CHECK:       ## %bb.0:
17; CHECK-NEXT:    vpternlogd $228, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm1, %ymm0
18; CHECK-NEXT:    retq
19  %tmp = tail call <8 x float> @llvm.copysign.v8f32( <8 x float> %a, <8 x float> %b )
20  ret <8 x float> %tmp
21}
22
23define <16 x float> @v16f32(<16 x float> %a, <16 x float> %b) nounwind {
24; CHECK-LABEL: v16f32:
25; CHECK:       ## %bb.0:
26; CHECK-NEXT:    vpternlogd $228, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to16}, %zmm1, %zmm0
27; CHECK-NEXT:    retq
28  %tmp = tail call <16 x float> @llvm.copysign.v16f32( <16 x float> %a, <16 x float> %b )
29  ret <16 x float> %tmp
30}
31
32define <2 x double> @v2f64(<2 x double> %a, <2 x double> %b) nounwind {
33; CHECK-LABEL: v2f64:
34; CHECK:       ## %bb.0:
35; CHECK-NEXT:    vpternlogq $228, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to2}, %xmm1, %xmm0
36; CHECK-NEXT:    retq
37  %tmp = tail call <2 x double> @llvm.copysign.v2f64( <2 x double> %a, <2 x double> %b )
38  ret <2 x double> %tmp
39}
40
41define <4 x double> @v4f64(<4 x double> %a, <4 x double> %b) nounwind {
42; CHECK-LABEL: v4f64:
43; CHECK:       ## %bb.0:
44; CHECK-NEXT:    vpternlogq $228, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %ymm1, %ymm0
45; CHECK-NEXT:    retq
46  %tmp = tail call <4 x double> @llvm.copysign.v4f64( <4 x double> %a, <4 x double> %b )
47  ret <4 x double> %tmp
48}
49
50define <8 x double> @v8f64(<8 x double> %a, <8 x double> %b) nounwind {
51; CHECK-LABEL: v8f64:
52; CHECK:       ## %bb.0:
53; CHECK-NEXT:    vpternlogq $228, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %zmm1, %zmm0
54; CHECK-NEXT:    retq
55  %tmp = tail call <8 x double> @llvm.copysign.v8f64( <8 x double> %a, <8 x double> %b )
56  ret <8 x double> %tmp
57}
58
59declare <4 x float>     @llvm.copysign.v4f32(<4 x float>  %Mag, <4 x float>  %Sgn)
60declare <8 x float>     @llvm.copysign.v8f32(<8 x float>  %Mag, <8 x float>  %Sgn)
61declare <16 x float>    @llvm.copysign.v16f32(<16 x float>  %Mag, <16 x float>  %Sgn)
62declare <2 x double>    @llvm.copysign.v2f64(<2 x double> %Mag, <2 x double> %Sgn)
63declare <4 x double>    @llvm.copysign.v4f64(<4 x double> %Mag, <4 x double> %Sgn)
64declare <8 x double>    @llvm.copysign.v8f64(<8 x double> %Mag, <8 x double> %Sgn)
65
66