xref: /llvm-project/llvm/test/CodeGen/X86/avx2-vperm.ll (revision 4318b033bddc64d5654f3e368fddde859ff4d02e)
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=X86
3; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=X64
4
5define <8 x i32> @perm_cl_int_8x32(<8 x i32> %A) nounwind readnone {
6; X86-LABEL: perm_cl_int_8x32:
7; X86:       # %bb.0: # %entry
8; X86-NEXT:    vmovaps {{.*#+}} ymm1 = [0,7,2,1,2,7,6,0]
9; X86-NEXT:    vpermps %ymm0, %ymm1, %ymm0
10; X86-NEXT:    retl
11;
12; X64-LABEL: perm_cl_int_8x32:
13; X64:       # %bb.0: # %entry
14; X64-NEXT:    vmovaps {{.*#+}} ymm1 = [0,7,2,1,2,7,6,0]
15; X64-NEXT:    vpermps %ymm0, %ymm1, %ymm0
16; X64-NEXT:    retq
17entry:
18  %B = shufflevector <8 x i32> %A, <8 x i32> undef, <8 x i32> <i32 0, i32 7, i32 2, i32 1, i32 2, i32 7, i32 6, i32 0>
19  ret <8 x i32> %B
20}
21
22
23define <8 x float> @perm_cl_fp_8x32(<8 x float> %A) nounwind readnone {
24; X86-LABEL: perm_cl_fp_8x32:
25; X86:       # %bb.0: # %entry
26; X86-NEXT:    vmovaps {{.*#+}} ymm1 = [u,7,2,u,4,u,1,6]
27; X86-NEXT:    vpermps %ymm0, %ymm1, %ymm0
28; X86-NEXT:    retl
29;
30; X64-LABEL: perm_cl_fp_8x32:
31; X64:       # %bb.0: # %entry
32; X64-NEXT:    vmovaps {{.*#+}} ymm1 = [u,7,2,u,4,u,1,6]
33; X64-NEXT:    vpermps %ymm0, %ymm1, %ymm0
34; X64-NEXT:    retq
35entry:
36  %B = shufflevector <8 x float> %A, <8 x float> undef, <8 x i32> <i32 undef, i32 7, i32 2, i32 undef, i32 4, i32 undef, i32 1, i32 6>
37  ret <8 x float> %B
38}
39
40define <4 x i64> @perm_cl_int_4x64(<4 x i64> %A) nounwind readnone {
41; X86-LABEL: perm_cl_int_4x64:
42; X86:       # %bb.0: # %entry
43; X86-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,1]
44; X86-NEXT:    retl
45;
46; X64-LABEL: perm_cl_int_4x64:
47; X64:       # %bb.0: # %entry
48; X64-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,1]
49; X64-NEXT:    retq
50entry:
51  %B = shufflevector <4 x i64> %A, <4 x i64> undef, <4 x i32> <i32 0, i32 3, i32 2, i32 1>
52  ret <4 x i64> %B
53}
54
55define <4 x double> @perm_cl_fp_4x64(<4 x double> %A) nounwind readnone {
56; X86-LABEL: perm_cl_fp_4x64:
57; X86:       # %bb.0: # %entry
58; X86-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,1]
59; X86-NEXT:    retl
60;
61; X64-LABEL: perm_cl_fp_4x64:
62; X64:       # %bb.0: # %entry
63; X64-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,1]
64; X64-NEXT:    retq
65entry:
66  %B = shufflevector <4 x double> %A, <4 x double> undef, <4 x i32> <i32 0, i32 3, i32 2, i32 1>
67  ret <4 x double> %B
68}
69