| blob | bafa33ff9a1c8ae3225e6864bcede59b2eed1ddb |
1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512vl,+gfni,+avx512bw --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X86BW
3 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+gfni,+avx512bw --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X64BW
4 ; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512vl,+gfni --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X86NOBW
5 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+gfni --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X64NOBW
7 declare <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8>, <16 x i8>, i8)
8 define { <16 x i8>, <16 x i8>, <16 x i8> } @test_vgf2p8affineinvqb_128(<16 x i8> %src1, <16 x i8> %src2, <16 x i8> %passthru, i16 %mask) {
9 ; X86BW-LABEL: test_vgf2p8affineinvqb_128:
10 ; X86BW: # %bb.0:
11 ; X86BW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
12 ; X86BW-NEXT: vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xd9,0x03]
13 ; X86BW-NEXT: vgf2p8affineinvqb $4, %xmm1, %xmm0, %xmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0x89,0xcf,0xe1,0x04]
14 ; X86BW-NEXT: vgf2p8affineinvqb $5, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0xcf,0xd1,0x05]
15 ; X86BW-NEXT: vmovdqa %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc3]
16 ; X86BW-NEXT: vmovdqa %xmm4, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcc]
17 ; X86BW-NEXT: retl # encoding: [0xc3]
18 ;
19 ; X64BW-LABEL: test_vgf2p8affineinvqb_128:
20 ; X64BW: # %bb.0:
21 ; X64BW-NEXT: kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
22 ; X64BW-NEXT: vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xd9,0x03]
23 ; X64BW-NEXT: vgf2p8affineinvqb $4, %xmm1, %xmm0, %xmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0x89,0xcf,0xe1,0x04]
24 ; X64BW-NEXT: vgf2p8affineinvqb $5, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0xcf,0xd1,0x05]
25 ; X64BW-NEXT: vmovdqa %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc3]
26 ; X64BW-NEXT: vmovdqa %xmm4, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcc]
27 ; X64BW-NEXT: retq # encoding: [0xc3]
28 ;
29 ; X86NOBW-LABEL: test_vgf2p8affineinvqb_128:
30 ; X86NOBW: # %bb.0:
31 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
32 ; X86NOBW-NEXT: vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xd9,0x03]
33 ; X86NOBW-NEXT: vgf2p8affineinvqb $4, %xmm1, %xmm0, %xmm4 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xe1,0x04]
34 ; X86NOBW-NEXT: vgf2p8affineinvqb $5, %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xc1,0x05]
35 ; X86NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
36 ; X86NOBW-NEXT: vpmovdb %zmm1, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xcd]
37 ; X86NOBW-NEXT: vpand %xmm4, %xmm5, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xd1,0xdb,0xcc]
38 ; X86NOBW-NEXT: vpternlogq $184, %xmm0, %xmm5, %xmm2 # encoding: [0x62,0xf3,0xd5,0x08,0x25,0xd0,0xb8]
39 ; X86NOBW-NEXT: vmovdqa %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc3]
40 ; X86NOBW-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77]
41 ; X86NOBW-NEXT: retl # encoding: [0xc3]
42 ;
43 ; X64NOBW-LABEL: test_vgf2p8affineinvqb_128:
44 ; X64NOBW: # %bb.0:
45 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
46 ; X64NOBW-NEXT: vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xd9,0x03]
47 ; X64NOBW-NEXT: vgf2p8affineinvqb $4, %xmm1, %xmm0, %xmm4 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xe1,0x04]
48 ; X64NOBW-NEXT: vgf2p8affineinvqb $5, %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xc1,0x05]
49 ; X64NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
50 ; X64NOBW-NEXT: vpmovdb %zmm1, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xcd]
51 ; X64NOBW-NEXT: vpand %xmm4, %xmm5, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xd1,0xdb,0xcc]
52 ; X64NOBW-NEXT: vpternlogq $184, %xmm0, %xmm5, %xmm2 # encoding: [0x62,0xf3,0xd5,0x08,0x25,0xd0,0xb8]
53 ; X64NOBW-NEXT: vmovdqa %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc3]
54 ; X64NOBW-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77]
55 ; X64NOBW-NEXT: retq # encoding: [0xc3]
56 %1 = bitcast i16 %mask to <16 x i1>
57 %2 = call <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 3)
58 %3 = call <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 4)
59 %4 = call <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 5)
60 %5 = select <16 x i1> %1, <16 x i8> %3, <16 x i8> zeroinitializer
61 %6 = select <16 x i1> %1, <16 x i8> %4, <16 x i8> %passthru
62 %7 = insertvalue { <16 x i8>, <16 x i8>, <16 x i8> } undef, <16 x i8> %2, 0
63 %8 = insertvalue { <16 x i8>, <16 x i8>, <16 x i8> } %7, <16 x i8> %5, 1
64 %9 = insertvalue { <16 x i8>, <16 x i8>, <16 x i8> } %8, <16 x i8> %6, 2
65 ret { <16 x i8>, <16 x i8>, <16 x i8> } %9
66 }
68 declare <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8>, <32 x i8>, i8)
69 define { <32 x i8>, <32 x i8>, <32 x i8> } @test_vgf2p8affineinvqb_256(<32 x i8> %src1, <32 x i8> %src2, <32 x i8> %passthru, i32 %mask) {
70 ; X86BW-LABEL: test_vgf2p8affineinvqb_256:
71 ; X86BW: # %bb.0:
72 ; X86BW-NEXT: kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x04]
73 ; X86BW-NEXT: vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xd9,0x03]
74 ; X86BW-NEXT: vgf2p8affineinvqb $4, %ymm1, %ymm0, %ymm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0xcf,0xe1,0x04]
75 ; X86BW-NEXT: vgf2p8affineinvqb $5, %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0xcf,0xd1,0x05]
76 ; X86BW-NEXT: vmovdqa %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc3]
77 ; X86BW-NEXT: vmovdqa %ymm4, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcc]
78 ; X86BW-NEXT: retl # encoding: [0xc3]
79 ;
80 ; X64BW-LABEL: test_vgf2p8affineinvqb_256:
81 ; X64BW: # %bb.0:
82 ; X64BW-NEXT: kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
83 ; X64BW-NEXT: vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xd9,0x03]
84 ; X64BW-NEXT: vgf2p8affineinvqb $4, %ymm1, %ymm0, %ymm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0xcf,0xe1,0x04]
85 ; X64BW-NEXT: vgf2p8affineinvqb $5, %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0xcf,0xd1,0x05]
86 ; X64BW-NEXT: vmovdqa %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc3]
87 ; X64BW-NEXT: vmovdqa %ymm4, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcc]
88 ; X64BW-NEXT: retq # encoding: [0xc3]
89 ;
90 ; X86NOBW-LABEL: test_vgf2p8affineinvqb_256:
91 ; X86NOBW: # %bb.0:
92 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
93 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k2 # encoding: [0xc5,0xf8,0x90,0x54,0x24,0x06]
94 ; X86NOBW-NEXT: vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xd9,0x03]
95 ; X86NOBW-NEXT: vgf2p8affineinvqb $4, %ymm1, %ymm0, %ymm4 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xe1,0x04]
96 ; X86NOBW-NEXT: vgf2p8affineinvqb $5, %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xc1,0x05]
97 ; X86NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
98 ; X86NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
99 ; X86NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k2} {z} # encoding: [0x62,0xf3,0x55,0xca,0x25,0xed,0xff]
100 ; X86NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
101 ; X86NOBW-NEXT: vinserti128 $1, %xmm5, %ymm1, %ymm5 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xed,0x01]
102 ; X86NOBW-NEXT: vpand %ymm4, %ymm5, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xd5,0xdb,0xcc]
103 ; X86NOBW-NEXT: vpternlogq $184, %ymm0, %ymm5, %ymm2 # encoding: [0x62,0xf3,0xd5,0x28,0x25,0xd0,0xb8]
104 ; X86NOBW-NEXT: vmovdqa %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc3]
105 ; X86NOBW-NEXT: retl # encoding: [0xc3]
106 ;
107 ; X64NOBW-LABEL: test_vgf2p8affineinvqb_256:
108 ; X64NOBW: # %bb.0:
109 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
110 ; X64NOBW-NEXT: shrl $16, %edi # encoding: [0xc1,0xef,0x10]
111 ; X64NOBW-NEXT: kmovw %edi, %k2 # encoding: [0xc5,0xf8,0x92,0xd7]
112 ; X64NOBW-NEXT: vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xd9,0x03]
113 ; X64NOBW-NEXT: vgf2p8affineinvqb $4, %ymm1, %ymm0, %ymm4 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xe1,0x04]
114 ; X64NOBW-NEXT: vgf2p8affineinvqb $5, %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xc1,0x05]
115 ; X64NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
116 ; X64NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
117 ; X64NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k2} {z} # encoding: [0x62,0xf3,0x55,0xca,0x25,0xed,0xff]
118 ; X64NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
119 ; X64NOBW-NEXT: vinserti128 $1, %xmm5, %ymm1, %ymm5 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xed,0x01]
120 ; X64NOBW-NEXT: vpand %ymm4, %ymm5, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xd5,0xdb,0xcc]
121 ; X64NOBW-NEXT: vpternlogq $184, %ymm0, %ymm5, %ymm2 # encoding: [0x62,0xf3,0xd5,0x28,0x25,0xd0,0xb8]
122 ; X64NOBW-NEXT: vmovdqa %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc3]
123 ; X64NOBW-NEXT: retq # encoding: [0xc3]
124 %1 = bitcast i32 %mask to <32 x i1>
125 %2 = call <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 3)
126 %3 = call <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 4)
127 %4 = call <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 5)
128 %5 = select <32 x i1> %1, <32 x i8> %3, <32 x i8> zeroinitializer
129 %6 = select <32 x i1> %1, <32 x i8> %4, <32 x i8> %passthru
130 %7 = insertvalue { <32 x i8>, <32 x i8>, <32 x i8> } undef, <32 x i8> %2, 0
131 %8 = insertvalue { <32 x i8>, <32 x i8>, <32 x i8> } %7, <32 x i8> %5, 1
132 %9 = insertvalue { <32 x i8>, <32 x i8>, <32 x i8> } %8, <32 x i8> %6, 2
133 ret { <32 x i8>, <32 x i8>, <32 x i8> } %9
134 }
136 declare <64 x i8> @llvm.x86.vgf2p8affineinvqb.512(<64 x i8>, <64 x i8>, i8)
137 define { <64 x i8>, <64 x i8>, <64 x i8> } @test_vgf2p8affineinvqb_512(<64 x i8> %src1, <64 x i8> %src2, <64 x i8> %passthru, i64 %mask) {
138 ; X86BW-LABEL: test_vgf2p8affineinvqb_512:
139 ; X86BW: # %bb.0:
140 ; X86BW-NEXT: kmovq {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf8,0x90,0x4c,0x24,0x04]
141 ; X86BW-NEXT: vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0xcf,0xd9,0x03]
142 ; X86BW-NEXT: vgf2p8affineinvqb $4, %zmm1, %zmm0, %zmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0xcf,0xe1,0x04]
143 ; X86BW-NEXT: vgf2p8affineinvqb $5, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0xcf,0xd1,0x05]
144 ; X86BW-NEXT: vmovdqa64 %zmm3, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc3]
145 ; X86BW-NEXT: vmovdqa64 %zmm4, %zmm1 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xcc]
146 ; X86BW-NEXT: retl # encoding: [0xc3]
147 ;
148 ; X64BW-LABEL: test_vgf2p8affineinvqb_512:
149 ; X64BW: # %bb.0:
150 ; X64BW-NEXT: kmovq %rdi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
151 ; X64BW-NEXT: vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0xcf,0xd9,0x03]
152 ; X64BW-NEXT: vgf2p8affineinvqb $4, %zmm1, %zmm0, %zmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0xcf,0xe1,0x04]
153 ; X64BW-NEXT: vgf2p8affineinvqb $5, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0xcf,0xd1,0x05]
154 ; X64BW-NEXT: vmovdqa64 %zmm3, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc3]
155 ; X64BW-NEXT: vmovdqa64 %zmm4, %zmm1 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xcc]
156 ; X64BW-NEXT: retq # encoding: [0xc3]
157 ;
158 ; X86NOBW-LABEL: test_vgf2p8affineinvqb_512:
159 ; X86NOBW: # %bb.0:
160 ; X86NOBW-NEXT: vmovdqa64 %zmm0, %zmm3 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xd8]
161 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k2 # encoding: [0xc5,0xf8,0x90,0x54,0x24,0x04]
162 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x06]
163 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k3 # encoding: [0xc5,0xf8,0x90,0x5c,0x24,0x0a]
164 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k4 # encoding: [0xc5,0xf8,0x90,0x64,0x24,0x08]
165 ; X86NOBW-NEXT: vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf3,0xfd,0x48,0xcf,0xc1,0x03]
166 ; X86NOBW-NEXT: vgf2p8affineinvqb $4, %zmm1, %zmm3, %zmm4 # encoding: [0x62,0xf3,0xe5,0x48,0xcf,0xe1,0x04]
167 ; X86NOBW-NEXT: vgf2p8affineinvqb $5, %zmm1, %zmm3, %zmm3 # encoding: [0x62,0xf3,0xe5,0x48,0xcf,0xd9,0x05]
168 ; X86NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k4} {z} # encoding: [0x62,0xf3,0x75,0xcc,0x25,0xc9,0xff]
169 ; X86NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
170 ; X86NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k3} {z} # encoding: [0x62,0xf3,0x55,0xcb,0x25,0xed,0xff]
171 ; X86NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
172 ; X86NOBW-NEXT: vinserti128 $1, %xmm5, %ymm1, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xcd,0x01]
173 ; X86NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k2} {z} # encoding: [0x62,0xf3,0x55,0xca,0x25,0xed,0xff]
174 ; X86NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
175 ; X86NOBW-NEXT: vpternlogd $255, %zmm6, %zmm6, %zmm6 {%k1} {z} # encoding: [0x62,0xf3,0x4d,0xc9,0x25,0xf6,0xff]
176 ; X86NOBW-NEXT: vpmovdb %zmm6, %xmm6 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xf6]
177 ; X86NOBW-NEXT: vinserti128 $1, %xmm6, %ymm5, %ymm5 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x55,0x38,0xee,0x01]
178 ; X86NOBW-NEXT: vinserti64x4 $1, %ymm1, %zmm5, %zmm5 # encoding: [0x62,0xf3,0xd5,0x48,0x3a,0xe9,0x01]
179 ; X86NOBW-NEXT: vpandq %zmm4, %zmm5, %zmm1 # encoding: [0x62,0xf1,0xd5,0x48,0xdb,0xcc]
180 ; X86NOBW-NEXT: vpternlogq $184, %zmm3, %zmm5, %zmm2 # encoding: [0x62,0xf3,0xd5,0x48,0x25,0xd3,0xb8]
181 ; X86NOBW-NEXT: retl # encoding: [0xc3]
182 ;
183 ; X64NOBW-LABEL: test_vgf2p8affineinvqb_512:
184 ; X64NOBW: # %bb.0:
185 ; X64NOBW-NEXT: movq %rdi, %rax # encoding: [0x48,0x89,0xf8]
186 ; X64NOBW-NEXT: movl %edi, %ecx # encoding: [0x89,0xf9]
187 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
188 ; X64NOBW-NEXT: shrq $32, %rdi # encoding: [0x48,0xc1,0xef,0x20]
189 ; X64NOBW-NEXT: shrq $48, %rax # encoding: [0x48,0xc1,0xe8,0x30]
190 ; X64NOBW-NEXT: shrl $16, %ecx # encoding: [0xc1,0xe9,0x10]
191 ; X64NOBW-NEXT: kmovw %ecx, %k2 # encoding: [0xc5,0xf8,0x92,0xd1]
192 ; X64NOBW-NEXT: kmovw %eax, %k3 # encoding: [0xc5,0xf8,0x92,0xd8]
193 ; X64NOBW-NEXT: kmovw %edi, %k4 # encoding: [0xc5,0xf8,0x92,0xe7]
194 ; X64NOBW-NEXT: vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0xcf,0xd9,0x03]
195 ; X64NOBW-NEXT: vgf2p8affineinvqb $4, %zmm1, %zmm0, %zmm4 # encoding: [0x62,0xf3,0xfd,0x48,0xcf,0xe1,0x04]
196 ; X64NOBW-NEXT: vgf2p8affineinvqb $5, %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf3,0xfd,0x48,0xcf,0xc1,0x05]
197 ; X64NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k4} {z} # encoding: [0x62,0xf3,0x75,0xcc,0x25,0xc9,0xff]
198 ; X64NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
199 ; X64NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k3} {z} # encoding: [0x62,0xf3,0x55,0xcb,0x25,0xed,0xff]
200 ; X64NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
201 ; X64NOBW-NEXT: vinserti128 $1, %xmm5, %ymm1, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xcd,0x01]
202 ; X64NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k1} {z} # encoding: [0x62,0xf3,0x55,0xc9,0x25,0xed,0xff]
203 ; X64NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
204 ; X64NOBW-NEXT: vpternlogd $255, %zmm6, %zmm6, %zmm6 {%k2} {z} # encoding: [0x62,0xf3,0x4d,0xca,0x25,0xf6,0xff]
205 ; X64NOBW-NEXT: vpmovdb %zmm6, %xmm6 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xf6]
206 ; X64NOBW-NEXT: vinserti128 $1, %xmm6, %ymm5, %ymm5 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x55,0x38,0xee,0x01]
207 ; X64NOBW-NEXT: vinserti64x4 $1, %ymm1, %zmm5, %zmm5 # encoding: [0x62,0xf3,0xd5,0x48,0x3a,0xe9,0x01]
208 ; X64NOBW-NEXT: vpandq %zmm4, %zmm5, %zmm1 # encoding: [0x62,0xf1,0xd5,0x48,0xdb,0xcc]
209 ; X64NOBW-NEXT: vpternlogq $184, %zmm0, %zmm5, %zmm2 # encoding: [0x62,0xf3,0xd5,0x48,0x25,0xd0,0xb8]
210 ; X64NOBW-NEXT: vmovdqa64 %zmm3, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc3]
211 ; X64NOBW-NEXT: retq # encoding: [0xc3]
212 %1 = bitcast i64 %mask to <64 x i1>
213 %2 = call <64 x i8> @llvm.x86.vgf2p8affineinvqb.512(<64 x i8> %src1, <64 x i8> %src2, i8 3)
214 %3 = call <64 x i8> @llvm.x86.vgf2p8affineinvqb.512(<64 x i8> %src1, <64 x i8> %src2, i8 4)
215 %4 = call <64 x i8> @llvm.x86.vgf2p8affineinvqb.512(<64 x i8> %src1, <64 x i8> %src2, i8 5)
216 %5 = select <64 x i1> %1, <64 x i8> %3, <64 x i8> zeroinitializer
217 %6 = select <64 x i1> %1, <64 x i8> %4, <64 x i8> %passthru
218 %7 = insertvalue { <64 x i8>, <64 x i8>, <64 x i8> } undef, <64 x i8> %2, 0
219 %8 = insertvalue { <64 x i8>, <64 x i8>, <64 x i8> } %7, <64 x i8> %5, 1
220 %9 = insertvalue { <64 x i8>, <64 x i8>, <64 x i8> } %8, <64 x i8> %6, 2
221 ret { <64 x i8>, <64 x i8>, <64 x i8> } %9
222 }
224 declare <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8>, <16 x i8>, i8)
225 define { <16 x i8>, <16 x i8>, <16 x i8> } @test_vgf2p8affineqb_128(<16 x i8> %src1, <16 x i8> %src2, <16 x i8> %passthru, i16 %mask) {
226 ; X86BW-LABEL: test_vgf2p8affineqb_128:
227 ; X86BW: # %bb.0:
228 ; X86BW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
229 ; X86BW-NEXT: vgf2p8affineqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xd9,0x03]
230 ; X86BW-NEXT: vgf2p8affineqb $4, %xmm1, %xmm0, %xmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0x89,0xce,0xe1,0x04]
231 ; X86BW-NEXT: vgf2p8affineqb $5, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0xce,0xd1,0x05]
232 ; X86BW-NEXT: vmovdqa %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc3]
233 ; X86BW-NEXT: vmovdqa %xmm4, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcc]
234 ; X86BW-NEXT: retl # encoding: [0xc3]
235 ;
236 ; X64BW-LABEL: test_vgf2p8affineqb_128:
237 ; X64BW: # %bb.0:
238 ; X64BW-NEXT: kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
239 ; X64BW-NEXT: vgf2p8affineqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xd9,0x03]
240 ; X64BW-NEXT: vgf2p8affineqb $4, %xmm1, %xmm0, %xmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0x89,0xce,0xe1,0x04]
241 ; X64BW-NEXT: vgf2p8affineqb $5, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0xce,0xd1,0x05]
242 ; X64BW-NEXT: vmovdqa %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc3]
243 ; X64BW-NEXT: vmovdqa %xmm4, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcc]
244 ; X64BW-NEXT: retq # encoding: [0xc3]
245 ;
246 ; X86NOBW-LABEL: test_vgf2p8affineqb_128:
247 ; X86NOBW: # %bb.0:
248 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
249 ; X86NOBW-NEXT: vgf2p8affineqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xd9,0x03]
250 ; X86NOBW-NEXT: vgf2p8affineqb $4, %xmm1, %xmm0, %xmm4 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xe1,0x04]
251 ; X86NOBW-NEXT: vgf2p8affineqb $5, %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xc1,0x05]
252 ; X86NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
253 ; X86NOBW-NEXT: vpmovdb %zmm1, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xcd]
254 ; X86NOBW-NEXT: vpand %xmm4, %xmm5, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xd1,0xdb,0xcc]
255 ; X86NOBW-NEXT: vpternlogq $184, %xmm0, %xmm5, %xmm2 # encoding: [0x62,0xf3,0xd5,0x08,0x25,0xd0,0xb8]
256 ; X86NOBW-NEXT: vmovdqa %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc3]
257 ; X86NOBW-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77]
258 ; X86NOBW-NEXT: retl # encoding: [0xc3]
259 ;
260 ; X64NOBW-LABEL: test_vgf2p8affineqb_128:
261 ; X64NOBW: # %bb.0:
262 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
263 ; X64NOBW-NEXT: vgf2p8affineqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xd9,0x03]
264 ; X64NOBW-NEXT: vgf2p8affineqb $4, %xmm1, %xmm0, %xmm4 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xe1,0x04]
265 ; X64NOBW-NEXT: vgf2p8affineqb $5, %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xc1,0x05]
266 ; X64NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
267 ; X64NOBW-NEXT: vpmovdb %zmm1, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xcd]
268 ; X64NOBW-NEXT: vpand %xmm4, %xmm5, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xd1,0xdb,0xcc]
269 ; X64NOBW-NEXT: vpternlogq $184, %xmm0, %xmm5, %xmm2 # encoding: [0x62,0xf3,0xd5,0x08,0x25,0xd0,0xb8]
270 ; X64NOBW-NEXT: vmovdqa %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc3]
271 ; X64NOBW-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77]
272 ; X64NOBW-NEXT: retq # encoding: [0xc3]
273 %1 = bitcast i16 %mask to <16 x i1>
274 %2 = call <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 3)
275 %3 = call <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 4)
276 %4 = call <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 5)
277 %5 = select <16 x i1> %1, <16 x i8> %3, <16 x i8> zeroinitializer
278 %6 = select <16 x i1> %1, <16 x i8> %4, <16 x i8> %passthru
279 %7 = insertvalue { <16 x i8>, <16 x i8>, <16 x i8> } undef, <16 x i8> %2, 0
280 %8 = insertvalue { <16 x i8>, <16 x i8>, <16 x i8> } %7, <16 x i8> %5, 1
281 %9 = insertvalue { <16 x i8>, <16 x i8>, <16 x i8> } %8, <16 x i8> %6, 2
282 ret { <16 x i8>, <16 x i8>, <16 x i8> } %9
283 }
285 declare <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8>, <32 x i8>, i8)
286 define { <32 x i8>, <32 x i8>, <32 x i8> } @test_vgf2p8affineqb_256(<32 x i8> %src1, <32 x i8> %src2, <32 x i8> %passthru, i32 %mask) {
287 ; X86BW-LABEL: test_vgf2p8affineqb_256:
288 ; X86BW: # %bb.0:
289 ; X86BW-NEXT: kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x04]
290 ; X86BW-NEXT: vgf2p8affineqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xd9,0x03]
291 ; X86BW-NEXT: vgf2p8affineqb $4, %ymm1, %ymm0, %ymm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0xce,0xe1,0x04]
292 ; X86BW-NEXT: vgf2p8affineqb $5, %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0xce,0xd1,0x05]
293 ; X86BW-NEXT: vmovdqa %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc3]
294 ; X86BW-NEXT: vmovdqa %ymm4, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcc]
295 ; X86BW-NEXT: retl # encoding: [0xc3]
296 ;
297 ; X64BW-LABEL: test_vgf2p8affineqb_256:
298 ; X64BW: # %bb.0:
299 ; X64BW-NEXT: kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
300 ; X64BW-NEXT: vgf2p8affineqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xd9,0x03]
301 ; X64BW-NEXT: vgf2p8affineqb $4, %ymm1, %ymm0, %ymm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0xce,0xe1,0x04]
302 ; X64BW-NEXT: vgf2p8affineqb $5, %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0xce,0xd1,0x05]
303 ; X64BW-NEXT: vmovdqa %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc3]
304 ; X64BW-NEXT: vmovdqa %ymm4, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcc]
305 ; X64BW-NEXT: retq # encoding: [0xc3]
306 ;
307 ; X86NOBW-LABEL: test_vgf2p8affineqb_256:
308 ; X86NOBW: # %bb.0:
309 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
310 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k2 # encoding: [0xc5,0xf8,0x90,0x54,0x24,0x06]
311 ; X86NOBW-NEXT: vgf2p8affineqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xd9,0x03]
312 ; X86NOBW-NEXT: vgf2p8affineqb $4, %ymm1, %ymm0, %ymm4 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xe1,0x04]
313 ; X86NOBW-NEXT: vgf2p8affineqb $5, %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xc1,0x05]
314 ; X86NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
315 ; X86NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
316 ; X86NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k2} {z} # encoding: [0x62,0xf3,0x55,0xca,0x25,0xed,0xff]
317 ; X86NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
318 ; X86NOBW-NEXT: vinserti128 $1, %xmm5, %ymm1, %ymm5 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xed,0x01]
319 ; X86NOBW-NEXT: vpand %ymm4, %ymm5, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xd5,0xdb,0xcc]
320 ; X86NOBW-NEXT: vpternlogq $184, %ymm0, %ymm5, %ymm2 # encoding: [0x62,0xf3,0xd5,0x28,0x25,0xd0,0xb8]
321 ; X86NOBW-NEXT: vmovdqa %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc3]
322 ; X86NOBW-NEXT: retl # encoding: [0xc3]
323 ;
324 ; X64NOBW-LABEL: test_vgf2p8affineqb_256:
325 ; X64NOBW: # %bb.0:
326 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
327 ; X64NOBW-NEXT: shrl $16, %edi # encoding: [0xc1,0xef,0x10]
328 ; X64NOBW-NEXT: kmovw %edi, %k2 # encoding: [0xc5,0xf8,0x92,0xd7]
329 ; X64NOBW-NEXT: vgf2p8affineqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xd9,0x03]
330 ; X64NOBW-NEXT: vgf2p8affineqb $4, %ymm1, %ymm0, %ymm4 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xe1,0x04]
331 ; X64NOBW-NEXT: vgf2p8affineqb $5, %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xc1,0x05]
332 ; X64NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
333 ; X64NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
334 ; X64NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k2} {z} # encoding: [0x62,0xf3,0x55,0xca,0x25,0xed,0xff]
335 ; X64NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
336 ; X64NOBW-NEXT: vinserti128 $1, %xmm5, %ymm1, %ymm5 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xed,0x01]
337 ; X64NOBW-NEXT: vpand %ymm4, %ymm5, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xd5,0xdb,0xcc]
338 ; X64NOBW-NEXT: vpternlogq $184, %ymm0, %ymm5, %ymm2 # encoding: [0x62,0xf3,0xd5,0x28,0x25,0xd0,0xb8]
339 ; X64NOBW-NEXT: vmovdqa %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc3]
340 ; X64NOBW-NEXT: retq # encoding: [0xc3]
341 %1 = bitcast i32 %mask to <32 x i1>
342 %2 = call <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 3)
343 %3 = call <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 4)
344 %4 = call <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 5)
345 %5 = select <32 x i1> %1, <32 x i8> %3, <32 x i8> zeroinitializer
346 %6 = select <32 x i1> %1, <32 x i8> %4, <32 x i8> %passthru
347 %7 = insertvalue { <32 x i8>, <32 x i8>, <32 x i8> } undef, <32 x i8> %2, 0
348 %8 = insertvalue { <32 x i8>, <32 x i8>, <32 x i8> } %7, <32 x i8> %5, 1
349 %9 = insertvalue { <32 x i8>, <32 x i8>, <32 x i8> } %8, <32 x i8> %6, 2
350 ret { <32 x i8>, <32 x i8>, <32 x i8> } %9
351 }
353 declare <64 x i8> @llvm.x86.vgf2p8affineqb.512(<64 x i8>, <64 x i8>, i8)
354 define { <64 x i8>, <64 x i8>, <64 x i8> } @test_vgf2p8affineqb_512(<64 x i8> %src1, <64 x i8> %src2, <64 x i8> %passthru, i64 %mask) {
355 ; X86BW-LABEL: test_vgf2p8affineqb_512:
356 ; X86BW: # %bb.0:
357 ; X86BW-NEXT: kmovq {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf8,0x90,0x4c,0x24,0x04]
358 ; X86BW-NEXT: vgf2p8affineqb $3, %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0xce,0xd9,0x03]
359 ; X86BW-NEXT: vgf2p8affineqb $4, %zmm1, %zmm0, %zmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0xce,0xe1,0x04]
360 ; X86BW-NEXT: vgf2p8affineqb $5, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0xce,0xd1,0x05]
361 ; X86BW-NEXT: vmovdqa64 %zmm3, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc3]
362 ; X86BW-NEXT: vmovdqa64 %zmm4, %zmm1 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xcc]
363 ; X86BW-NEXT: retl # encoding: [0xc3]
364 ;
365 ; X64BW-LABEL: test_vgf2p8affineqb_512:
366 ; X64BW: # %bb.0:
367 ; X64BW-NEXT: kmovq %rdi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
368 ; X64BW-NEXT: vgf2p8affineqb $3, %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0xce,0xd9,0x03]
369 ; X64BW-NEXT: vgf2p8affineqb $4, %zmm1, %zmm0, %zmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0xce,0xe1,0x04]
370 ; X64BW-NEXT: vgf2p8affineqb $5, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0xce,0xd1,0x05]
371 ; X64BW-NEXT: vmovdqa64 %zmm3, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc3]
372 ; X64BW-NEXT: vmovdqa64 %zmm4, %zmm1 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xcc]
373 ; X64BW-NEXT: retq # encoding: [0xc3]
374 ;
375 ; X86NOBW-LABEL: test_vgf2p8affineqb_512:
376 ; X86NOBW: # %bb.0:
377 ; X86NOBW-NEXT: vmovdqa64 %zmm0, %zmm3 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xd8]
378 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k2 # encoding: [0xc5,0xf8,0x90,0x54,0x24,0x04]
379 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x06]
380 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k3 # encoding: [0xc5,0xf8,0x90,0x5c,0x24,0x0a]
381 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k4 # encoding: [0xc5,0xf8,0x90,0x64,0x24,0x08]
382 ; X86NOBW-NEXT: vgf2p8affineqb $3, %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf3,0xfd,0x48,0xce,0xc1,0x03]
383 ; X86NOBW-NEXT: vgf2p8affineqb $4, %zmm1, %zmm3, %zmm4 # encoding: [0x62,0xf3,0xe5,0x48,0xce,0xe1,0x04]
384 ; X86NOBW-NEXT: vgf2p8affineqb $5, %zmm1, %zmm3, %zmm3 # encoding: [0x62,0xf3,0xe5,0x48,0xce,0xd9,0x05]
385 ; X86NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k4} {z} # encoding: [0x62,0xf3,0x75,0xcc,0x25,0xc9,0xff]
386 ; X86NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
387 ; X86NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k3} {z} # encoding: [0x62,0xf3,0x55,0xcb,0x25,0xed,0xff]
388 ; X86NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
389 ; X86NOBW-NEXT: vinserti128 $1, %xmm5, %ymm1, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xcd,0x01]
390 ; X86NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k2} {z} # encoding: [0x62,0xf3,0x55,0xca,0x25,0xed,0xff]
391 ; X86NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
392 ; X86NOBW-NEXT: vpternlogd $255, %zmm6, %zmm6, %zmm6 {%k1} {z} # encoding: [0x62,0xf3,0x4d,0xc9,0x25,0xf6,0xff]
393 ; X86NOBW-NEXT: vpmovdb %zmm6, %xmm6 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xf6]
394 ; X86NOBW-NEXT: vinserti128 $1, %xmm6, %ymm5, %ymm5 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x55,0x38,0xee,0x01]
395 ; X86NOBW-NEXT: vinserti64x4 $1, %ymm1, %zmm5, %zmm5 # encoding: [0x62,0xf3,0xd5,0x48,0x3a,0xe9,0x01]
396 ; X86NOBW-NEXT: vpandq %zmm4, %zmm5, %zmm1 # encoding: [0x62,0xf1,0xd5,0x48,0xdb,0xcc]
397 ; X86NOBW-NEXT: vpternlogq $184, %zmm3, %zmm5, %zmm2 # encoding: [0x62,0xf3,0xd5,0x48,0x25,0xd3,0xb8]
398 ; X86NOBW-NEXT: retl # encoding: [0xc3]
399 ;
400 ; X64NOBW-LABEL: test_vgf2p8affineqb_512:
401 ; X64NOBW: # %bb.0:
402 ; X64NOBW-NEXT: movq %rdi, %rax # encoding: [0x48,0x89,0xf8]
403 ; X64NOBW-NEXT: movl %edi, %ecx # encoding: [0x89,0xf9]
404 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
405 ; X64NOBW-NEXT: shrq $32, %rdi # encoding: [0x48,0xc1,0xef,0x20]
406 ; X64NOBW-NEXT: shrq $48, %rax # encoding: [0x48,0xc1,0xe8,0x30]
407 ; X64NOBW-NEXT: shrl $16, %ecx # encoding: [0xc1,0xe9,0x10]
408 ; X64NOBW-NEXT: kmovw %ecx, %k2 # encoding: [0xc5,0xf8,0x92,0xd1]
409 ; X64NOBW-NEXT: kmovw %eax, %k3 # encoding: [0xc5,0xf8,0x92,0xd8]
410 ; X64NOBW-NEXT: kmovw %edi, %k4 # encoding: [0xc5,0xf8,0x92,0xe7]
411 ; X64NOBW-NEXT: vgf2p8affineqb $3, %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0xce,0xd9,0x03]
412 ; X64NOBW-NEXT: vgf2p8affineqb $4, %zmm1, %zmm0, %zmm4 # encoding: [0x62,0xf3,0xfd,0x48,0xce,0xe1,0x04]
413 ; X64NOBW-NEXT: vgf2p8affineqb $5, %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf3,0xfd,0x48,0xce,0xc1,0x05]
414 ; X64NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k4} {z} # encoding: [0x62,0xf3,0x75,0xcc,0x25,0xc9,0xff]
415 ; X64NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
416 ; X64NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k3} {z} # encoding: [0x62,0xf3,0x55,0xcb,0x25,0xed,0xff]
417 ; X64NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
418 ; X64NOBW-NEXT: vinserti128 $1, %xmm5, %ymm1, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xcd,0x01]
419 ; X64NOBW-NEXT: vpternlogd $255, %zmm5, %zmm5, %zmm5 {%k1} {z} # encoding: [0x62,0xf3,0x55,0xc9,0x25,0xed,0xff]
420 ; X64NOBW-NEXT: vpmovdb %zmm5, %xmm5 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xed]
421 ; X64NOBW-NEXT: vpternlogd $255, %zmm6, %zmm6, %zmm6 {%k2} {z} # encoding: [0x62,0xf3,0x4d,0xca,0x25,0xf6,0xff]
422 ; X64NOBW-NEXT: vpmovdb %zmm6, %xmm6 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xf6]
423 ; X64NOBW-NEXT: vinserti128 $1, %xmm6, %ymm5, %ymm5 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x55,0x38,0xee,0x01]
424 ; X64NOBW-NEXT: vinserti64x4 $1, %ymm1, %zmm5, %zmm5 # encoding: [0x62,0xf3,0xd5,0x48,0x3a,0xe9,0x01]
425 ; X64NOBW-NEXT: vpandq %zmm4, %zmm5, %zmm1 # encoding: [0x62,0xf1,0xd5,0x48,0xdb,0xcc]
426 ; X64NOBW-NEXT: vpternlogq $184, %zmm0, %zmm5, %zmm2 # encoding: [0x62,0xf3,0xd5,0x48,0x25,0xd0,0xb8]
427 ; X64NOBW-NEXT: vmovdqa64 %zmm3, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc3]
428 ; X64NOBW-NEXT: retq # encoding: [0xc3]
429 %1 = bitcast i64 %mask to <64 x i1>
430 %2 = call <64 x i8> @llvm.x86.vgf2p8affineqb.512(<64 x i8> %src1, <64 x i8> %src2, i8 3)
431 %3 = call <64 x i8> @llvm.x86.vgf2p8affineqb.512(<64 x i8> %src1, <64 x i8> %src2, i8 4)
432 %4 = call <64 x i8> @llvm.x86.vgf2p8affineqb.512(<64 x i8> %src1, <64 x i8> %src2, i8 5)
433 %5 = select <64 x i1> %1, <64 x i8> %3, <64 x i8> zeroinitializer
434 %6 = select <64 x i1> %1, <64 x i8> %4, <64 x i8> %passthru
435 %7 = insertvalue { <64 x i8>, <64 x i8>, <64 x i8> } undef, <64 x i8> %2, 0
436 %8 = insertvalue { <64 x i8>, <64 x i8>, <64 x i8> } %7, <64 x i8> %5, 1
437 %9 = insertvalue { <64 x i8>, <64 x i8>, <64 x i8> } %8, <64 x i8> %6, 2
438 ret { <64 x i8>, <64 x i8>, <64 x i8> } %9
439 }
441 declare <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8>, <16 x i8>)
442 define <16 x i8> @test_vgf2p8mulb_128(<16 x i8> %src1, <16 x i8> %src2) {
443 ; CHECK-LABEL: test_vgf2p8mulb_128:
444 ; CHECK: # %bb.0:
445 ; CHECK-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xcf,0xc1]
446 ; CHECK-NEXT: ret{{[l|q]}} # encoding: [0xc3]
447 %1 = call <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8> %src1, <16 x i8> %src2)
448 ret <16 x i8> %1
449 }
451 define <16 x i8> @test_vgf2p8mulb_128_mask(<16 x i8> %src1, <16 x i8> %src2, <16 x i8> %passthru, i16 %mask) {
452 ; X86BW-LABEL: test_vgf2p8mulb_128_mask:
453 ; X86BW: # %bb.0:
454 ; X86BW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
455 ; X86BW-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xcf,0xd1]
456 ; X86BW-NEXT: vmovdqa %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc2]
457 ; X86BW-NEXT: retl # encoding: [0xc3]
458 ;
459 ; X64BW-LABEL: test_vgf2p8mulb_128_mask:
460 ; X64BW: # %bb.0:
461 ; X64BW-NEXT: kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
462 ; X64BW-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xcf,0xd1]
463 ; X64BW-NEXT: vmovdqa %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc2]
464 ; X64BW-NEXT: retq # encoding: [0xc3]
465 ;
466 ; X86NOBW-LABEL: test_vgf2p8mulb_128_mask:
467 ; X86NOBW: # %bb.0:
468 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
469 ; X86NOBW-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm1 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xcf,0xc9]
470 ; X86NOBW-NEXT: vpternlogd $255, %zmm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x25,0xc0,0xff]
471 ; X86NOBW-NEXT: vpmovdb %zmm0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc0]
472 ; X86NOBW-NEXT: vpternlogq $202, %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf3,0xf5,0x08,0x25,0xc2,0xca]
473 ; X86NOBW-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77]
474 ; X86NOBW-NEXT: retl # encoding: [0xc3]
475 ;
476 ; X64NOBW-LABEL: test_vgf2p8mulb_128_mask:
477 ; X64NOBW: # %bb.0:
478 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
479 ; X64NOBW-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm1 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xcf,0xc9]
480 ; X64NOBW-NEXT: vpternlogd $255, %zmm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x25,0xc0,0xff]
481 ; X64NOBW-NEXT: vpmovdb %zmm0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc0]
482 ; X64NOBW-NEXT: vpternlogq $202, %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf3,0xf5,0x08,0x25,0xc2,0xca]
483 ; X64NOBW-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77]
484 ; X64NOBW-NEXT: retq # encoding: [0xc3]
485 %1 = bitcast i16 %mask to <16 x i1>
486 %2 = call <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8> %src1, <16 x i8> %src2)
487 %3 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> %passthru
488 ret <16 x i8> %3
489 }
491 define <16 x i8> @test_vgf2p8mulb_128_maskz(<16 x i8> %src1, <16 x i8> %src2, i16 %mask) {
492 ; X86BW-LABEL: test_vgf2p8mulb_128_maskz:
493 ; X86BW: # %bb.0:
494 ; X86BW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
495 ; X86BW-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0x89,0xcf,0xc1]
496 ; X86BW-NEXT: retl # encoding: [0xc3]
497 ;
498 ; X64BW-LABEL: test_vgf2p8mulb_128_maskz:
499 ; X64BW: # %bb.0:
500 ; X64BW-NEXT: kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
501 ; X64BW-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0x89,0xcf,0xc1]
502 ; X64BW-NEXT: retq # encoding: [0xc3]
503 ;
504 ; X86NOBW-LABEL: test_vgf2p8mulb_128_maskz:
505 ; X86NOBW: # %bb.0:
506 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
507 ; X86NOBW-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xcf,0xc1]
508 ; X86NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
509 ; X86NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
510 ; X86NOBW-NEXT: vpand %xmm0, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0xdb,0xc0]
511 ; X86NOBW-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77]
512 ; X86NOBW-NEXT: retl # encoding: [0xc3]
513 ;
514 ; X64NOBW-LABEL: test_vgf2p8mulb_128_maskz:
515 ; X64NOBW: # %bb.0:
516 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
517 ; X64NOBW-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xcf,0xc1]
518 ; X64NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
519 ; X64NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
520 ; X64NOBW-NEXT: vpand %xmm0, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0xdb,0xc0]
521 ; X64NOBW-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77]
522 ; X64NOBW-NEXT: retq # encoding: [0xc3]
523 %1 = bitcast i16 %mask to <16 x i1>
524 %2 = call <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8> %src1, <16 x i8> %src2)
525 %3 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> zeroinitializer
526 ret <16 x i8> %3
527 }
529 declare <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8>, <32 x i8>)
530 define <32 x i8> @test_vgf2p8mulb_256(<32 x i8> %src1, <32 x i8> %src2) {
531 ; CHECK-LABEL: test_vgf2p8mulb_256:
532 ; CHECK: # %bb.0:
533 ; CHECK-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0xcf,0xc1]
534 ; CHECK-NEXT: ret{{[l|q]}} # encoding: [0xc3]
535 %1 = call <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8> %src1, <32 x i8> %src2)
536 ret <32 x i8> %1
537 }
539 define <32 x i8> @test_vgf2p8mulb_256_mask(<32 x i8> %src1, <32 x i8> %src2, <32 x i8> %passthru, i32 %mask) {
540 ; X86BW-LABEL: test_vgf2p8mulb_256_mask:
541 ; X86BW: # %bb.0:
542 ; X86BW-NEXT: kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x04]
543 ; X86BW-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xcf,0xd1]
544 ; X86BW-NEXT: vmovdqa %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc2]
545 ; X86BW-NEXT: retl # encoding: [0xc3]
546 ;
547 ; X64BW-LABEL: test_vgf2p8mulb_256_mask:
548 ; X64BW: # %bb.0:
549 ; X64BW-NEXT: kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
550 ; X64BW-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xcf,0xd1]
551 ; X64BW-NEXT: vmovdqa %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc2]
552 ; X64BW-NEXT: retq # encoding: [0xc3]
553 ;
554 ; X86NOBW-LABEL: test_vgf2p8mulb_256_mask:
555 ; X86NOBW: # %bb.0:
556 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
557 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k2 # encoding: [0xc5,0xf8,0x90,0x54,0x24,0x06]
558 ; X86NOBW-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0xcf,0xc9]
559 ; X86NOBW-NEXT: vpternlogd $255, %zmm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x25,0xc0,0xff]
560 ; X86NOBW-NEXT: vpmovdb %zmm0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc0]
561 ; X86NOBW-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k2} {z} # encoding: [0x62,0xf3,0x65,0xca,0x25,0xdb,0xff]
562 ; X86NOBW-NEXT: vpmovdb %zmm3, %xmm3 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xdb]
563 ; X86NOBW-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xc3,0x01]
564 ; X86NOBW-NEXT: vpternlogq $202, %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf3,0xf5,0x28,0x25,0xc2,0xca]
565 ; X86NOBW-NEXT: retl # encoding: [0xc3]
566 ;
567 ; X64NOBW-LABEL: test_vgf2p8mulb_256_mask:
568 ; X64NOBW: # %bb.0:
569 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
570 ; X64NOBW-NEXT: shrl $16, %edi # encoding: [0xc1,0xef,0x10]
571 ; X64NOBW-NEXT: kmovw %edi, %k2 # encoding: [0xc5,0xf8,0x92,0xd7]
572 ; X64NOBW-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0xcf,0xc9]
573 ; X64NOBW-NEXT: vpternlogd $255, %zmm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x25,0xc0,0xff]
574 ; X64NOBW-NEXT: vpmovdb %zmm0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc0]
575 ; X64NOBW-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k2} {z} # encoding: [0x62,0xf3,0x65,0xca,0x25,0xdb,0xff]
576 ; X64NOBW-NEXT: vpmovdb %zmm3, %xmm3 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xdb]
577 ; X64NOBW-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xc3,0x01]
578 ; X64NOBW-NEXT: vpternlogq $202, %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf3,0xf5,0x28,0x25,0xc2,0xca]
579 ; X64NOBW-NEXT: retq # encoding: [0xc3]
580 %1 = bitcast i32 %mask to <32 x i1>
581 %2 = call <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8> %src1, <32 x i8> %src2)
582 %3 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> %passthru
583 ret <32 x i8> %3
584 }
586 define <32 x i8> @test_vgf2p8mulb_256_maskz(<32 x i8> %src1, <32 x i8> %src2, i32 %mask) {
587 ; X86BW-LABEL: test_vgf2p8mulb_256_maskz:
588 ; X86BW: # %bb.0:
589 ; X86BW-NEXT: kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x04]
590 ; X86BW-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xa9,0xcf,0xc1]
591 ; X86BW-NEXT: retl # encoding: [0xc3]
592 ;
593 ; X64BW-LABEL: test_vgf2p8mulb_256_maskz:
594 ; X64BW: # %bb.0:
595 ; X64BW-NEXT: kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
596 ; X64BW-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xa9,0xcf,0xc1]
597 ; X64BW-NEXT: retq # encoding: [0xc3]
598 ;
599 ; X86NOBW-LABEL: test_vgf2p8mulb_256_maskz:
600 ; X86NOBW: # %bb.0:
601 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
602 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k2 # encoding: [0xc5,0xf8,0x90,0x54,0x24,0x06]
603 ; X86NOBW-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0xcf,0xc1]
604 ; X86NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
605 ; X86NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
606 ; X86NOBW-NEXT: vpternlogd $255, %zmm2, %zmm2, %zmm2 {%k2} {z} # encoding: [0x62,0xf3,0x6d,0xca,0x25,0xd2,0xff]
607 ; X86NOBW-NEXT: vpmovdb %zmm2, %xmm2 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xd2]
608 ; X86NOBW-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xca,0x01]
609 ; X86NOBW-NEXT: vpand %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf5,0xdb,0xc0]
610 ; X86NOBW-NEXT: retl # encoding: [0xc3]
611 ;
612 ; X64NOBW-LABEL: test_vgf2p8mulb_256_maskz:
613 ; X64NOBW: # %bb.0:
614 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
615 ; X64NOBW-NEXT: shrl $16, %edi # encoding: [0xc1,0xef,0x10]
616 ; X64NOBW-NEXT: kmovw %edi, %k2 # encoding: [0xc5,0xf8,0x92,0xd7]
617 ; X64NOBW-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0xcf,0xc1]
618 ; X64NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z} # encoding: [0x62,0xf3,0x75,0xc9,0x25,0xc9,0xff]
619 ; X64NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
620 ; X64NOBW-NEXT: vpternlogd $255, %zmm2, %zmm2, %zmm2 {%k2} {z} # encoding: [0x62,0xf3,0x6d,0xca,0x25,0xd2,0xff]
621 ; X64NOBW-NEXT: vpmovdb %zmm2, %xmm2 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xd2]
622 ; X64NOBW-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xca,0x01]
623 ; X64NOBW-NEXT: vpand %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf5,0xdb,0xc0]
624 ; X64NOBW-NEXT: retq # encoding: [0xc3]
625 %1 = bitcast i32 %mask to <32 x i1>
626 %2 = call <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8> %src1, <32 x i8> %src2)
627 %3 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> zeroinitializer
628 ret <32 x i8> %3
629 }
631 declare <64 x i8> @llvm.x86.vgf2p8mulb.512(<64 x i8>, <64 x i8>)
632 define <64 x i8> @test_vgf2p8mulb_512(<64 x i8> %src1, <64 x i8> %src2) {
633 ; CHECK-LABEL: test_vgf2p8mulb_512:
634 ; CHECK: # %bb.0:
635 ; CHECK-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf2,0x7d,0x48,0xcf,0xc1]
636 ; CHECK-NEXT: ret{{[l|q]}} # encoding: [0xc3]
637 %1 = call <64 x i8> @llvm.x86.vgf2p8mulb.512(<64 x i8> %src1, <64 x i8> %src2)
638 ret <64 x i8> %1
639 }
641 define <64 x i8> @test_vgf2p8mulb_512_mask(<64 x i8> %src1, <64 x i8> %src2, <64 x i8> %passthru, i64 %mask) {
642 ; X86BW-LABEL: test_vgf2p8mulb_512_mask:
643 ; X86BW: # %bb.0:
644 ; X86BW-NEXT: kmovq {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf8,0x90,0x4c,0x24,0x04]
645 ; X86BW-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0xcf,0xd1]
646 ; X86BW-NEXT: vmovdqa64 %zmm2, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc2]
647 ; X86BW-NEXT: retl # encoding: [0xc3]
648 ;
649 ; X64BW-LABEL: test_vgf2p8mulb_512_mask:
650 ; X64BW: # %bb.0:
651 ; X64BW-NEXT: kmovq %rdi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
652 ; X64BW-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0xcf,0xd1]
653 ; X64BW-NEXT: vmovdqa64 %zmm2, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc2]
654 ; X64BW-NEXT: retq # encoding: [0xc3]
655 ;
656 ; X86NOBW-LABEL: test_vgf2p8mulb_512_mask:
657 ; X86NOBW: # %bb.0:
658 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
659 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k2 # encoding: [0xc5,0xf8,0x90,0x54,0x24,0x06]
660 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k3 # encoding: [0xc5,0xf8,0x90,0x5c,0x24,0x0a]
661 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k4 # encoding: [0xc5,0xf8,0x90,0x64,0x24,0x08]
662 ; X86NOBW-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm1 # encoding: [0x62,0xf2,0x7d,0x48,0xcf,0xc9]
663 ; X86NOBW-NEXT: vpternlogd $255, %zmm0, %zmm0, %zmm0 {%k4} {z} # encoding: [0x62,0xf3,0x7d,0xcc,0x25,0xc0,0xff]
664 ; X86NOBW-NEXT: vpmovdb %zmm0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc0]
665 ; X86NOBW-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k3} {z} # encoding: [0x62,0xf3,0x65,0xcb,0x25,0xdb,0xff]
666 ; X86NOBW-NEXT: vpmovdb %zmm3, %xmm3 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xdb]
667 ; X86NOBW-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xc3,0x01]
668 ; X86NOBW-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k1} {z} # encoding: [0x62,0xf3,0x65,0xc9,0x25,0xdb,0xff]
669 ; X86NOBW-NEXT: vpmovdb %zmm3, %xmm3 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xdb]
670 ; X86NOBW-NEXT: vpternlogd $255, %zmm4, %zmm4, %zmm4 {%k2} {z} # encoding: [0x62,0xf3,0x5d,0xca,0x25,0xe4,0xff]
671 ; X86NOBW-NEXT: vpmovdb %zmm4, %xmm4 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xe4]
672 ; X86NOBW-NEXT: vinserti128 $1, %xmm4, %ymm3, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x65,0x38,0xdc,0x01]
673 ; X86NOBW-NEXT: vinserti64x4 $1, %ymm0, %zmm3, %zmm0 # encoding: [0x62,0xf3,0xe5,0x48,0x3a,0xc0,0x01]
674 ; X86NOBW-NEXT: vpternlogq $202, %zmm2, %zmm1, %zmm0 # encoding: [0x62,0xf3,0xf5,0x48,0x25,0xc2,0xca]
675 ; X86NOBW-NEXT: retl # encoding: [0xc3]
676 ;
677 ; X64NOBW-LABEL: test_vgf2p8mulb_512_mask:
678 ; X64NOBW: # %bb.0:
679 ; X64NOBW-NEXT: movq %rdi, %rax # encoding: [0x48,0x89,0xf8]
680 ; X64NOBW-NEXT: movl %edi, %ecx # encoding: [0x89,0xf9]
681 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
682 ; X64NOBW-NEXT: shrq $32, %rdi # encoding: [0x48,0xc1,0xef,0x20]
683 ; X64NOBW-NEXT: shrq $48, %rax # encoding: [0x48,0xc1,0xe8,0x30]
684 ; X64NOBW-NEXT: shrl $16, %ecx # encoding: [0xc1,0xe9,0x10]
685 ; X64NOBW-NEXT: kmovw %ecx, %k2 # encoding: [0xc5,0xf8,0x92,0xd1]
686 ; X64NOBW-NEXT: kmovw %eax, %k3 # encoding: [0xc5,0xf8,0x92,0xd8]
687 ; X64NOBW-NEXT: kmovw %edi, %k4 # encoding: [0xc5,0xf8,0x92,0xe7]
688 ; X64NOBW-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm1 # encoding: [0x62,0xf2,0x7d,0x48,0xcf,0xc9]
689 ; X64NOBW-NEXT: vpternlogd $255, %zmm0, %zmm0, %zmm0 {%k4} {z} # encoding: [0x62,0xf3,0x7d,0xcc,0x25,0xc0,0xff]
690 ; X64NOBW-NEXT: vpmovdb %zmm0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc0]
691 ; X64NOBW-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k3} {z} # encoding: [0x62,0xf3,0x65,0xcb,0x25,0xdb,0xff]
692 ; X64NOBW-NEXT: vpmovdb %zmm3, %xmm3 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xdb]
693 ; X64NOBW-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xc3,0x01]
694 ; X64NOBW-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k1} {z} # encoding: [0x62,0xf3,0x65,0xc9,0x25,0xdb,0xff]
695 ; X64NOBW-NEXT: vpmovdb %zmm3, %xmm3 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xdb]
696 ; X64NOBW-NEXT: vpternlogd $255, %zmm4, %zmm4, %zmm4 {%k2} {z} # encoding: [0x62,0xf3,0x5d,0xca,0x25,0xe4,0xff]
697 ; X64NOBW-NEXT: vpmovdb %zmm4, %xmm4 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xe4]
698 ; X64NOBW-NEXT: vinserti128 $1, %xmm4, %ymm3, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x65,0x38,0xdc,0x01]
699 ; X64NOBW-NEXT: vinserti64x4 $1, %ymm0, %zmm3, %zmm0 # encoding: [0x62,0xf3,0xe5,0x48,0x3a,0xc0,0x01]
700 ; X64NOBW-NEXT: vpternlogq $202, %zmm2, %zmm1, %zmm0 # encoding: [0x62,0xf3,0xf5,0x48,0x25,0xc2,0xca]
701 ; X64NOBW-NEXT: retq # encoding: [0xc3]
702 %1 = bitcast i64 %mask to <64 x i1>
703 %2 = call <64 x i8> @llvm.x86.vgf2p8mulb.512(<64 x i8> %src1, <64 x i8> %src2)
704 %3 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> %passthru
705 ret <64 x i8> %3
706 }
708 define <64 x i8> @test_vgf2p8mulb_512_maskz(<64 x i8> %src1, <64 x i8> %src2, i64 %mask) {
709 ; X86BW-LABEL: test_vgf2p8mulb_512_maskz:
710 ; X86BW: # %bb.0:
711 ; X86BW-NEXT: kmovq {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf8,0x90,0x4c,0x24,0x04]
712 ; X86BW-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xc9,0xcf,0xc1]
713 ; X86BW-NEXT: retl # encoding: [0xc3]
714 ;
715 ; X64BW-LABEL: test_vgf2p8mulb_512_maskz:
716 ; X64BW: # %bb.0:
717 ; X64BW-NEXT: kmovq %rdi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
718 ; X64BW-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xc9,0xcf,0xc1]
719 ; X64BW-NEXT: retq # encoding: [0xc3]
720 ;
721 ; X86NOBW-LABEL: test_vgf2p8mulb_512_maskz:
722 ; X86NOBW: # %bb.0:
723 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
724 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k2 # encoding: [0xc5,0xf8,0x90,0x54,0x24,0x06]
725 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k3 # encoding: [0xc5,0xf8,0x90,0x5c,0x24,0x0a]
726 ; X86NOBW-NEXT: kmovw {{[0-9]+}}(%esp), %k4 # encoding: [0xc5,0xf8,0x90,0x64,0x24,0x08]
727 ; X86NOBW-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf2,0x7d,0x48,0xcf,0xc1]
728 ; X86NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k4} {z} # encoding: [0x62,0xf3,0x75,0xcc,0x25,0xc9,0xff]
729 ; X86NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
730 ; X86NOBW-NEXT: vpternlogd $255, %zmm2, %zmm2, %zmm2 {%k3} {z} # encoding: [0x62,0xf3,0x6d,0xcb,0x25,0xd2,0xff]
731 ; X86NOBW-NEXT: vpmovdb %zmm2, %xmm2 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xd2]
732 ; X86NOBW-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xca,0x01]
733 ; X86NOBW-NEXT: vpternlogd $255, %zmm2, %zmm2, %zmm2 {%k1} {z} # encoding: [0x62,0xf3,0x6d,0xc9,0x25,0xd2,0xff]
734 ; X86NOBW-NEXT: vpmovdb %zmm2, %xmm2 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xd2]
735 ; X86NOBW-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k2} {z} # encoding: [0x62,0xf3,0x65,0xca,0x25,0xdb,0xff]
736 ; X86NOBW-NEXT: vpmovdb %zmm3, %xmm3 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xdb]
737 ; X86NOBW-NEXT: vinserti128 $1, %xmm3, %ymm2, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x6d,0x38,0xd3,0x01]
738 ; X86NOBW-NEXT: vinserti64x4 $1, %ymm1, %zmm2, %zmm1 # encoding: [0x62,0xf3,0xed,0x48,0x3a,0xc9,0x01]
739 ; X86NOBW-NEXT: vpandq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xdb,0xc0]
740 ; X86NOBW-NEXT: retl # encoding: [0xc3]
741 ;
742 ; X64NOBW-LABEL: test_vgf2p8mulb_512_maskz:
743 ; X64NOBW: # %bb.0:
744 ; X64NOBW-NEXT: movq %rdi, %rax # encoding: [0x48,0x89,0xf8]
745 ; X64NOBW-NEXT: movl %edi, %ecx # encoding: [0x89,0xf9]
746 ; X64NOBW-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
747 ; X64NOBW-NEXT: shrq $32, %rdi # encoding: [0x48,0xc1,0xef,0x20]
748 ; X64NOBW-NEXT: shrq $48, %rax # encoding: [0x48,0xc1,0xe8,0x30]
749 ; X64NOBW-NEXT: shrl $16, %ecx # encoding: [0xc1,0xe9,0x10]
750 ; X64NOBW-NEXT: kmovw %ecx, %k2 # encoding: [0xc5,0xf8,0x92,0xd1]
751 ; X64NOBW-NEXT: kmovw %eax, %k3 # encoding: [0xc5,0xf8,0x92,0xd8]
752 ; X64NOBW-NEXT: kmovw %edi, %k4 # encoding: [0xc5,0xf8,0x92,0xe7]
753 ; X64NOBW-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf2,0x7d,0x48,0xcf,0xc1]
754 ; X64NOBW-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k4} {z} # encoding: [0x62,0xf3,0x75,0xcc,0x25,0xc9,0xff]
755 ; X64NOBW-NEXT: vpmovdb %zmm1, %xmm1 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xc9]
756 ; X64NOBW-NEXT: vpternlogd $255, %zmm2, %zmm2, %zmm2 {%k3} {z} # encoding: [0x62,0xf3,0x6d,0xcb,0x25,0xd2,0xff]
757 ; X64NOBW-NEXT: vpmovdb %zmm2, %xmm2 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xd2]
758 ; X64NOBW-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x75,0x38,0xca,0x01]
759 ; X64NOBW-NEXT: vpternlogd $255, %zmm2, %zmm2, %zmm2 {%k1} {z} # encoding: [0x62,0xf3,0x6d,0xc9,0x25,0xd2,0xff]
760 ; X64NOBW-NEXT: vpmovdb %zmm2, %xmm2 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xd2]
761 ; X64NOBW-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k2} {z} # encoding: [0x62,0xf3,0x65,0xca,0x25,0xdb,0xff]
762 ; X64NOBW-NEXT: vpmovdb %zmm3, %xmm3 # encoding: [0x62,0xf2,0x7e,0x48,0x31,0xdb]
763 ; X64NOBW-NEXT: vinserti128 $1, %xmm3, %ymm2, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x6d,0x38,0xd3,0x01]
764 ; X64NOBW-NEXT: vinserti64x4 $1, %ymm1, %zmm2, %zmm1 # encoding: [0x62,0xf3,0xed,0x48,0x3a,0xc9,0x01]
765 ; X64NOBW-NEXT: vpandq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xdb,0xc0]
766 ; X64NOBW-NEXT: retq # encoding: [0xc3]
767 %1 = bitcast i64 %mask to <64 x i1>
768 %2 = call <64 x i8> @llvm.x86.vgf2p8mulb.512(<64 x i8> %src1, <64 x i8> %src2)
769 %3 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> zeroinitializer
770 ret <64 x i8> %3
771 }