• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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,X86
3; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+gfni,+avx512bw --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X64
4
5declare <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8>, <16 x i8>, i8)
6define <16 x i8> @test_vgf2p8affineinvqb_128(<16 x i8> %src1, <16 x i8> %src2, <16 x i8> %passthru, i16 %mask) {
7; X86-LABEL: test_vgf2p8affineinvqb_128:
8; X86:       # %bb.0:
9; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
10; X86-NEXT:    vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xd9,0x03]
11; X86-NEXT:    vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0x89,0xcf,0xe1,0x03]
12; X86-NEXT:    vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0xcf,0xd1,0x03]
13; X86-NEXT:    vpxor %xmm3, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0xef,0xc3]
14; X86-NEXT:    vpxor %xmm0, %xmm4, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xd9,0xef,0xc0]
15; X86-NEXT:    retl # encoding: [0xc3]
16;
17; X64-LABEL: test_vgf2p8affineinvqb_128:
18; X64:       # %bb.0:
19; X64-NEXT:    kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
20; X64-NEXT:    vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xd9,0x03]
21; X64-NEXT:    vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0x89,0xcf,0xe1,0x03]
22; X64-NEXT:    vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0xcf,0xd1,0x03]
23; X64-NEXT:    vpxor %xmm3, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0xef,0xc3]
24; X64-NEXT:    vpxor %xmm0, %xmm4, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xd9,0xef,0xc0]
25; X64-NEXT:    retq # encoding: [0xc3]
26  %1 = bitcast i16 %mask to <16 x i1>
27  %2 = call <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 3)
28  %3 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> zeroinitializer
29  %4 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> %passthru
30  %5 = xor <16 x i8> %3, %4
31  %6 = xor <16 x i8> %5, %2
32  ret <16 x i8> %6
33}
34
35declare <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8>, <32 x i8>, i8)
36define <32 x i8> @test_vgf2p8affineinvqb_256(<32 x i8> %src1, <32 x i8> %src2, <32 x i8> %passthru, i32 %mask) {
37; X86-LABEL: test_vgf2p8affineinvqb_256:
38; X86:       # %bb.0:
39; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x04]
40; X86-NEXT:    vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xd9,0x03]
41; X86-NEXT:    vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0xcf,0xe1,0x03]
42; X86-NEXT:    vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0xcf,0xd1,0x03]
43; X86-NEXT:    vpxor %ymm3, %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xed,0xef,0xc3]
44; X86-NEXT:    vpxor %ymm0, %ymm4, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xdd,0xef,0xc0]
45; X86-NEXT:    retl # encoding: [0xc3]
46;
47; X64-LABEL: test_vgf2p8affineinvqb_256:
48; X64:       # %bb.0:
49; X64-NEXT:    kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
50; X64-NEXT:    vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xd9,0x03]
51; X64-NEXT:    vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0xcf,0xe1,0x03]
52; X64-NEXT:    vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0xcf,0xd1,0x03]
53; X64-NEXT:    vpxor %ymm3, %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xed,0xef,0xc3]
54; X64-NEXT:    vpxor %ymm0, %ymm4, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xdd,0xef,0xc0]
55; X64-NEXT:    retq # encoding: [0xc3]
56  %1 = bitcast i32 %mask to <32 x i1>
57  %2 = call <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 3)
58  %3 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> zeroinitializer
59  %4 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> %passthru
60  %5 = xor <32 x i8> %3, %4
61  %6 = xor <32 x i8> %5, %2
62  ret <32 x i8> %6
63}
64
65declare <64 x i8> @llvm.x86.vgf2p8affineinvqb.512(<64 x i8>, <64 x i8>, i8)
66define <64 x i8> @test_vgf2p8affineinvqb_512(<64 x i8> %src1, <64 x i8> %src2, <64 x i8> %passthru, i64 %mask) {
67; X86-LABEL: test_vgf2p8affineinvqb_512:
68; X86:       # %bb.0:
69; X86-NEXT:    kmovq {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf8,0x90,0x4c,0x24,0x04]
70; X86-NEXT:    vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0xcf,0xd9,0x03]
71; X86-NEXT:    vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0xcf,0xe1,0x03]
72; X86-NEXT:    vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0xcf,0xd1,0x03]
73; X86-NEXT:    vpxorq %zmm3, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0xef,0xc3]
74; X86-NEXT:    vpxorq %zmm0, %zmm4, %zmm0 # encoding: [0x62,0xf1,0xdd,0x48,0xef,0xc0]
75; X86-NEXT:    retl # encoding: [0xc3]
76;
77; X64-LABEL: test_vgf2p8affineinvqb_512:
78; X64:       # %bb.0:
79; X64-NEXT:    kmovq %rdi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
80; X64-NEXT:    vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0xcf,0xd9,0x03]
81; X64-NEXT:    vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0xcf,0xe1,0x03]
82; X64-NEXT:    vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0xcf,0xd1,0x03]
83; X64-NEXT:    vpxorq %zmm3, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0xef,0xc3]
84; X64-NEXT:    vpxorq %zmm0, %zmm4, %zmm0 # encoding: [0x62,0xf1,0xdd,0x48,0xef,0xc0]
85; X64-NEXT:    retq # encoding: [0xc3]
86  %1 = bitcast i64 %mask to <64 x i1>
87  %2 = call <64 x i8> @llvm.x86.vgf2p8affineinvqb.512(<64 x i8> %src1, <64 x i8> %src2, i8 3)
88  %3 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> zeroinitializer
89  %4 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> %passthru
90  %5 = xor <64 x i8> %3, %4
91  %6 = xor <64 x i8> %5, %2
92  ret <64 x i8> %6
93}
94
95declare <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8>, <16 x i8>, i8)
96define <16 x i8> @test_vgf2p8affineqb_128(<16 x i8> %src1, <16 x i8> %src2, <16 x i8> %passthru, i16 %mask) {
97; X86-LABEL: test_vgf2p8affineqb_128:
98; X86:       # %bb.0:
99; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
100; X86-NEXT:    vgf2p8affineqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xd9,0x03]
101; X86-NEXT:    vgf2p8affineqb $3, %xmm1, %xmm0, %xmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0x89,0xce,0xe1,0x03]
102; X86-NEXT:    vgf2p8affineqb $3, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0xce,0xd1,0x03]
103; X86-NEXT:    vpxor %xmm3, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0xef,0xc3]
104; X86-NEXT:    vpxor %xmm0, %xmm4, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xd9,0xef,0xc0]
105; X86-NEXT:    retl # encoding: [0xc3]
106;
107; X64-LABEL: test_vgf2p8affineqb_128:
108; X64:       # %bb.0:
109; X64-NEXT:    kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
110; X64-NEXT:    vgf2p8affineqb $3, %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xd9,0x03]
111; X64-NEXT:    vgf2p8affineqb $3, %xmm1, %xmm0, %xmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0x89,0xce,0xe1,0x03]
112; X64-NEXT:    vgf2p8affineqb $3, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0xce,0xd1,0x03]
113; X64-NEXT:    vpxor %xmm3, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0xef,0xc3]
114; X64-NEXT:    vpxor %xmm0, %xmm4, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xd9,0xef,0xc0]
115; X64-NEXT:    retq # encoding: [0xc3]
116  %1 = bitcast i16 %mask to <16 x i1>
117  %2 = call <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 3)
118  %3 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> zeroinitializer
119  %4 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> %passthru
120  %5 = xor <16 x i8> %3, %4
121  %6 = xor <16 x i8> %5, %2
122  ret <16 x i8> %6
123}
124
125declare <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8>, <32 x i8>, i8)
126define <32 x i8> @test_vgf2p8affineqb_256(<32 x i8> %src1, <32 x i8> %src2, <32 x i8> %passthru, i32 %mask) {
127; X86-LABEL: test_vgf2p8affineqb_256:
128; X86:       # %bb.0:
129; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x04]
130; X86-NEXT:    vgf2p8affineqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xd9,0x03]
131; X86-NEXT:    vgf2p8affineqb $3, %ymm1, %ymm0, %ymm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0xce,0xe1,0x03]
132; X86-NEXT:    vgf2p8affineqb $3, %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0xce,0xd1,0x03]
133; X86-NEXT:    vpxor %ymm3, %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xed,0xef,0xc3]
134; X86-NEXT:    vpxor %ymm0, %ymm4, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xdd,0xef,0xc0]
135; X86-NEXT:    retl # encoding: [0xc3]
136;
137; X64-LABEL: test_vgf2p8affineqb_256:
138; X64:       # %bb.0:
139; X64-NEXT:    kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
140; X64-NEXT:    vgf2p8affineqb $3, %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xd9,0x03]
141; X64-NEXT:    vgf2p8affineqb $3, %ymm1, %ymm0, %ymm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0xce,0xe1,0x03]
142; X64-NEXT:    vgf2p8affineqb $3, %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0xce,0xd1,0x03]
143; X64-NEXT:    vpxor %ymm3, %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xed,0xef,0xc3]
144; X64-NEXT:    vpxor %ymm0, %ymm4, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xdd,0xef,0xc0]
145; X64-NEXT:    retq # encoding: [0xc3]
146  %1 = bitcast i32 %mask to <32 x i1>
147  %2 = call <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 3)
148  %3 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> zeroinitializer
149  %4 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> %passthru
150  %5 = xor <32 x i8> %3, %4
151  %6 = xor <32 x i8> %5, %2
152  ret <32 x i8> %6
153}
154
155declare <64 x i8> @llvm.x86.vgf2p8affineqb.512(<64 x i8>, <64 x i8>, i8)
156define <64 x i8> @test_vgf2p8affineqb_512(<64 x i8> %src1, <64 x i8> %src2, <64 x i8> %passthru, i64 %mask) {
157; X86-LABEL: test_vgf2p8affineqb_512:
158; X86:       # %bb.0:
159; X86-NEXT:    kmovq {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf8,0x90,0x4c,0x24,0x04]
160; X86-NEXT:    vgf2p8affineqb $3, %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0xce,0xd9,0x03]
161; X86-NEXT:    vgf2p8affineqb $3, %zmm1, %zmm0, %zmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0xce,0xe1,0x03]
162; X86-NEXT:    vgf2p8affineqb $3, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0xce,0xd1,0x03]
163; X86-NEXT:    vpxorq %zmm3, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0xef,0xc3]
164; X86-NEXT:    vpxorq %zmm0, %zmm4, %zmm0 # encoding: [0x62,0xf1,0xdd,0x48,0xef,0xc0]
165; X86-NEXT:    retl # encoding: [0xc3]
166;
167; X64-LABEL: test_vgf2p8affineqb_512:
168; X64:       # %bb.0:
169; X64-NEXT:    kmovq %rdi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
170; X64-NEXT:    vgf2p8affineqb $3, %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0xce,0xd9,0x03]
171; X64-NEXT:    vgf2p8affineqb $3, %zmm1, %zmm0, %zmm4 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0xce,0xe1,0x03]
172; X64-NEXT:    vgf2p8affineqb $3, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0xce,0xd1,0x03]
173; X64-NEXT:    vpxorq %zmm3, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0xef,0xc3]
174; X64-NEXT:    vpxorq %zmm0, %zmm4, %zmm0 # encoding: [0x62,0xf1,0xdd,0x48,0xef,0xc0]
175; X64-NEXT:    retq # encoding: [0xc3]
176  %1 = bitcast i64 %mask to <64 x i1>
177  %2 = call <64 x i8> @llvm.x86.vgf2p8affineqb.512(<64 x i8> %src1, <64 x i8> %src2, i8 3)
178  %3 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> zeroinitializer
179  %4 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> %passthru
180  %5 = xor <64 x i8> %3, %4
181  %6 = xor <64 x i8> %5, %2
182  ret <64 x i8> %6
183}
184
185declare <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8>, <16 x i8>)
186define <16 x i8> @test_vgf2p8mulb_128(<16 x i8> %src1, <16 x i8> %src2, <16 x i8> %passthru, i16 %mask) {
187; X86-LABEL: test_vgf2p8mulb_128:
188; X86:       # %bb.0:
189; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
190; X86-NEXT:    vgf2p8mulb %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xcf,0xd9]
191; X86-NEXT:    vgf2p8mulb %xmm1, %xmm0, %xmm4 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0x89,0xcf,0xe1]
192; X86-NEXT:    vgf2p8mulb %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xcf,0xd1]
193; X86-NEXT:    vpxor %xmm3, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0xef,0xc3]
194; X86-NEXT:    vpxor %xmm0, %xmm4, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xd9,0xef,0xc0]
195; X86-NEXT:    retl # encoding: [0xc3]
196;
197; X64-LABEL: test_vgf2p8mulb_128:
198; X64:       # %bb.0:
199; X64-NEXT:    kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
200; X64-NEXT:    vgf2p8mulb %xmm1, %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xcf,0xd9]
201; X64-NEXT:    vgf2p8mulb %xmm1, %xmm0, %xmm4 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0x89,0xcf,0xe1]
202; X64-NEXT:    vgf2p8mulb %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xcf,0xd1]
203; X64-NEXT:    vpxor %xmm3, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0xef,0xc3]
204; X64-NEXT:    vpxor %xmm0, %xmm4, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xd9,0xef,0xc0]
205; X64-NEXT:    retq # encoding: [0xc3]
206  %1 = bitcast i16 %mask to <16 x i1>
207  %2 = call <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8> %src1, <16 x i8> %src2)
208  %3 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> zeroinitializer
209  %4 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> %passthru
210  %5 = xor <16 x i8> %3, %4
211  %6 = xor <16 x i8> %5, %2
212  ret <16 x i8> %6
213}
214
215declare <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8>, <32 x i8>)
216define <32 x i8> @test_vgf2p8mulb_256(<32 x i8> %src1, <32 x i8> %src2, <32 x i8> %passthru, i32 %mask) {
217; X86-LABEL: test_vgf2p8mulb_256:
218; X86:       # %bb.0:
219; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x04]
220; X86-NEXT:    vgf2p8mulb %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0xcf,0xd9]
221; X86-NEXT:    vgf2p8mulb %ymm1, %ymm0, %ymm4 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xa9,0xcf,0xe1]
222; X86-NEXT:    vgf2p8mulb %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xcf,0xd1]
223; X86-NEXT:    vpxor %ymm3, %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xed,0xef,0xc3]
224; X86-NEXT:    vpxor %ymm0, %ymm4, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xdd,0xef,0xc0]
225; X86-NEXT:    retl # encoding: [0xc3]
226;
227; X64-LABEL: test_vgf2p8mulb_256:
228; X64:       # %bb.0:
229; X64-NEXT:    kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
230; X64-NEXT:    vgf2p8mulb %ymm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0xcf,0xd9]
231; X64-NEXT:    vgf2p8mulb %ymm1, %ymm0, %ymm4 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xa9,0xcf,0xe1]
232; X64-NEXT:    vgf2p8mulb %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xcf,0xd1]
233; X64-NEXT:    vpxor %ymm3, %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xed,0xef,0xc3]
234; X64-NEXT:    vpxor %ymm0, %ymm4, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xdd,0xef,0xc0]
235; X64-NEXT:    retq # encoding: [0xc3]
236  %1 = bitcast i32 %mask to <32 x i1>
237  %2 = call <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8> %src1, <32 x i8> %src2)
238  %3 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> zeroinitializer
239  %4 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> %passthru
240  %5 = xor <32 x i8> %3, %4
241  %6 = xor <32 x i8> %5, %2
242  ret <32 x i8> %6
243}
244
245declare <64 x i8> @llvm.x86.vgf2p8mulb.512(<64 x i8>, <64 x i8>)
246define <64 x i8> @test_vgf2p8mulb_512(<64 x i8> %src1, <64 x i8> %src2, <64 x i8> %passthru, i64 %mask) {
247; X86-LABEL: test_vgf2p8mulb_512:
248; X86:       # %bb.0:
249; X86-NEXT:    kmovq {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf8,0x90,0x4c,0x24,0x04]
250; X86-NEXT:    vgf2p8mulb %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf2,0x7d,0x48,0xcf,0xd9]
251; X86-NEXT:    vgf2p8mulb %zmm1, %zmm0, %zmm4 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xc9,0xcf,0xe1]
252; X86-NEXT:    vgf2p8mulb %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0xcf,0xd1]
253; X86-NEXT:    vpxorq %zmm3, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0xef,0xc3]
254; X86-NEXT:    vpxorq %zmm0, %zmm4, %zmm0 # encoding: [0x62,0xf1,0xdd,0x48,0xef,0xc0]
255; X86-NEXT:    retl # encoding: [0xc3]
256;
257; X64-LABEL: test_vgf2p8mulb_512:
258; X64:       # %bb.0:
259; X64-NEXT:    kmovq %rdi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
260; X64-NEXT:    vgf2p8mulb %zmm1, %zmm0, %zmm3 # encoding: [0x62,0xf2,0x7d,0x48,0xcf,0xd9]
261; X64-NEXT:    vgf2p8mulb %zmm1, %zmm0, %zmm4 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xc9,0xcf,0xe1]
262; X64-NEXT:    vgf2p8mulb %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0xcf,0xd1]
263; X64-NEXT:    vpxorq %zmm3, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0xef,0xc3]
264; X64-NEXT:    vpxorq %zmm0, %zmm4, %zmm0 # encoding: [0x62,0xf1,0xdd,0x48,0xef,0xc0]
265; X64-NEXT:    retq # encoding: [0xc3]
266  %1 = bitcast i64 %mask to <64 x i1>
267  %2 = call <64 x i8> @llvm.x86.vgf2p8mulb.512(<64 x i8> %src1, <64 x i8> %src2)
268  %3 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> zeroinitializer
269  %4 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> %passthru
270  %5 = xor <64 x i8> %3, %4
271  %6 = xor <64 x i8> %5, %2
272  ret <64 x i8> %6
273}
274
275