• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc < %s -fast-isel -mtriple=i386-unknown-unknown -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefix=ALL --check-prefix=X32
3; RUN: llc < %s -fast-isel -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefix=ALL --check-prefix=X64
4
5; NOTE: This should use IR equivalent to what is generated by clang/test/CodeGen/avx512bw-builtins.c
6
7define <8 x i64> @test_mm512_broadcastb_epi8(<2 x i64> %a0) {
8; X32-LABEL: test_mm512_broadcastb_epi8:
9; X32:       # BB#0:
10; X32-NEXT:    vpbroadcastb %xmm0, %zmm0
11; X32-NEXT:    retl
12;
13; X64-LABEL: test_mm512_broadcastb_epi8:
14; X64:       # BB#0:
15; X64-NEXT:    vpbroadcastb %xmm0, %zmm0
16; X64-NEXT:    retq
17  %arg0 = bitcast <2 x i64> %a0 to <16 x i8>
18  %res0 = shufflevector <16 x i8> %arg0, <16 x i8> undef, <64 x i32> zeroinitializer
19  %res1 = bitcast <64 x i8> %res0 to <8 x i64>
20  ret <8 x i64> %res1
21}
22
23define <8 x i64> @test_mm512_mask_broadcastb_epi8(<8 x i64> %a0, i64* %a1, <2 x i64> %a2) {
24; X32-LABEL: test_mm512_mask_broadcastb_epi8:
25; X32:       # BB#0:
26; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
27; X32-NEXT:    kmovq (%eax), %k1
28; X32-NEXT:    vpbroadcastb %xmm1, %zmm0 {%k1}
29; X32-NEXT:    retl
30;
31; X64-LABEL: test_mm512_mask_broadcastb_epi8:
32; X64:       # BB#0:
33; X64-NEXT:    kmovq (%rdi), %k1
34; X64-NEXT:    vpbroadcastb %xmm1, %zmm0 {%k1}
35; X64-NEXT:    retq
36  %arg0 = bitcast <8 x i64> %a0 to <64 x i8>
37  %bc1 = bitcast i64* %a1 to <64 x i1>*
38  %arg1 = load <64 x i1>, <64 x i1>* %bc1
39  %arg2 = bitcast <2 x i64> %a2 to <16 x i8>
40  %res0 = shufflevector <16 x i8> %arg2, <16 x i8> undef, <64 x i32> zeroinitializer
41  %res1 = select <64 x i1> %arg1, <64 x i8> %res0, <64 x i8> %arg0
42  %res2 = bitcast <64 x i8> %res1 to <8 x i64>
43  ret <8 x i64> %res2
44}
45
46define <8 x i64> @test_mm512_maskz_broadcastb_epi8(i64* %a0, <2 x i64> %a1) {
47; X32-LABEL: test_mm512_maskz_broadcastb_epi8:
48; X32:       # BB#0:
49; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
50; X32-NEXT:    kmovq (%eax), %k1
51; X32-NEXT:    vpbroadcastb %xmm0, %zmm0 {%k1} {z}
52; X32-NEXT:    retl
53;
54; X64-LABEL: test_mm512_maskz_broadcastb_epi8:
55; X64:       # BB#0:
56; X64-NEXT:    kmovq (%rdi), %k1
57; X64-NEXT:    vpbroadcastb %xmm0, %zmm0 {%k1} {z}
58; X64-NEXT:    retq
59  %bc0 = bitcast i64* %a0 to <64 x i1>*
60  %arg0 = load <64 x i1>, <64 x i1>* %bc0
61  %arg1 = bitcast <2 x i64> %a1 to <16 x i8>
62  %res0 = shufflevector <16 x i8> %arg1, <16 x i8> undef, <64 x i32> zeroinitializer
63  %res1 = select <64 x i1> %arg0, <64 x i8> %res0, <64 x i8> zeroinitializer
64  %res2 = bitcast <64 x i8> %res1 to <8 x i64>
65  ret <8 x i64> %res2
66}
67
68define <8 x i64> @test_mm512_broadcastw_epi16(<2 x i64> %a0) {
69; X32-LABEL: test_mm512_broadcastw_epi16:
70; X32:       # BB#0:
71; X32-NEXT:    vpbroadcastw %xmm0, %zmm0
72; X32-NEXT:    retl
73;
74; X64-LABEL: test_mm512_broadcastw_epi16:
75; X64:       # BB#0:
76; X64-NEXT:    vpbroadcastw %xmm0, %zmm0
77; X64-NEXT:    retq
78  %arg0 = bitcast <2 x i64> %a0 to <8 x i16>
79  %res0 = shufflevector <8 x i16> %arg0, <8 x i16> undef, <32 x i32> zeroinitializer
80  %res1 = bitcast <32 x i16> %res0 to <8 x i64>
81  ret <8 x i64> %res1
82}
83
84define <8 x i64> @test_mm512_mask_broadcastw_epi16(<8 x i64> %a0, i32 %a1, <2 x i64> %a2) {
85; X32-LABEL: test_mm512_mask_broadcastw_epi16:
86; X32:       # BB#0:
87; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
88; X32-NEXT:    kmovd %eax, %k1
89; X32-NEXT:    vpbroadcastw %xmm1, %zmm0 {%k1}
90; X32-NEXT:    retl
91;
92; X64-LABEL: test_mm512_mask_broadcastw_epi16:
93; X64:       # BB#0:
94; X64-NEXT:    kmovd %edi, %k1
95; X64-NEXT:    vpbroadcastw %xmm1, %zmm0 {%k1}
96; X64-NEXT:    retq
97  %arg0 = bitcast <8 x i64> %a0 to <32 x i16>
98  %arg1 = bitcast i32 %a1 to <32 x i1>
99  %arg2 = bitcast <2 x i64> %a2 to <8 x i16>
100  %res0 = shufflevector <8 x i16> %arg2, <8 x i16> undef, <32 x i32> zeroinitializer
101  %res1 = select <32 x i1> %arg1, <32 x i16> %res0, <32 x i16> %arg0
102  %res2 = bitcast <32 x i16> %res1 to <8 x i64>
103  ret <8 x i64> %res2
104}
105
106define <8 x i64> @test_mm512_maskz_broadcastw_epi16(i32 %a0, <2 x i64> %a1) {
107; X32-LABEL: test_mm512_maskz_broadcastw_epi16:
108; X32:       # BB#0:
109; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
110; X32-NEXT:    kmovd %eax, %k1
111; X32-NEXT:    vpbroadcastw %xmm0, %zmm0 {%k1} {z}
112; X32-NEXT:    retl
113;
114; X64-LABEL: test_mm512_maskz_broadcastw_epi16:
115; X64:       # BB#0:
116; X64-NEXT:    kmovd %edi, %k1
117; X64-NEXT:    vpbroadcastw %xmm0, %zmm0 {%k1} {z}
118; X64-NEXT:    retq
119  %arg0 = bitcast i32 %a0 to <32 x i1>
120  %arg1 = bitcast <2 x i64> %a1 to <8 x i16>
121  %res0 = shufflevector <8 x i16> %arg1, <8 x i16> undef, <32 x i32> zeroinitializer
122  %res1 = select <32 x i1> %arg0, <32 x i16> %res0, <32 x i16> zeroinitializer
123  %res2 = bitcast <32 x i16> %res1 to <8 x i64>
124  ret <8 x i64> %res2
125}
126
127define <8 x i64> @test_mm512_bslli_epi128(<8 x i64> %a0) {
128; X32-LABEL: test_mm512_bslli_epi128:
129; X32:       # BB#0:
130; X32-NEXT:    vpsrldq {{.*#+}} zmm0 = zmm0[11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[27,28,29,30,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[43,44,45,46,47],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[59,60,61,62,63],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
131; X32-NEXT:    retl
132;
133; X64-LABEL: test_mm512_bslli_epi128:
134; X64:       # BB#0:
135; X64-NEXT:    vpsrldq {{.*#+}} zmm0 = zmm0[11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[27,28,29,30,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[43,44,45,46,47],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[59,60,61,62,63],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
136; X64-NEXT:    retq
137  %arg0 = bitcast <8 x i64> %a0 to <64 x i8>
138  %res0 = shufflevector <64 x i8> %arg0, <64 x i8> zeroinitializer, <64 x i32> <i32 11, i32 12, i32 13, i32 14, i32 15, i32 64, i32 65, i32 66, i32 67, i32 68, i32 69, i32 70, i32 71, i32 72, i32 73, i32 74, i32 27, i32 28, i32 29, i32 30, i32 31, i32 80, i32 81, i32 82, i32 83, i32 84, i32 85, i32 86, i32 87, i32 88, i32 89, i32 90, i32 43, i32 44, i32 45, i32 46, i32 47, i32 96, i32 97, i32 98, i32 99, i32 100, i32 101, i32 102, i32 103, i32 104, i32 105, i32 106, i32 59, i32 60, i32 61, i32 62, i32 63, i32 112, i32 113, i32 114, i32 115, i32 116, i32 117, i32 118, i32 119, i32 120, i32 121, i32 122>
139  %res1 = bitcast <64 x i8> %res0 to <8 x i64>
140  ret <8 x i64> %res1
141}
142
143define <8 x i64> @test_mm512_bsrli_epi128(<8 x i64> %a0) {
144; X32-LABEL: test_mm512_bsrli_epi128:
145; X32:       # BB#0:
146; X32-NEXT:    vpsrldq {{.*#+}} zmm0 = zmm0[5,6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zmm0[21,22,23,24,25,26,27,28,29,30,31],zero,zero,zero,zero,zero,zmm0[37,38,39,40,41,42,43,44,45,46,47],zero,zero,zero,zero,zero,zmm0[53,54,55,56,57,58,59,60,61,62,63],zero,zero,zero,zero,zero
147; X32-NEXT:    retl
148;
149; X64-LABEL: test_mm512_bsrli_epi128:
150; X64:       # BB#0:
151; X64-NEXT:    vpsrldq {{.*#+}} zmm0 = zmm0[5,6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zmm0[21,22,23,24,25,26,27,28,29,30,31],zero,zero,zero,zero,zero,zmm0[37,38,39,40,41,42,43,44,45,46,47],zero,zero,zero,zero,zero,zmm0[53,54,55,56,57,58,59,60,61,62,63],zero,zero,zero,zero,zero
152; X64-NEXT:    retq
153  %arg0 = bitcast <8 x i64> %a0 to <64 x i8>
154  %res0 = shufflevector <64 x i8> %arg0, <64 x i8> zeroinitializer, <64 x i32> <i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 64, i32 65, i32 66, i32 67, i32 68, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 80, i32 81, i32 82, i32 83, i32 84, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 96, i32 97, i32 98, i32 99, i32 100, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63, i32 112, i32 113, i32 114, i32 115, i32 116>
155  %res1 = bitcast <64 x i8> %res0 to <8 x i64>
156  ret <8 x i64> %res1
157}
158
159define <8 x i64> @test_mm512_unpackhi_epi8(<8 x i64> %a0, <8 x i64> %a1) {
160; X32-LABEL: test_mm512_unpackhi_epi8:
161; X32:       # BB#0:
162; X32-NEXT:    vpunpckhbw {{.*#+}} zmm0 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
163; X32-NEXT:    retl
164;
165; X64-LABEL: test_mm512_unpackhi_epi8:
166; X64:       # BB#0:
167; X64-NEXT:    vpunpckhbw {{.*#+}} zmm0 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
168; X64-NEXT:    retq
169  %arg0 = bitcast <8 x i64> %a0 to <64 x i8>
170  %arg1 = bitcast <8 x i64> %a1 to <64 x i8>
171  %res0 = shufflevector <64 x i8> %arg0, <64 x i8> %arg1, <64 x i32> <i32 8, i32 72, i32 9, i32 73, i32 10, i32 74, i32 11, i32 75, i32 12, i32 76, i32 13, i32 77, i32 14, i32 78, i32 15, i32 79, i32 24, i32 88, i32 25, i32 89, i32 26, i32 90, i32 27, i32 91, i32 28, i32 92, i32 29, i32 93, i32 30, i32 94, i32 31, i32 95, i32 40, i32 104, i32 41, i32 105, i32 42, i32 106, i32 43, i32 107, i32 44, i32 108, i32 45, i32 109, i32 46, i32 110, i32 47, i32 111, i32 56, i32 120, i32 57, i32 121, i32 58, i32 122, i32 59, i32 123, i32 60, i32 124, i32 61, i32 125, i32 62, i32 126, i32 63, i32 127>
172  %res1 = bitcast <64 x i8> %res0 to <8 x i64>
173  ret <8 x i64> %res1
174}
175
176; TODO - improve support for i64 -> mmask64 on 32-bit targets
177define <8 x i64> @test_mm512_mask_unpackhi_epi8(<8 x i64> %a0, i64* %a1, <8 x i64> %a2, <8 x i64> %a3) {
178; X32-LABEL: test_mm512_mask_unpackhi_epi8:
179; X32:       # BB#0:
180; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
181; X32-NEXT:    kmovq (%eax), %k1
182; X32-NEXT:    vpunpckhbw {{.*#+}} zmm0 {%k1} = zmm1[8],zmm2[8],zmm1[9],zmm2[9],zmm1[10],zmm2[10],zmm1[11],zmm2[11],zmm1[12],zmm2[12],zmm1[13],zmm2[13],zmm1[14],zmm2[14],zmm1[15],zmm2[15],zmm1[24],zmm2[24],zmm1[25],zmm2[25],zmm1[26],zmm2[26],zmm1[27],zmm2[27],zmm1[28],zmm2[28],zmm1[29],zmm2[29],zmm1[30],zmm2[30],zmm1[31],zmm2[31],zmm1[40],zmm2[40],zmm1[41],zmm2[41],zmm1[42],zmm2[42],zmm1[43],zmm2[43],zmm1[44],zmm2[44],zmm1[45],zmm2[45],zmm1[46],zmm2[46],zmm1[47],zmm2[47],zmm1[56],zmm2[56],zmm1[57],zmm2[57],zmm1[58],zmm2[58],zmm1[59],zmm2[59],zmm1[60],zmm2[60],zmm1[61],zmm2[61],zmm1[62],zmm2[62],zmm1[63],zmm2[63]
183; X32-NEXT:    retl
184;
185; X64-LABEL: test_mm512_mask_unpackhi_epi8:
186; X64:       # BB#0:
187; X64-NEXT:    kmovq (%rdi), %k1
188; X64-NEXT:    vpunpckhbw {{.*#+}} zmm0 {%k1} = zmm1[8],zmm2[8],zmm1[9],zmm2[9],zmm1[10],zmm2[10],zmm1[11],zmm2[11],zmm1[12],zmm2[12],zmm1[13],zmm2[13],zmm1[14],zmm2[14],zmm1[15],zmm2[15],zmm1[24],zmm2[24],zmm1[25],zmm2[25],zmm1[26],zmm2[26],zmm1[27],zmm2[27],zmm1[28],zmm2[28],zmm1[29],zmm2[29],zmm1[30],zmm2[30],zmm1[31],zmm2[31],zmm1[40],zmm2[40],zmm1[41],zmm2[41],zmm1[42],zmm2[42],zmm1[43],zmm2[43],zmm1[44],zmm2[44],zmm1[45],zmm2[45],zmm1[46],zmm2[46],zmm1[47],zmm2[47],zmm1[56],zmm2[56],zmm1[57],zmm2[57],zmm1[58],zmm2[58],zmm1[59],zmm2[59],zmm1[60],zmm2[60],zmm1[61],zmm2[61],zmm1[62],zmm2[62],zmm1[63],zmm2[63]
189; X64-NEXT:    retq
190  %arg0 = bitcast <8 x i64> %a0 to <64 x i8>
191  %arg1 = bitcast i64* %a1 to <64 x i1>*
192  %sel1 = load <64 x i1>, <64 x i1>* %arg1
193  %arg2 = bitcast <8 x i64> %a2 to <64 x i8>
194  %arg3 = bitcast <8 x i64> %a3 to <64 x i8>
195  %res0 = shufflevector <64 x i8> %arg2, <64 x i8> %arg3, <64 x i32> <i32 8, i32 72, i32 9, i32 73, i32 10, i32 74, i32 11, i32 75, i32 12, i32 76, i32 13, i32 77, i32 14, i32 78, i32 15, i32 79, i32 24, i32 88, i32 25, i32 89, i32 26, i32 90, i32 27, i32 91, i32 28, i32 92, i32 29, i32 93, i32 30, i32 94, i32 31, i32 95, i32 40, i32 104, i32 41, i32 105, i32 42, i32 106, i32 43, i32 107, i32 44, i32 108, i32 45, i32 109, i32 46, i32 110, i32 47, i32 111, i32 56, i32 120, i32 57, i32 121, i32 58, i32 122, i32 59, i32 123, i32 60, i32 124, i32 61, i32 125, i32 62, i32 126, i32 63, i32 127>
196  %res1 = select <64 x i1> %sel1, <64 x i8> %res0, <64 x i8> %arg0
197  %res2 = bitcast <64 x i8> %res1 to <8 x i64>
198  ret <8 x i64> %res2
199}
200
201define <8 x i64> @test_mm512_maskz_unpackhi_epi8(i64* %a0, <8 x i64> %a1, <8 x i64> %a2) {
202; X32-LABEL: test_mm512_maskz_unpackhi_epi8:
203; X32:       # BB#0:
204; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
205; X32-NEXT:    kmovq (%eax), %k1
206; X32-NEXT:    vpunpckhbw {{.*#+}} zmm0 {%k1} {z} = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
207; X32-NEXT:    retl
208;
209; X64-LABEL: test_mm512_maskz_unpackhi_epi8:
210; X64:       # BB#0:
211; X64-NEXT:    kmovq (%rdi), %k1
212; X64-NEXT:    vpunpckhbw {{.*#+}} zmm0 {%k1} {z} = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
213; X64-NEXT:    retq
214  %arg0 = bitcast i64* %a0 to <64 x i1>*
215  %sel0 = load <64 x i1>, <64 x i1>* %arg0
216  %arg1 = bitcast <8 x i64> %a1 to <64 x i8>
217  %arg2 = bitcast <8 x i64> %a2 to <64 x i8>
218  %res0 = shufflevector <64 x i8> %arg1, <64 x i8> %arg2, <64 x i32> <i32 8, i32 72, i32 9, i32 73, i32 10, i32 74, i32 11, i32 75, i32 12, i32 76, i32 13, i32 77, i32 14, i32 78, i32 15, i32 79, i32 24, i32 88, i32 25, i32 89, i32 26, i32 90, i32 27, i32 91, i32 28, i32 92, i32 29, i32 93, i32 30, i32 94, i32 31, i32 95, i32 40, i32 104, i32 41, i32 105, i32 42, i32 106, i32 43, i32 107, i32 44, i32 108, i32 45, i32 109, i32 46, i32 110, i32 47, i32 111, i32 56, i32 120, i32 57, i32 121, i32 58, i32 122, i32 59, i32 123, i32 60, i32 124, i32 61, i32 125, i32 62, i32 126, i32 63, i32 127>
219  %res1 = select <64 x i1> %sel0, <64 x i8> %res0, <64 x i8> zeroinitializer
220  %res2 = bitcast <64 x i8> %res1 to <8 x i64>
221  ret <8 x i64> %res2
222}
223
224define <8 x i64> @test_mm512_unpackhi_epi16(<8 x i64> %a0, <8 x i64> %a1) {
225; X32-LABEL: test_mm512_unpackhi_epi16:
226; X32:       # BB#0:
227; X32-NEXT:    vpunpckhwd {{.*#+}} zmm0 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31]
228; X32-NEXT:    retl
229;
230; X64-LABEL: test_mm512_unpackhi_epi16:
231; X64:       # BB#0:
232; X64-NEXT:    vpunpckhwd {{.*#+}} zmm0 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31]
233; X64-NEXT:    retq
234  %arg0 = bitcast <8 x i64> %a0 to <32 x i16>
235  %arg1 = bitcast <8 x i64> %a1 to <32 x i16>
236  %res0 = shufflevector <32 x i16> %arg0, <32 x i16> %arg1, <32 x i32> <i32 4, i32 36, i32 5, i32 37, i32 6, i32 38, i32 7, i32 39, i32 12, i32 44, i32 13, i32 45, i32 14, i32 46, i32 15, i32 47, i32 20, i32 52, i32 21, i32 53, i32 22, i32 54, i32 23, i32 55, i32 28, i32 60, i32 29, i32 61, i32 30, i32 62, i32 31, i32 63>
237  %res1 = bitcast <32 x i16> %res0 to <8 x i64>
238  ret <8 x i64> %res1
239}
240
241define <8 x i64> @test_mm512_mask_unpackhi_epi16(<8 x i64> %a0, i32 %a1, <8 x i64> %a2, <8 x i64> %a3) {
242; X32-LABEL: test_mm512_mask_unpackhi_epi16:
243; X32:       # BB#0:
244; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
245; X32-NEXT:    kmovd %eax, %k1
246; X32-NEXT:    vpunpckhwd {{.*#+}} zmm0 {%k1} = zmm1[4],zmm2[4],zmm1[5],zmm2[5],zmm1[6],zmm2[6],zmm1[7],zmm2[7],zmm1[12],zmm2[12],zmm1[13],zmm2[13],zmm1[14],zmm2[14],zmm1[15],zmm2[15],zmm1[20],zmm2[20],zmm1[21],zmm2[21],zmm1[22],zmm2[22],zmm1[23],zmm2[23],zmm1[28],zmm2[28],zmm1[29],zmm2[29],zmm1[30],zmm2[30],zmm1[31],zmm2[31]
247; X32-NEXT:    retl
248;
249; X64-LABEL: test_mm512_mask_unpackhi_epi16:
250; X64:       # BB#0:
251; X64-NEXT:    kmovd %edi, %k1
252; X64-NEXT:    vpunpckhwd {{.*#+}} zmm0 {%k1} = zmm1[4],zmm2[4],zmm1[5],zmm2[5],zmm1[6],zmm2[6],zmm1[7],zmm2[7],zmm1[12],zmm2[12],zmm1[13],zmm2[13],zmm1[14],zmm2[14],zmm1[15],zmm2[15],zmm1[20],zmm2[20],zmm1[21],zmm2[21],zmm1[22],zmm2[22],zmm1[23],zmm2[23],zmm1[28],zmm2[28],zmm1[29],zmm2[29],zmm1[30],zmm2[30],zmm1[31],zmm2[31]
253; X64-NEXT:    retq
254  %arg0 = bitcast <8 x i64> %a0 to <32 x i16>
255  %arg1 = bitcast i32 %a1 to <32 x i1>
256  %arg2 = bitcast <8 x i64> %a2 to <32 x i16>
257  %arg3 = bitcast <8 x i64> %a3 to <32 x i16>
258  %res0 = shufflevector <32 x i16> %arg2, <32 x i16> %arg3, <32 x i32> <i32 4, i32 36, i32 5, i32 37, i32 6, i32 38, i32 7, i32 39, i32 12, i32 44, i32 13, i32 45, i32 14, i32 46, i32 15, i32 47, i32 20, i32 52, i32 21, i32 53, i32 22, i32 54, i32 23, i32 55, i32 28, i32 60, i32 29, i32 61, i32 30, i32 62, i32 31, i32 63>
259  %res1 = select <32 x i1> %arg1, <32 x i16> %res0, <32 x i16> %arg0
260  %res2 = bitcast <32 x i16> %res1 to <8 x i64>
261  ret <8 x i64> %res2
262}
263
264define <8 x i64> @test_mm512_maskz_unpackhi_epi16(i32 %a0, <8 x i64> %a1, <8 x i64> %a2) {
265; X32-LABEL: test_mm512_maskz_unpackhi_epi16:
266; X32:       # BB#0:
267; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
268; X32-NEXT:    kmovd %eax, %k1
269; X32-NEXT:    vpunpckhwd {{.*#+}} zmm0 {%k1} {z} = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31]
270; X32-NEXT:    retl
271;
272; X64-LABEL: test_mm512_maskz_unpackhi_epi16:
273; X64:       # BB#0:
274; X64-NEXT:    kmovd %edi, %k1
275; X64-NEXT:    vpunpckhwd {{.*#+}} zmm0 {%k1} {z} = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31]
276; X64-NEXT:    retq
277  %arg0 = bitcast i32 %a0 to <32 x i1>
278  %arg1 = bitcast <8 x i64> %a1 to <32 x i16>
279  %arg2 = bitcast <8 x i64> %a2 to <32 x i16>
280  %res0 = shufflevector <32 x i16> %arg1, <32 x i16> %arg2, <32 x i32> <i32 4, i32 36, i32 5, i32 37, i32 6, i32 38, i32 7, i32 39, i32 12, i32 44, i32 13, i32 45, i32 14, i32 46, i32 15, i32 47, i32 20, i32 52, i32 21, i32 53, i32 22, i32 54, i32 23, i32 55, i32 28, i32 60, i32 29, i32 61, i32 30, i32 62, i32 31, i32 63>
281  %res1 = select <32 x i1> %arg0, <32 x i16> %res0, <32 x i16> zeroinitializer
282  %res2 = bitcast <32 x i16> %res1 to <8 x i64>
283  ret <8 x i64> %res2
284}
285
286define <8 x i64> @test_mm512_unpacklo_epi8(<8 x i64> %a0, <8 x i64> %a1) {
287; X32-LABEL: test_mm512_unpacklo_epi8:
288; X32:       # BB#0:
289; X32-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
290; X32-NEXT:    retl
291;
292; X64-LABEL: test_mm512_unpacklo_epi8:
293; X64:       # BB#0:
294; X64-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
295; X64-NEXT:    retq
296  %arg0 = bitcast <8 x i64> %a0 to <64 x i8>
297  %arg1 = bitcast <8 x i64> %a1 to <64 x i8>
298  %res0 = shufflevector <64 x i8> %arg0, <64 x i8> %arg1, <64 x i32> <i32 0, i32 64, i32 1, i32 65, i32 2, i32 66, i32 3, i32 67, i32 4, i32 68, i32 5, i32 69, i32 6, i32 70, i32 7, i32 71, i32 16, i32 80, i32 17, i32 81, i32 18, i32 82, i32 19, i32 83, i32 20, i32 84, i32 21, i32 85, i32 22, i32 86, i32 23, i32 87, i32 32, i32 96, i32 33, i32 97, i32 34, i32 98, i32 35, i32 99, i32 36, i32 100, i32 37, i32 101, i32 38, i32 102, i32 39, i32 103, i32 48, i32 112, i32 49, i32 113, i32 50, i32 114, i32 51, i32 115, i32 52, i32 116, i32 53, i32 117, i32 54, i32 118, i32 55, i32 119>
299  %res1 = bitcast <64 x i8> %res0 to <8 x i64>
300  ret <8 x i64> %res1
301}
302
303define <8 x i64> @test_mm512_mask_unpacklo_epi8(<8 x i64> %a0, i64* %a1, <8 x i64> %a2, <8 x i64> %a3) {
304; X32-LABEL: test_mm512_mask_unpacklo_epi8:
305; X32:       # BB#0:
306; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
307; X32-NEXT:    kmovq (%eax), %k1
308; X32-NEXT:    vpunpcklbw {{.*#+}} zmm0 {%k1} = zmm1[0],zmm2[0],zmm1[1],zmm2[1],zmm1[2],zmm2[2],zmm1[3],zmm2[3],zmm1[4],zmm2[4],zmm1[5],zmm2[5],zmm1[6],zmm2[6],zmm1[7],zmm2[7],zmm1[16],zmm2[16],zmm1[17],zmm2[17],zmm1[18],zmm2[18],zmm1[19],zmm2[19],zmm1[20],zmm2[20],zmm1[21],zmm2[21],zmm1[22],zmm2[22],zmm1[23],zmm2[23],zmm1[32],zmm2[32],zmm1[33],zmm2[33],zmm1[34],zmm2[34],zmm1[35],zmm2[35],zmm1[36],zmm2[36],zmm1[37],zmm2[37],zmm1[38],zmm2[38],zmm1[39],zmm2[39],zmm1[48],zmm2[48],zmm1[49],zmm2[49],zmm1[50],zmm2[50],zmm1[51],zmm2[51],zmm1[52],zmm2[52],zmm1[53],zmm2[53],zmm1[54],zmm2[54],zmm1[55],zmm2[55]
309; X32-NEXT:    retl
310;
311; X64-LABEL: test_mm512_mask_unpacklo_epi8:
312; X64:       # BB#0:
313; X64-NEXT:    kmovq (%rdi), %k1
314; X64-NEXT:    vpunpcklbw {{.*#+}} zmm0 {%k1} = zmm1[0],zmm2[0],zmm1[1],zmm2[1],zmm1[2],zmm2[2],zmm1[3],zmm2[3],zmm1[4],zmm2[4],zmm1[5],zmm2[5],zmm1[6],zmm2[6],zmm1[7],zmm2[7],zmm1[16],zmm2[16],zmm1[17],zmm2[17],zmm1[18],zmm2[18],zmm1[19],zmm2[19],zmm1[20],zmm2[20],zmm1[21],zmm2[21],zmm1[22],zmm2[22],zmm1[23],zmm2[23],zmm1[32],zmm2[32],zmm1[33],zmm2[33],zmm1[34],zmm2[34],zmm1[35],zmm2[35],zmm1[36],zmm2[36],zmm1[37],zmm2[37],zmm1[38],zmm2[38],zmm1[39],zmm2[39],zmm1[48],zmm2[48],zmm1[49],zmm2[49],zmm1[50],zmm2[50],zmm1[51],zmm2[51],zmm1[52],zmm2[52],zmm1[53],zmm2[53],zmm1[54],zmm2[54],zmm1[55],zmm2[55]
315; X64-NEXT:    retq
316  %arg0 = bitcast <8 x i64> %a0 to <64 x i8>
317  %arg1 = bitcast i64* %a1 to <64 x i1>*
318  %sel1 = load <64 x i1>, <64 x i1>* %arg1
319  %arg2 = bitcast <8 x i64> %a2 to <64 x i8>
320  %arg3 = bitcast <8 x i64> %a3 to <64 x i8>
321  %res0 = shufflevector <64 x i8> %arg2, <64 x i8> %arg3, <64 x i32> <i32 0, i32 64, i32 1, i32 65, i32 2, i32 66, i32 3, i32 67, i32 4, i32 68, i32 5, i32 69, i32 6, i32 70, i32 7, i32 71, i32 16, i32 80, i32 17, i32 81, i32 18, i32 82, i32 19, i32 83, i32 20, i32 84, i32 21, i32 85, i32 22, i32 86, i32 23, i32 87, i32 32, i32 96, i32 33, i32 97, i32 34, i32 98, i32 35, i32 99, i32 36, i32 100, i32 37, i32 101, i32 38, i32 102, i32 39, i32 103, i32 48, i32 112, i32 49, i32 113, i32 50, i32 114, i32 51, i32 115, i32 52, i32 116, i32 53, i32 117, i32 54, i32 118, i32 55, i32 119>
322  %res1 = select <64 x i1> %sel1, <64 x i8> %res0, <64 x i8> %arg0
323  %res2 = bitcast <64 x i8> %res1 to <8 x i64>
324  ret <8 x i64> %res2
325}
326
327define <8 x i64> @test_mm512_maskz_unpacklo_epi8(i64* %a0, <8 x i64> %a1, <8 x i64> %a2) {
328; X32-LABEL: test_mm512_maskz_unpacklo_epi8:
329; X32:       # BB#0:
330; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
331; X32-NEXT:    kmovq (%eax), %k1
332; X32-NEXT:    vpunpcklbw {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
333; X32-NEXT:    retl
334;
335; X64-LABEL: test_mm512_maskz_unpacklo_epi8:
336; X64:       # BB#0:
337; X64-NEXT:    kmovq (%rdi), %k1
338; X64-NEXT:    vpunpcklbw {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
339; X64-NEXT:    retq
340  %arg0 = bitcast i64* %a0 to <64 x i1>*
341  %sel0 = load <64 x i1>, <64 x i1>* %arg0
342  %arg1 = bitcast <8 x i64> %a1 to <64 x i8>
343  %arg2 = bitcast <8 x i64> %a2 to <64 x i8>
344  %res0 = shufflevector <64 x i8> %arg1, <64 x i8> %arg2, <64 x i32> <i32 0, i32 64, i32 1, i32 65, i32 2, i32 66, i32 3, i32 67, i32 4, i32 68, i32 5, i32 69, i32 6, i32 70, i32 7, i32 71, i32 16, i32 80, i32 17, i32 81, i32 18, i32 82, i32 19, i32 83, i32 20, i32 84, i32 21, i32 85, i32 22, i32 86, i32 23, i32 87, i32 32, i32 96, i32 33, i32 97, i32 34, i32 98, i32 35, i32 99, i32 36, i32 100, i32 37, i32 101, i32 38, i32 102, i32 39, i32 103, i32 48, i32 112, i32 49, i32 113, i32 50, i32 114, i32 51, i32 115, i32 52, i32 116, i32 53, i32 117, i32 54, i32 118, i32 55, i32 119>
345  %res1 = select <64 x i1> %sel0, <64 x i8> %res0, <64 x i8> zeroinitializer
346  %res2 = bitcast <64 x i8> %res1 to <8 x i64>
347  ret <8 x i64> %res2
348}
349
350define <8 x i64> @test_mm512_unpacklo_epi16(<8 x i64> %a0, <8 x i64> %a1) {
351; X32-LABEL: test_mm512_unpacklo_epi16:
352; X32:       # BB#0:
353; X32-NEXT:    vpunpcklwd {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27]
354; X32-NEXT:    retl
355;
356; X64-LABEL: test_mm512_unpacklo_epi16:
357; X64:       # BB#0:
358; X64-NEXT:    vpunpcklwd {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27]
359; X64-NEXT:    retq
360  %arg0 = bitcast <8 x i64> %a0 to <32 x i16>
361  %arg1 = bitcast <8 x i64> %a1 to <32 x i16>
362  %res0 = shufflevector <32 x i16> %arg0, <32 x i16> %arg1, <32 x i32> <i32 0, i32 32, i32 1, i32 33, i32 2, i32 34, i32 3, i32 35, i32 8, i32 40, i32 9, i32 41, i32 10, i32 42, i32 11, i32 43, i32 16, i32 48, i32 17, i32 49, i32 18, i32 50, i32 19, i32 51, i32 24, i32 56, i32 25, i32 57, i32 26, i32 58, i32 27, i32 59>
363  %res1 = bitcast <32 x i16> %res0 to <8 x i64>
364  ret <8 x i64> %res1
365}
366
367define <8 x i64> @test_mm512_mask_unpacklo_epi16(<8 x i64> %a0, i32 %a1, <8 x i64> %a2, <8 x i64> %a3) {
368; X32-LABEL: test_mm512_mask_unpacklo_epi16:
369; X32:       # BB#0:
370; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
371; X32-NEXT:    kmovd %eax, %k1
372; X32-NEXT:    vpunpcklwd {{.*#+}} zmm0 {%k1} = zmm1[0],zmm2[0],zmm1[1],zmm2[1],zmm1[2],zmm2[2],zmm1[3],zmm2[3],zmm1[8],zmm2[8],zmm1[9],zmm2[9],zmm1[10],zmm2[10],zmm1[11],zmm2[11],zmm1[16],zmm2[16],zmm1[17],zmm2[17],zmm1[18],zmm2[18],zmm1[19],zmm2[19],zmm1[24],zmm2[24],zmm1[25],zmm2[25],zmm1[26],zmm2[26],zmm1[27],zmm2[27]
373; X32-NEXT:    retl
374;
375; X64-LABEL: test_mm512_mask_unpacklo_epi16:
376; X64:       # BB#0:
377; X64-NEXT:    kmovd %edi, %k1
378; X64-NEXT:    vpunpcklwd {{.*#+}} zmm0 {%k1} = zmm1[0],zmm2[0],zmm1[1],zmm2[1],zmm1[2],zmm2[2],zmm1[3],zmm2[3],zmm1[8],zmm2[8],zmm1[9],zmm2[9],zmm1[10],zmm2[10],zmm1[11],zmm2[11],zmm1[16],zmm2[16],zmm1[17],zmm2[17],zmm1[18],zmm2[18],zmm1[19],zmm2[19],zmm1[24],zmm2[24],zmm1[25],zmm2[25],zmm1[26],zmm2[26],zmm1[27],zmm2[27]
379; X64-NEXT:    retq
380  %arg0 = bitcast <8 x i64> %a0 to <32 x i16>
381  %arg1 = bitcast i32 %a1 to <32 x i1>
382  %arg2 = bitcast <8 x i64> %a2 to <32 x i16>
383  %arg3 = bitcast <8 x i64> %a3 to <32 x i16>
384  %res0 = shufflevector <32 x i16> %arg2, <32 x i16> %arg3, <32 x i32> <i32 0, i32 32, i32 1, i32 33, i32 2, i32 34, i32 3, i32 35, i32 8, i32 40, i32 9, i32 41, i32 10, i32 42, i32 11, i32 43, i32 16, i32 48, i32 17, i32 49, i32 18, i32 50, i32 19, i32 51, i32 24, i32 56, i32 25, i32 57, i32 26, i32 58, i32 27, i32 59>
385  %res1 = select <32 x i1> %arg1, <32 x i16> %res0, <32 x i16> %arg0
386  %res2 = bitcast <32 x i16> %res1 to <8 x i64>
387  ret <8 x i64> %res2
388}
389
390define <8 x i64> @test_mm512_maskz_unpacklo_epi16(i32 %a0, <8 x i64> %a1, <8 x i64> %a2) {
391; X32-LABEL: test_mm512_maskz_unpacklo_epi16:
392; X32:       # BB#0:
393; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
394; X32-NEXT:    kmovd %eax, %k1
395; X32-NEXT:    vpunpcklwd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27]
396; X32-NEXT:    retl
397;
398; X64-LABEL: test_mm512_maskz_unpacklo_epi16:
399; X64:       # BB#0:
400; X64-NEXT:    kmovd %edi, %k1
401; X64-NEXT:    vpunpcklwd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27]
402; X64-NEXT:    retq
403  %arg0 = bitcast i32 %a0 to <32 x i1>
404  %arg1 = bitcast <8 x i64> %a1 to <32 x i16>
405  %arg2 = bitcast <8 x i64> %a2 to <32 x i16>
406  %res0 = shufflevector <32 x i16> %arg1, <32 x i16> %arg2, <32 x i32> <i32 0, i32 32, i32 1, i32 33, i32 2, i32 34, i32 3, i32 35, i32 8, i32 40, i32 9, i32 41, i32 10, i32 42, i32 11, i32 43, i32 16, i32 48, i32 17, i32 49, i32 18, i32 50, i32 19, i32 51, i32 24, i32 56, i32 25, i32 57, i32 26, i32 58, i32 27, i32 59>
407  %res1 = select <32 x i1> %arg0, <32 x i16> %res0, <32 x i16> zeroinitializer
408  %res2 = bitcast <32 x i16> %res1 to <8 x i64>
409  ret <8 x i64> %res2
410}
411
412!0 = !{i32 1}
413
414