1 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx -x86-experimental-vector-shuffle-lowering | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1
2 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx2 -x86-experimental-vector-shuffle-lowering | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
4 target triple = "x86_64-unknown-unknown"
6 define <4 x double> @shuffle_v4f64_0001(<4 x double> %a, <4 x double> %b) {
7 ; ALL-LABEL: @shuffle_v4f64_0001
9 ; ALL-NEXT: vunpcklpd {{.*}} # xmm1 = xmm0[0,0]
10 ; ALL-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
12 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
13 ret <4 x double> %shuffle
16 define <4 x double> @shuffle_v4f64_0020(<4 x double> %a, <4 x double> %b) {
17 ; ALL-LABEL: @shuffle_v4f64_0020
19 ; ALL-NEXT: vextractf128 $1, %ymm0, %xmm1
20 ; ALL-NEXT: vunpcklpd {{.*}} # xmm1 = xmm1[0],xmm0[0]
21 ; ALL-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
22 ; ALL-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
24 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
25 ret <4 x double> %shuffle
28 define <4 x double> @shuffle_v4f64_0300(<4 x double> %a, <4 x double> %b) {
29 ; ALL-LABEL: @shuffle_v4f64_0300
31 ; ALL-NEXT: vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
32 ; ALL-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[0,1,2,2]
33 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
35 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
36 ret <4 x double> %shuffle
39 define <4 x double> @shuffle_v4f64_1000(<4 x double> %a, <4 x double> %b) {
40 ; ALL-LABEL: @shuffle_v4f64_1000
42 ; ALL-NEXT: vpermilpd {{.*}} # xmm1 = xmm0[1,0]
43 ; ALL-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
44 ; ALL-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
46 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
47 ret <4 x double> %shuffle
50 define <4 x double> @shuffle_v4f64_2200(<4 x double> %a, <4 x double> %b) {
51 ; ALL-LABEL: @shuffle_v4f64_2200
53 ; ALL-NEXT: vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
54 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,2]
56 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
57 ret <4 x double> %shuffle
60 define <4 x double> @shuffle_v4f64_3330(<4 x double> %a, <4 x double> %b) {
61 ; ALL-LABEL: @shuffle_v4f64_3330
63 ; ALL-NEXT: vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
64 ; ALL-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[1,1,2,2]
65 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,3,2]
66 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2],ymm1[3]
68 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
69 ret <4 x double> %shuffle
72 define <4 x double> @shuffle_v4f64_3210(<4 x double> %a, <4 x double> %b) {
73 ; ALL-LABEL: @shuffle_v4f64_3210
75 ; ALL-NEXT: vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
76 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,3,2]
78 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
79 ret <4 x double> %shuffle
82 define <4 x double> @shuffle_v4f64_0023(<4 x double> %a, <4 x double> %b) {
83 ; ALL-LABEL: @shuffle_v4f64_0023
85 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,3]
87 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 3>
88 ret <4 x double> %shuffle
91 define <4 x double> @shuffle_v4f64_0022(<4 x double> %a, <4 x double> %b) {
92 ; ALL-LABEL: @shuffle_v4f64_0022
94 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,2]
96 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
97 ret <4 x double> %shuffle
100 define <4 x double> @shuffle_v4f64_1032(<4 x double> %a, <4 x double> %b) {
101 ; ALL-LABEL: @shuffle_v4f64_1032
103 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,3,2]
105 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
106 ret <4 x double> %shuffle
109 define <4 x double> @shuffle_v4f64_1133(<4 x double> %a, <4 x double> %b) {
110 ; ALL-LABEL: @shuffle_v4f64_1133
112 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,1,3,3]
114 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
115 ret <4 x double> %shuffle
118 define <4 x double> @shuffle_v4f64_1023(<4 x double> %a, <4 x double> %b) {
119 ; ALL-LABEL: @shuffle_v4f64_1023
121 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,2,3]
123 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 3>
124 ret <4 x double> %shuffle
127 define <4 x double> @shuffle_v4f64_1022(<4 x double> %a, <4 x double> %b) {
128 ; ALL-LABEL: @shuffle_v4f64_1022
130 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,2,2]
132 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 2>
133 ret <4 x double> %shuffle
136 define <4 x double> @shuffle_v4f64_0423(<4 x double> %a, <4 x double> %b) {
137 ; ALL-LABEL: @shuffle_v4f64_0423
139 ; ALL-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[0,0,2,2]
140 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
142 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 3>
143 ret <4 x double> %shuffle
146 define <4 x double> @shuffle_v4f64_0462(<4 x double> %a, <4 x double> %b) {
147 ; ALL-LABEL: @shuffle_v4f64_0462
149 ; ALL-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[0,0,2,2]
150 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,2]
151 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
153 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 6, i32 2>
154 ret <4 x double> %shuffle
157 define <4 x double> @shuffle_v4f64_0426(<4 x double> %a, <4 x double> %b) {
158 ; ALL-LABEL: @shuffle_v4f64_0426
160 ; ALL-NEXT: vunpcklpd {{.*}} # ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
162 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
163 ret <4 x double> %shuffle
166 define <4 x double> @shuffle_v4f64_1537(<4 x double> %a, <4 x double> %b) {
167 ; ALL-LABEL: @shuffle_v4f64_1537
169 ; ALL-NEXT: vunpckhpd {{.*}} # ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
171 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
172 ret <4 x double> %shuffle
175 define <4 x double> @shuffle_v4f64_4062(<4 x double> %a, <4 x double> %b) {
176 ; ALL-LABEL: @shuffle_v4f64_4062
178 ; ALL-NEXT: vunpcklpd {{.*}} # ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
180 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
181 ret <4 x double> %shuffle
184 define <4 x double> @shuffle_v4f64_5173(<4 x double> %a, <4 x double> %b) {
185 ; ALL-LABEL: @shuffle_v4f64_5173
187 ; ALL-NEXT: vunpckhpd {{.*}} # ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
189 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 7, i32 3>
190 ret <4 x double> %shuffle
193 define <4 x double> @shuffle_v4f64_5163(<4 x double> %a, <4 x double> %b) {
194 ; ALL-LABEL: @shuffle_v4f64_5163
196 ; ALL-NEXT: vshufpd {{.*}} # ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
198 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
199 ret <4 x double> %shuffle
202 define <4 x double> @shuffle_v4f64_0527(<4 x double> %a, <4 x double> %b) {
203 ; ALL-LABEL: @shuffle_v4f64_0527
205 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
207 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
208 ret <4 x double> %shuffle
211 define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
212 ; ALL-LABEL: @shuffle_v4f64_4163
214 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
216 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
217 ret <4 x double> %shuffle
220 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
221 ; ALL-LABEL: @shuffle_v4f64_0145
223 ; ALL-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
225 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
226 ret <4 x double> %shuffle
229 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
230 ; ALL-LABEL: @shuffle_v4f64_4501
232 ; ALL-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
234 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
235 ret <4 x double> %shuffle
238 define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
239 ; ALL-LABEL: @shuffle_v4f64_0167
241 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3]
243 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
244 ret <4 x double> %shuffle
247 define <4 x i64> @shuffle_v4i64_0001(<4 x i64> %a, <4 x i64> %b) {
248 ; AVX1-LABEL: @shuffle_v4i64_0001
250 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm1 = xmm0[0,0]
251 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
254 ; AVX2-LABEL: @shuffle_v4i64_0001
256 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,0,0,1]
258 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
259 ret <4 x i64> %shuffle
262 define <4 x i64> @shuffle_v4i64_0020(<4 x i64> %a, <4 x i64> %b) {
263 ; AVX1-LABEL: @shuffle_v4i64_0020
265 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
266 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm1 = xmm1[0],xmm0[0]
267 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
268 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
271 ; AVX2-LABEL: @shuffle_v4i64_0020
273 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,0,2,0]
275 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
276 ret <4 x i64> %shuffle
279 define <4 x i64> @shuffle_v4i64_0112(<4 x i64> %a, <4 x i64> %b) {
280 ; AVX1-LABEL: @shuffle_v4i64_0112
282 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
283 ; AVX1-NEXT: vshufpd {{.*}} # xmm1 = xmm0[1],xmm1[0]
284 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
287 ; AVX2-LABEL: @shuffle_v4i64_0112
289 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,1,2]
291 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 1, i32 2>
292 ret <4 x i64> %shuffle
295 define <4 x i64> @shuffle_v4i64_0300(<4 x i64> %a, <4 x i64> %b) {
296 ; AVX1-LABEL: @shuffle_v4i64_0300
298 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
299 ; AVX1-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[0,1,2,2]
300 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
303 ; AVX2-LABEL: @shuffle_v4i64_0300
305 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,3,0,0]
307 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
308 ret <4 x i64> %shuffle
311 define <4 x i64> @shuffle_v4i64_1000(<4 x i64> %a, <4 x i64> %b) {
312 ; AVX1-LABEL: @shuffle_v4i64_1000
314 ; AVX1-NEXT: vpermilpd {{.*}} # xmm1 = xmm0[1,0]
315 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
316 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
319 ; AVX2-LABEL: @shuffle_v4i64_1000
321 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[1,0,0,0]
323 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
324 ret <4 x i64> %shuffle
327 define <4 x i64> @shuffle_v4i64_2200(<4 x i64> %a, <4 x i64> %b) {
328 ; AVX1-LABEL: @shuffle_v4i64_2200
330 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
331 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,2]
334 ; AVX2-LABEL: @shuffle_v4i64_2200
336 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[2,2,0,0]
338 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
339 ret <4 x i64> %shuffle
342 define <4 x i64> @shuffle_v4i64_3330(<4 x i64> %a, <4 x i64> %b) {
343 ; AVX1-LABEL: @shuffle_v4i64_3330
345 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
346 ; AVX1-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[1,1,2,2]
347 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,3,2]
348 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2],ymm1[3]
351 ; AVX2-LABEL: @shuffle_v4i64_3330
353 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[3,3,3,0]
355 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
356 ret <4 x i64> %shuffle
359 define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
360 ; AVX1-LABEL: @shuffle_v4i64_3210
362 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
363 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,3,2]
366 ; AVX2-LABEL: @shuffle_v4i64_3210
368 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[3,2,1,0]
370 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
371 ret <4 x i64> %shuffle
374 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
375 ; AVX1-LABEL: @shuffle_v4i64_0124
377 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
378 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm1 = xmm1[0,0]
379 ; AVX1-NEXT: vblendpd {{.*}} # xmm1 = xmm2[0],xmm1[1]
380 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
383 ; AVX2-LABEL: @shuffle_v4i64_0124
385 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,1,2,0]
386 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
388 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
389 ret <4 x i64> %shuffle
392 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
393 ; AVX1-LABEL: @shuffle_v4i64_0142
395 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
396 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm2 = xmm2[0,0]
397 ; AVX1-NEXT: vblendpd {{.*}} # xmm1 = xmm1[0],xmm2[1]
398 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
401 ; AVX2-LABEL: @shuffle_v4i64_0142
403 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,1,0,3]
404 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,2,2]
405 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
407 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
408 ret <4 x i64> %shuffle
411 define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
412 ; AVX1-LABEL: @shuffle_v4i64_0412
414 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
415 ; AVX1-NEXT: vshufpd {{.*}} # xmm2 = xmm0[1],xmm2[0]
416 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm1 = xmm1[0,0]
417 ; AVX1-NEXT: vblendpd {{.*}} # xmm0 = xmm0[0],xmm1[1]
418 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
421 ; AVX2-LABEL: @shuffle_v4i64_0412
423 ; AVX2-NEXT: vpshufd {{.*}} # ymm1 = ymm1[0,1,0,1,4,5,4,5]
424 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,1,2]
425 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
427 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
428 ret <4 x i64> %shuffle
431 define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
432 ; AVX1-LABEL: @shuffle_v4i64_4012
434 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
435 ; AVX1-NEXT: vshufpd {{.*}} # xmm2 = xmm0[1],xmm2[0]
436 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
437 ; AVX1-NEXT: vblendpd {{.*}} # xmm0 = xmm1[0],xmm0[1]
438 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
441 ; AVX2-LABEL: @shuffle_v4i64_4012
443 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,0,1,2]
444 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
446 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
447 ret <4 x i64> %shuffle
450 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
451 ; AVX1-LABEL: @shuffle_v4i64_0145
453 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
456 ; AVX2-LABEL: @shuffle_v4i64_0145
458 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,1,0,1]
459 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
461 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
462 ret <4 x i64> %shuffle
465 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
466 ; AVX1-LABEL: @shuffle_v4i64_0451
468 ; AVX1-NEXT: vpermilpd {{.*}} # xmm2 = xmm1[1,0]
469 ; AVX1-NEXT: vblendpd {{.*}} # xmm2 = xmm2[0],xmm0[1]
470 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm1 = xmm1[0,0]
471 ; AVX1-NEXT: vblendpd {{.*}} # xmm0 = xmm0[0],xmm1[1]
472 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
475 ; AVX2-LABEL: @shuffle_v4i64_0451
477 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,0,1,3]
478 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,2,1]
479 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
481 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
482 ret <4 x i64> %shuffle
485 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
486 ; AVX1-LABEL: @shuffle_v4i64_4501
488 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
491 ; AVX2-LABEL: @shuffle_v4i64_4501
493 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,0,1]
494 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
496 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
497 ret <4 x i64> %shuffle
500 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
501 ; AVX1-LABEL: @shuffle_v4i64_4015
503 ; AVX1-NEXT: vpermilpd {{.*}} # xmm2 = xmm0[1,0]
504 ; AVX1-NEXT: vblendpd {{.*}} # xmm2 = xmm2[0],xmm1[1]
505 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
506 ; AVX1-NEXT: vblendpd {{.*}} # xmm0 = xmm1[0],xmm0[1]
507 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
510 ; AVX2-LABEL: @shuffle_v4i64_4015
512 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,1,2,1]
513 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,0,1,3]
514 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
516 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
517 ret <4 x i64> %shuffle
520 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
521 ; AVX1-LABEL: @stress_test1
523 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm0 = ymm1[2,3,0,1]
524 ; AVX1-NEXT: vpermilpd {{.*}} # ymm2 = ymm0[1,0,3,2]
525 ; AVX1-NEXT: vblendpd {{.*}} # ymm1 = ymm2[0],ymm1[1],ymm2[2,3]
526 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm1
527 ; AVX1-NEXT: vpermilpd {{.*}} # xmm1 = xmm1[1,0]
528 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,1,3,2]
529 ; AVX1-NEXT: vpermilpd {{.*}} # xmm0 = xmm0[1,0]
530 ; AVX1-NEXT: vblendpd {{.*}} # xmm0 = xmm0[0],xmm1[1]
531 ; AVX1-NEXT: vpermilpd {{.*}} # xmm1 = xmm1[1,0]
532 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
535 ; AVX2-LABEL: @stress_test1
537 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm1[3,1,1,0]
538 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[3,1,2,3]
539 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[3,3,1,3]
540 ; AVX2-NEXT: vpshufd {{.*}} # ymm1 = ymm1[2,3,2,3,6,7,6,7]
541 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,1,0]
542 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
544 %c = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 1, i32 1, i32 0>
545 %d = shufflevector <4 x i64> %c, <4 x i64> undef, <4 x i32> <i32 3, i32 undef, i32 2, i32 undef>
546 %e = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 undef>
547 %f = shufflevector <4 x i64> %d, <4 x i64> %e, <4 x i32> <i32 5, i32 1, i32 1, i32 0>
552 define <4 x i64> @insert_reg_and_zero_v4i64(i64 %a) {
553 ; AVX1-LABEL: @insert_reg_and_zero_v4i64
555 ; AVX1-NEXT: vmovq %rdi, %xmm0
556 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
557 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
560 ; AVX2-LABEL: @insert_reg_and_zero_v4i64
562 ; AVX2-NEXT: vmovq %rdi, %xmm0
563 ; AVX2-NEXT: vpxor %ymm1, %ymm1, %ymm1
564 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
566 %v = insertelement <4 x i64> undef, i64 %a, i64 0
567 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
568 ret <4 x i64> %shuffle
571 define <4 x i64> @insert_mem_and_zero_v4i64(i64* %ptr) {
572 ; AVX1-LABEL: @insert_mem_and_zero_v4i64
574 ; AVX1-NEXT: vmovq (%rdi), %xmm0
575 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
576 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
579 ; AVX2-LABEL: @insert_mem_and_zero_v4i64
581 ; AVX2-NEXT: vmovq (%rdi), %xmm0
582 ; AVX2-NEXT: vpxor %ymm1, %ymm1, %ymm1
583 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
586 %v = insertelement <4 x i64> undef, i64 %a, i64 0
587 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
588 ret <4 x i64> %shuffle
591 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
592 ; ALL-LABEL: @insert_reg_and_zero_v4f64
594 ; ALL: vxorpd %ymm1, %ymm1, %ymm1
595 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
597 %v = insertelement <4 x double> undef, double %a, i32 0
598 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
599 ret <4 x double> %shuffle
602 define <4 x double> @insert_mem_and_zero_v4f64(double* %ptr) {
603 ; ALL-LABEL: @insert_mem_and_zero_v4f64
605 ; ALL-NEXT: vmovsd (%rdi), %xmm0
606 ; ALL-NEXT: vxorpd %ymm1, %ymm1, %ymm1
607 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
609 %a = load double* %ptr
610 %v = insertelement <4 x double> undef, double %a, i32 0
611 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
612 ret <4 x double> %shuffle