1 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx | FileCheck %s --check-prefix=ALL --check-prefix=AVX1
2 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX2
3 ; RUN: llc < %s -mcpu=knl -mattr=+avx512vl | FileCheck %s --check-prefix=ALL --check-prefix=AVX512VL
5 target triple = "x86_64-unknown-unknown"
7 define <4 x double> @shuffle_v4f64_0000(<4 x double> %a, <4 x double> %b) {
8 ; AVX1-LABEL: shuffle_v4f64_0000:
10 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
11 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
14 ; AVX2-LABEL: shuffle_v4f64_0000:
16 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
19 ; AVX512VL-LABEL: shuffle_v4f64_0000:
21 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
23 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
24 ret <4 x double> %shuffle
27 define <4 x double> @shuffle_v4f64_0001(<4 x double> %a, <4 x double> %b) {
28 ; AVX1-LABEL: shuffle_v4f64_0001:
30 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm0[0,0]
31 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
34 ; AVX2-LABEL: shuffle_v4f64_0001:
36 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,0,0,1]
39 ; AVX512VL-LABEL: shuffle_v4f64_0001:
41 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,0,0,1]
43 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
44 ret <4 x double> %shuffle
47 define <4 x double> @shuffle_v4f64_0020(<4 x double> %a, <4 x double> %b) {
48 ; AVX1-LABEL: shuffle_v4f64_0020:
50 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
51 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
52 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
53 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
56 ; AVX2-LABEL: shuffle_v4f64_0020:
58 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,0,2,0]
61 ; AVX512VL-LABEL: shuffle_v4f64_0020:
63 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,0,2,0]
65 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
66 ret <4 x double> %shuffle
69 define <4 x double> @shuffle_v4f64_0300(<4 x double> %a, <4 x double> %b) {
70 ; AVX1-LABEL: shuffle_v4f64_0300:
72 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
73 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
74 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
77 ; AVX2-LABEL: shuffle_v4f64_0300:
79 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,0,0]
82 ; AVX512VL-LABEL: shuffle_v4f64_0300:
84 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,0,0]
86 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
87 ret <4 x double> %shuffle
90 define <4 x double> @shuffle_v4f64_1000(<4 x double> %a, <4 x double> %b) {
91 ; AVX1-LABEL: shuffle_v4f64_1000:
93 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
94 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
95 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
98 ; AVX2-LABEL: shuffle_v4f64_1000:
100 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,0,0]
103 ; AVX512VL-LABEL: shuffle_v4f64_1000:
105 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,0,0]
106 ; AVX512VL-NEXT: retq
107 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
108 ret <4 x double> %shuffle
111 define <4 x double> @shuffle_v4f64_2200(<4 x double> %a, <4 x double> %b) {
112 ; AVX1-LABEL: shuffle_v4f64_2200:
114 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
115 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
118 ; AVX2-LABEL: shuffle_v4f64_2200:
120 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,0,0]
123 ; AVX512VL-LABEL: shuffle_v4f64_2200:
125 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,0,0]
126 ; AVX512VL-NEXT: retq
127 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
128 ret <4 x double> %shuffle
131 define <4 x double> @shuffle_v4f64_3330(<4 x double> %a, <4 x double> %b) {
132 ; AVX1-LABEL: shuffle_v4f64_3330:
134 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
135 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3]
136 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
139 ; AVX2-LABEL: shuffle_v4f64_3330:
141 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,0]
144 ; AVX512VL-LABEL: shuffle_v4f64_3330:
146 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,0]
147 ; AVX512VL-NEXT: retq
148 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
149 ret <4 x double> %shuffle
152 define <4 x double> @shuffle_v4f64_3210(<4 x double> %a, <4 x double> %b) {
153 ; AVX1-LABEL: shuffle_v4f64_3210:
155 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
156 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
159 ; AVX2-LABEL: shuffle_v4f64_3210:
161 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,1,0]
164 ; AVX512VL-LABEL: shuffle_v4f64_3210:
166 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,1,0]
167 ; AVX512VL-NEXT: retq
168 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
169 ret <4 x double> %shuffle
172 define <4 x double> @shuffle_v4f64_0023(<4 x double> %a, <4 x double> %b) {
173 ; AVX1-LABEL: shuffle_v4f64_0023:
175 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,3]
178 ; AVX2-LABEL: shuffle_v4f64_0023:
180 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,3]
183 ; AVX512VL-LABEL: shuffle_v4f64_0023:
185 ; AVX512VL-NEXT: vpermilpd $8, %ymm0, %ymm0
186 ; AVX512VL-NEXT: retq
188 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 3>
189 ret <4 x double> %shuffle
192 define <4 x double> @shuffle_v4f64_0022(<4 x double> %a, <4 x double> %b) {
193 ; ALL-LABEL: shuffle_v4f64_0022:
195 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
197 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
198 ret <4 x double> %shuffle
201 define <4 x double> @shuffle_v4f64_1032(<4 x double> %a, <4 x double> %b) {
202 ; AVX1-LABEL: shuffle_v4f64_1032:
204 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
207 ; AVX2-LABEL: shuffle_v4f64_1032:
209 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
212 ; AVX512VL-LABEL: shuffle_v4f64_1032:
214 ; AVX512VL-NEXT: vpermilpd $5, %ymm0, %ymm0
215 ; AVX512VL-NEXT: retq
216 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
217 ret <4 x double> %shuffle
220 define <4 x double> @shuffle_v4f64_1133(<4 x double> %a, <4 x double> %b) {
221 ; AVX1-LABEL: shuffle_v4f64_1133:
223 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
226 ; AVX2-LABEL: shuffle_v4f64_1133:
228 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
231 ; AVX512VL-LABEL: shuffle_v4f64_1133:
233 ; AVX512VL-NEXT: vpermilpd $15, %ymm0, %ymm0
234 ; AVX512VL-NEXT: retq
235 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
236 ret <4 x double> %shuffle
239 define <4 x double> @shuffle_v4f64_1023(<4 x double> %a, <4 x double> %b) {
240 ; AVX1-LABEL: shuffle_v4f64_1023:
242 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,3]
245 ; AVX2-LABEL: shuffle_v4f64_1023:
247 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,3]
250 ; AVX512VL-LABEL: shuffle_v4f64_1023:
252 ; AVX512VL-NEXT: vpermilpd $9, %ymm0, %ymm0
253 ; AVX512VL-NEXT: retq
254 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 3>
255 ret <4 x double> %shuffle
258 define <4 x double> @shuffle_v4f64_1022(<4 x double> %a, <4 x double> %b) {
259 ; AVX1-LABEL: shuffle_v4f64_1022:
261 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
264 ; AVX2-LABEL: shuffle_v4f64_1022:
266 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
269 ; AVX512VL-LABEL: shuffle_v4f64_1022:
271 ; AVX512VL-NEXT: vpermilpd $1, %ymm0, %ymm0
272 ; AVX512VL-NEXT: retq
273 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 2>
274 ret <4 x double> %shuffle
277 define <4 x double> @shuffle_v4f64_0423(<4 x double> %a, <4 x double> %b) {
278 ; AVX1-LABEL: shuffle_v4f64_0423:
280 ; AVX1-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
281 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
284 ; AVX2-LABEL: shuffle_v4f64_0423:
286 ; AVX2-NEXT: vbroadcastsd %xmm1, %ymm1
287 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
290 ; AVX512VL-LABEL: shuffle_v4f64_0423:
292 ; AVX512VL-NEXT: vbroadcastsd %xmm1, %ymm1
293 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
294 ; AVX512VL-NEXT: retq
295 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 3>
296 ret <4 x double> %shuffle
299 define <4 x double> @shuffle_v4f64_0462(<4 x double> %a, <4 x double> %b) {
300 ; ALL-LABEL: shuffle_v4f64_0462:
302 ; ALL-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
303 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
304 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
306 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 6, i32 2>
307 ret <4 x double> %shuffle
310 define <4 x double> @shuffle_v4f64_0426(<4 x double> %a, <4 x double> %b) {
311 ; ALL-LABEL: shuffle_v4f64_0426:
313 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
315 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
316 ret <4 x double> %shuffle
319 define <4 x double> @shuffle_v4f64_1537(<4 x double> %a, <4 x double> %b) {
320 ; ALL-LABEL: shuffle_v4f64_1537:
322 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
324 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
325 ret <4 x double> %shuffle
328 define <4 x double> @shuffle_v4f64_4062(<4 x double> %a, <4 x double> %b) {
329 ; ALL-LABEL: shuffle_v4f64_4062:
331 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
333 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
334 ret <4 x double> %shuffle
337 define <4 x double> @shuffle_v4f64_5173(<4 x double> %a, <4 x double> %b) {
338 ; ALL-LABEL: shuffle_v4f64_5173:
340 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
342 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 7, i32 3>
343 ret <4 x double> %shuffle
346 define <4 x double> @shuffle_v4f64_5163(<4 x double> %a, <4 x double> %b) {
347 ; AVX1-LABEL: shuffle_v4f64_5163:
349 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
352 ; AVX2-LABEL: shuffle_v4f64_5163:
354 ; AVX2-NEXT: vshufpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
357 ; AVX512VL-LABEL: shuffle_v4f64_5163:
359 ; AVX512VL-NEXT: vshufpd $11, %ymm0, %ymm1, %ymm0
360 ; AVX512VL-NEXT: retq
361 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
362 ret <4 x double> %shuffle
365 define <4 x double> @shuffle_v4f64_0527(<4 x double> %a, <4 x double> %b) {
366 ; ALL-LABEL: shuffle_v4f64_0527:
368 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
370 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
371 ret <4 x double> %shuffle
374 define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
375 ; ALL-LABEL: shuffle_v4f64_4163:
377 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
379 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
380 ret <4 x double> %shuffle
383 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
384 ; AVX1-LABEL: shuffle_v4f64_0145:
386 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
389 ; AVX2-LABEL: shuffle_v4f64_0145:
391 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
394 ; AVX512VL-LABEL: shuffle_v4f64_0145:
396 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
397 ; AVX512VL-NEXT: retq
398 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
399 ret <4 x double> %shuffle
402 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
403 ; AVX1-LABEL: shuffle_v4f64_4501:
405 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
408 ; AVX2-LABEL: shuffle_v4f64_4501:
410 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
413 ; AVX512VL-LABEL: shuffle_v4f64_4501:
415 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm1, %ymm0
416 ; AVX512VL-NEXT: retq
417 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
418 ret <4 x double> %shuffle
421 define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
422 ; ALL-LABEL: shuffle_v4f64_0167:
424 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
426 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
427 ret <4 x double> %shuffle
430 define <4 x double> @shuffle_v4f64_1054(<4 x double> %a, <4 x double> %b) {
431 ; AVX1-LABEL: shuffle_v4f64_1054:
433 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
434 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
437 ; AVX2-LABEL: shuffle_v4f64_1054:
439 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
440 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
443 ; AVX512VL-LABEL: shuffle_v4f64_1054:
445 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
446 ; AVX512VL-NEXT: vpermilpd $5, %ymm0, %ymm0
447 ; AVX512VL-NEXT: retq
448 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
449 ret <4 x double> %shuffle
452 define <4 x double> @shuffle_v4f64_3254(<4 x double> %a, <4 x double> %b) {
453 ; AVX1-LABEL: shuffle_v4f64_3254:
455 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
456 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
459 ; AVX2-LABEL: shuffle_v4f64_3254:
461 ; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
462 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
465 ; AVX512VL-LABEL: shuffle_v4f64_3254:
467 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
468 ; AVX512VL-NEXT: vpermilpd $5, %ymm0, %ymm0
469 ; AVX512VL-NEXT: retq
470 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
471 ret <4 x double> %shuffle
474 define <4 x double> @shuffle_v4f64_3276(<4 x double> %a, <4 x double> %b) {
475 ; AVX1-LABEL: shuffle_v4f64_3276:
477 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
478 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
481 ; AVX2-LABEL: shuffle_v4f64_3276:
483 ; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
484 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
487 ; AVX512VL-LABEL: shuffle_v4f64_3276:
489 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
490 ; AVX512VL-NEXT: vpermilpd $5, %ymm0, %ymm0
491 ; AVX512VL-NEXT: retq
492 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
493 ret <4 x double> %shuffle
496 define <4 x double> @shuffle_v4f64_1076(<4 x double> %a, <4 x double> %b) {
497 ; AVX1-LABEL: shuffle_v4f64_1076:
499 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
500 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
503 ; AVX2-LABEL: shuffle_v4f64_1076:
505 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
506 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
509 ; AVX512VL-LABEL: shuffle_v4f64_1076:
511 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
512 ; AVX512VL-NEXT: vpermilpd $5, %ymm0, %ymm0
513 ; AVX512VL-NEXT: retq
514 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
515 ret <4 x double> %shuffle
518 define <4 x double> @shuffle_v4f64_0415(<4 x double> %a, <4 x double> %b) {
519 ; AVX1-LABEL: shuffle_v4f64_0415:
521 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
522 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
523 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
526 ; AVX2-LABEL: shuffle_v4f64_0415:
528 ; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
529 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
530 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
533 ; AVX512VL-LABEL: shuffle_v4f64_0415:
535 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
536 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
537 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
538 ; AVX512VL-NEXT: retq
539 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
540 ret <4 x double> %shuffle
543 define <4 x double> @shuffle_v4f64_u062(<4 x double> %a, <4 x double> %b) {
544 ; ALL-LABEL: shuffle_v4f64_u062:
546 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
548 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 undef, i32 0, i32 6, i32 2>
549 ret <4 x double> %shuffle
552 define <4 x i64> @shuffle_v4i64_0000(<4 x i64> %a, <4 x i64> %b) {
553 ; AVX1-LABEL: shuffle_v4i64_0000:
555 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
556 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
559 ; AVX2-LABEL: shuffle_v4i64_0000:
561 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
564 ; AVX512VL-LABEL: shuffle_v4i64_0000:
566 ; AVX512VL-NEXT: vpbroadcastq %xmm0, %ymm0
567 ; AVX512VL-NEXT: retq
568 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
569 ret <4 x i64> %shuffle
572 define <4 x i64> @shuffle_v4i64_0001(<4 x i64> %a, <4 x i64> %b) {
573 ; AVX1-LABEL: shuffle_v4i64_0001:
575 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm0[0,0]
576 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
579 ; AVX2-LABEL: shuffle_v4i64_0001:
581 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
584 ; AVX512VL-LABEL: shuffle_v4i64_0001:
586 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
587 ; AVX512VL-NEXT: retq
588 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
589 ret <4 x i64> %shuffle
592 define <4 x i64> @shuffle_v4i64_0020(<4 x i64> %a, <4 x i64> %b) {
593 ; AVX1-LABEL: shuffle_v4i64_0020:
595 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
596 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
597 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
598 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
601 ; AVX2-LABEL: shuffle_v4i64_0020:
603 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
606 ; AVX512VL-LABEL: shuffle_v4i64_0020:
608 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
609 ; AVX512VL-NEXT: retq
610 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
611 ret <4 x i64> %shuffle
614 define <4 x i64> @shuffle_v4i64_0112(<4 x i64> %a, <4 x i64> %b) {
615 ; AVX1-LABEL: shuffle_v4i64_0112:
617 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
618 ; AVX1-NEXT: vshufpd {{.*#+}} xmm1 = xmm0[1],xmm1[0]
619 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
622 ; AVX2-LABEL: shuffle_v4i64_0112:
624 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
627 ; AVX512VL-LABEL: shuffle_v4i64_0112:
629 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
630 ; AVX512VL-NEXT: retq
631 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 1, i32 2>
632 ret <4 x i64> %shuffle
635 define <4 x i64> @shuffle_v4i64_0300(<4 x i64> %a, <4 x i64> %b) {
636 ; AVX1-LABEL: shuffle_v4i64_0300:
638 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
639 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
640 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
643 ; AVX2-LABEL: shuffle_v4i64_0300:
645 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
648 ; AVX512VL-LABEL: shuffle_v4i64_0300:
650 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
651 ; AVX512VL-NEXT: retq
652 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
653 ret <4 x i64> %shuffle
656 define <4 x i64> @shuffle_v4i64_1000(<4 x i64> %a, <4 x i64> %b) {
657 ; AVX1-LABEL: shuffle_v4i64_1000:
659 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
660 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
661 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
664 ; AVX2-LABEL: shuffle_v4i64_1000:
666 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
669 ; AVX512VL-LABEL: shuffle_v4i64_1000:
671 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
672 ; AVX512VL-NEXT: retq
673 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
674 ret <4 x i64> %shuffle
677 define <4 x i64> @shuffle_v4i64_2200(<4 x i64> %a, <4 x i64> %b) {
678 ; AVX1-LABEL: shuffle_v4i64_2200:
680 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
681 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
684 ; AVX2-LABEL: shuffle_v4i64_2200:
686 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
689 ; AVX512VL-LABEL: shuffle_v4i64_2200:
691 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
692 ; AVX512VL-NEXT: retq
693 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
694 ret <4 x i64> %shuffle
697 define <4 x i64> @shuffle_v4i64_3330(<4 x i64> %a, <4 x i64> %b) {
698 ; AVX1-LABEL: shuffle_v4i64_3330:
700 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
701 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3]
702 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
705 ; AVX2-LABEL: shuffle_v4i64_3330:
707 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
710 ; AVX512VL-LABEL: shuffle_v4i64_3330:
712 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
713 ; AVX512VL-NEXT: retq
714 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
715 ret <4 x i64> %shuffle
718 define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
719 ; AVX1-LABEL: shuffle_v4i64_3210:
721 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
722 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
725 ; AVX2-LABEL: shuffle_v4i64_3210:
727 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
730 ; AVX512VL-LABEL: shuffle_v4i64_3210:
732 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
733 ; AVX512VL-NEXT: retq
734 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
735 ret <4 x i64> %shuffle
738 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
739 ; AVX1-LABEL: shuffle_v4i64_0124:
741 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
742 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
743 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3]
746 ; AVX2-LABEL: shuffle_v4i64_0124:
748 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
749 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
752 ; AVX512VL-LABEL: shuffle_v4i64_0124:
754 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
755 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
756 ; AVX512VL-NEXT: retq
757 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
758 ret <4 x i64> %shuffle
761 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
762 ; AVX1-LABEL: shuffle_v4i64_0142:
764 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm1, %ymm1
765 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,1,2,2]
766 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
769 ; AVX2-LABEL: shuffle_v4i64_0142:
771 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
772 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
773 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
776 ; AVX512VL-LABEL: shuffle_v4i64_0142:
778 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm1, %ymm1
779 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
780 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
781 ; AVX512VL-NEXT: retq
782 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
783 ret <4 x i64> %shuffle
786 define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
787 ; AVX1-LABEL: shuffle_v4i64_0412:
789 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
790 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
791 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
792 ; AVX1-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
793 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
796 ; AVX2-LABEL: shuffle_v4i64_0412:
798 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
799 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
800 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
803 ; AVX512VL-LABEL: shuffle_v4i64_0412:
805 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
806 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
807 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
808 ; AVX512VL-NEXT: retq
809 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
810 ret <4 x i64> %shuffle
813 define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
814 ; AVX1-LABEL: shuffle_v4i64_4012:
816 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
817 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
818 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
819 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
820 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
823 ; AVX2-LABEL: shuffle_v4i64_4012:
825 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
826 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
829 ; AVX512VL-LABEL: shuffle_v4i64_4012:
831 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
832 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
833 ; AVX512VL-NEXT: retq
834 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
835 ret <4 x i64> %shuffle
838 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
839 ; AVX1-LABEL: shuffle_v4i64_0145:
841 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
844 ; AVX2-LABEL: shuffle_v4i64_0145:
846 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
849 ; AVX512VL-LABEL: shuffle_v4i64_0145:
851 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
852 ; AVX512VL-NEXT: retq
853 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
854 ret <4 x i64> %shuffle
857 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
858 ; AVX1-LABEL: shuffle_v4i64_0451:
860 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm1[1],xmm0[1]
861 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
862 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
865 ; AVX2-LABEL: shuffle_v4i64_0451:
867 ; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
868 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
869 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
872 ; AVX512VL-LABEL: shuffle_v4i64_0451:
874 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm0, %ymm0
875 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
876 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
877 ; AVX512VL-NEXT: retq
878 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
879 ret <4 x i64> %shuffle
882 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
883 ; AVX1-LABEL: shuffle_v4i64_4501:
885 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
888 ; AVX2-LABEL: shuffle_v4i64_4501:
890 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
893 ; AVX512VL-LABEL: shuffle_v4i64_4501:
895 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm1, %ymm0
896 ; AVX512VL-NEXT: retq
897 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
898 ret <4 x i64> %shuffle
901 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
902 ; AVX1-LABEL: shuffle_v4i64_4015:
904 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
905 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
906 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
909 ; AVX2-LABEL: shuffle_v4i64_4015:
911 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
912 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
913 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
916 ; AVX512VL-LABEL: shuffle_v4i64_4015:
918 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm1, %ymm1
919 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
920 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
921 ; AVX512VL-NEXT: retq
922 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
923 ret <4 x i64> %shuffle
926 define <4 x i64> @shuffle_v4i64_2u35(<4 x i64> %a, <4 x i64> %b) {
927 ; AVX1-LABEL: shuffle_v4i64_2u35:
929 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
930 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm1 = xmm0[1],xmm1[1]
931 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
934 ; AVX2-LABEL: shuffle_v4i64_2u35:
936 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
937 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
940 ; AVX512VL-LABEL: shuffle_v4i64_2u35:
942 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
943 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
944 ; AVX512VL-NEXT: retq
945 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 undef, i32 3, i32 5>
946 ret <4 x i64> %shuffle
949 define <4 x i64> @shuffle_v4i64_1251(<4 x i64> %a, <4 x i64> %b) {
950 ; AVX1-LABEL: shuffle_v4i64_1251:
952 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
953 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm2[0],ymm0[2],ymm2[3]
954 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
955 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
956 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
959 ; AVX2-LABEL: shuffle_v4i64_1251:
961 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
962 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
963 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
966 ; AVX512VL-LABEL: shuffle_v4i64_1251:
968 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
969 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
970 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
971 ; AVX512VL-NEXT: retq
972 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 5, i32 1>
973 ret <4 x i64> %shuffle
976 define <4 x i64> @shuffle_v4i64_1054(<4 x i64> %a, <4 x i64> %b) {
977 ; AVX1-LABEL: shuffle_v4i64_1054:
979 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
980 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
983 ; AVX2-LABEL: shuffle_v4i64_1054:
985 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
986 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
989 ; AVX512VL-LABEL: shuffle_v4i64_1054:
991 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
992 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
993 ; AVX512VL-NEXT: retq
994 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
995 ret <4 x i64> %shuffle
998 define <4 x i64> @shuffle_v4i64_3254(<4 x i64> %a, <4 x i64> %b) {
999 ; AVX1-LABEL: shuffle_v4i64_3254:
1001 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
1002 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1005 ; AVX2-LABEL: shuffle_v4i64_3254:
1007 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
1008 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1011 ; AVX512VL-LABEL: shuffle_v4i64_3254:
1013 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
1014 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1015 ; AVX512VL-NEXT: retq
1016 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
1017 ret <4 x i64> %shuffle
1020 define <4 x i64> @shuffle_v4i64_3276(<4 x i64> %a, <4 x i64> %b) {
1021 ; AVX1-LABEL: shuffle_v4i64_3276:
1023 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1024 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1027 ; AVX2-LABEL: shuffle_v4i64_3276:
1029 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1030 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1033 ; AVX512VL-LABEL: shuffle_v4i64_3276:
1035 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1036 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1037 ; AVX512VL-NEXT: retq
1038 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
1039 ret <4 x i64> %shuffle
1042 define <4 x i64> @shuffle_v4i64_1076(<4 x i64> %a, <4 x i64> %b) {
1043 ; AVX1-LABEL: shuffle_v4i64_1076:
1045 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1046 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1049 ; AVX2-LABEL: shuffle_v4i64_1076:
1051 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1052 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1055 ; AVX512VL-LABEL: shuffle_v4i64_1076:
1057 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1058 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1059 ; AVX512VL-NEXT: retq
1060 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
1061 ret <4 x i64> %shuffle
1064 define <4 x i64> @shuffle_v4i64_0415(<4 x i64> %a, <4 x i64> %b) {
1065 ; AVX1-LABEL: shuffle_v4i64_0415:
1067 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
1068 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
1069 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
1072 ; AVX2-LABEL: shuffle_v4i64_0415:
1074 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1075 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1076 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1079 ; AVX512VL-LABEL: shuffle_v4i64_0415:
1081 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1082 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1083 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1084 ; AVX512VL-NEXT: retq
1085 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
1086 ret <4 x i64> %shuffle
1089 define <4 x i64> @shuffle_v4i64_z4z6(<4 x i64> %a) {
1090 ; AVX1-LABEL: shuffle_v4i64_z4z6:
1092 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1093 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1096 ; AVX2-LABEL: shuffle_v4i64_z4z6:
1098 ; AVX2-NEXT: vpslldq {{.*#+}} ymm0 = zero,zero,zero,zero,zero,zero,zero,zero,ymm0[0,1,2,3,4,5,6,7],zero,zero,zero,zero,zero,zero,zero,zero,ymm0[16,17,18,19,20,21,22,23]
1101 ; AVX512VL-LABEL: shuffle_v4i64_z4z6:
1103 ; AVX512VL-NEXT: vpslldq {{.*#+}} ymm0 = zero,zero,zero,zero,zero,zero,zero,zero,ymm0[0,1,2,3,4,5,6,7],zero,zero,zero,zero,zero,zero,zero,zero,ymm0[16,17,18,19,20,21,22,23]
1104 ; AVX512VL-NEXT: retq
1105 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 0, i32 4, i32 0, i32 6>
1106 ret <4 x i64> %shuffle
1109 define <4 x i64> @shuffle_v4i64_5zuz(<4 x i64> %a) {
1110 ; AVX1-LABEL: shuffle_v4i64_5zuz:
1112 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1113 ; AVX1-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
1116 ; AVX2-LABEL: shuffle_v4i64_5zuz:
1118 ; AVX2-NEXT: vpsrldq {{.*#+}} ymm0 = ymm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,ymm0[24,25,26,27,28,29,30,31],zero,zero,zero,zero,zero,zero,zero,zero
1121 ; AVX512VL-LABEL: shuffle_v4i64_5zuz:
1123 ; AVX512VL-NEXT: vpsrldq {{.*#+}} ymm0 = ymm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,ymm0[24,25,26,27,28,29,30,31],zero,zero,zero,zero,zero,zero,zero,zero
1124 ; AVX512VL-NEXT: retq
1125 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 5, i32 0, i32 undef, i32 0>
1126 ret <4 x i64> %shuffle
1129 define <4 x i64> @shuffle_v4i64_40u2(<4 x i64> %a, <4 x i64> %b) {
1130 ; AVX1-LABEL: shuffle_v4i64_40u2:
1132 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1135 ; AVX2-LABEL: shuffle_v4i64_40u2:
1137 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1140 ; AVX512VL-LABEL: shuffle_v4i64_40u2:
1142 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1143 ; AVX512VL-NEXT: retq
1144 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 undef, i32 2>
1145 ret <4 x i64> %shuffle
1148 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
1149 ; ALL-LABEL: stress_test1:
1151 %c = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 1, i32 1, i32 0>
1152 %d = shufflevector <4 x i64> %c, <4 x i64> undef, <4 x i32> <i32 3, i32 undef, i32 2, i32 undef>
1153 %e = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 undef>
1154 %f = shufflevector <4 x i64> %d, <4 x i64> %e, <4 x i32> <i32 5, i32 1, i32 1, i32 0>
1159 define <4 x i64> @insert_reg_and_zero_v4i64(i64 %a) {
1160 ; ALL-LABEL: insert_reg_and_zero_v4i64:
1162 ; ALL-NEXT: vmovq %rdi, %xmm0
1164 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1165 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1166 ret <4 x i64> %shuffle
1169 define <4 x i64> @insert_mem_and_zero_v4i64(i64* %ptr) {
1170 ; AVX1-LABEL: insert_mem_and_zero_v4i64:
1172 ; AVX1-NEXT: vmovq {{.*#+}} xmm0 = mem[0],zero
1175 ; AVX2-LABEL: insert_mem_and_zero_v4i64:
1177 ; AVX2-NEXT: vmovq {{.*#+}} xmm0 = mem[0],zero
1180 ; AVX512VL-LABEL: insert_mem_and_zero_v4i64:
1182 ; AVX512VL-NEXT: vmovq (%rdi), %xmm0
1183 ; AVX512VL-NEXT: retq
1184 %a = load i64, i64* %ptr
1185 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1186 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1187 ret <4 x i64> %shuffle
1190 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
1191 ; AVX1-LABEL: insert_reg_and_zero_v4f64:
1193 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1194 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1197 ; AVX2-LABEL: insert_reg_and_zero_v4f64:
1199 ; AVX2-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1200 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1203 ; AVX512VL-LABEL: insert_reg_and_zero_v4f64:
1205 ; AVX512VL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
1206 ; AVX512VL-NEXT: vmovsd %xmm0, %xmm1, %xmm0
1207 ; AVX512VL-NEXT: retq
1208 %v = insertelement <4 x double> undef, double %a, i32 0
1209 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1210 ret <4 x double> %shuffle
1213 define <4 x double> @insert_mem_and_zero_v4f64(double* %ptr) {
1214 ; AVX1-LABEL: insert_mem_and_zero_v4f64:
1216 ; AVX1-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1219 ; AVX2-LABEL: insert_mem_and_zero_v4f64:
1221 ; AVX2-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1224 ; AVX512VL-LABEL: insert_mem_and_zero_v4f64:
1226 ; AVX512VL-NEXT: vmovsd (%rdi), %xmm0
1227 ; AVX512VL-NEXT: retq
1228 %a = load double, double* %ptr
1229 %v = insertelement <4 x double> undef, double %a, i32 0
1230 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1231 ret <4 x double> %shuffle
1234 define <4 x double> @splat_mem_v4f64(double* %ptr) {
1235 ; ALL-LABEL: splat_mem_v4f64:
1237 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1239 %a = load double, double* %ptr
1240 %v = insertelement <4 x double> undef, double %a, i32 0
1241 %shuffle = shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1242 ret <4 x double> %shuffle
1245 define <4 x i64> @splat_mem_v4i64(i64* %ptr) {
1246 ; AVX1-LABEL: splat_mem_v4i64:
1248 ; AVX1-NEXT: vbroadcastsd (%rdi), %ymm0
1251 ; AVX2-LABEL: splat_mem_v4i64:
1253 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1256 ; AVX512VL-LABEL: splat_mem_v4i64:
1258 ; AVX512VL-NEXT: vpbroadcastq (%rdi), %ymm0
1259 ; AVX512VL-NEXT: retq
1260 %a = load i64, i64* %ptr
1261 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1262 %shuffle = shufflevector <4 x i64> %v, <4 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1263 ret <4 x i64> %shuffle
1266 define <4 x double> @splat_mem_v4f64_2(double* %p) {
1267 ; ALL-LABEL: splat_mem_v4f64_2:
1269 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1271 %1 = load double, double* %p
1272 %2 = insertelement <2 x double> undef, double %1, i32 0
1273 %3 = shufflevector <2 x double> %2, <2 x double> undef, <4 x i32> zeroinitializer
1277 define <4 x double> @splat_v4f64(<2 x double> %r) {
1278 ; AVX1-LABEL: splat_v4f64:
1280 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
1281 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1284 ; AVX2-LABEL: splat_v4f64:
1286 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
1289 ; AVX512VL-LABEL: splat_v4f64:
1291 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
1292 ; AVX512VL-NEXT: retq
1293 %1 = shufflevector <2 x double> %r, <2 x double> undef, <4 x i32> zeroinitializer
1297 define <4 x i64> @splat_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1298 ; AVX1-LABEL: splat_mem_v4i64_from_v2i64:
1300 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
1301 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1304 ; AVX2-LABEL: splat_mem_v4i64_from_v2i64:
1306 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1309 ; AVX512VL-LABEL: splat_mem_v4i64_from_v2i64:
1311 ; AVX512VL-NEXT: vmovdqa64 (%rdi), %xmm0
1312 ; AVX512VL-NEXT: vpbroadcastq %xmm0, %ymm0
1313 ; AVX512VL-NEXT: retq
1314 %v = load <2 x i64>, <2 x i64>* %ptr
1315 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1316 ret <4 x i64> %shuffle
1319 define <4 x double> @splat_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1320 ; AVX1-LABEL: splat_mem_v4f64_from_v2f64:
1322 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
1323 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1326 ; AVX2-LABEL: splat_mem_v4f64_from_v2f64:
1328 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1331 ; AVX512VL-LABEL: splat_mem_v4f64_from_v2f64:
1333 ; AVX512VL-NEXT: vbroadcastsd (%rdi), %ymm0
1334 ; AVX512VL-NEXT: retq
1335 %v = load <2 x double>, <2 x double>* %ptr
1336 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1337 ret <4 x double> %shuffle
1340 define <4 x i64> @splat128_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1341 ; AVX1-LABEL: splat128_mem_v4i64_from_v2i64:
1343 ; AVX1-NEXT: vmovaps (%rdi), %xmm0
1344 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1347 ; AVX2-LABEL: splat128_mem_v4i64_from_v2i64:
1349 ; AVX2-NEXT: vmovaps (%rdi), %xmm0
1350 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1353 ; AVX512VL-LABEL: splat128_mem_v4i64_from_v2i64:
1355 ; AVX512VL-NEXT: vmovdqa64 (%rdi), %xmm0
1356 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm0, %ymm0
1357 ; AVX512VL-NEXT: retq
1358 %v = load <2 x i64>, <2 x i64>* %ptr
1359 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1360 ret <4 x i64> %shuffle
1363 define <4 x double> @splat128_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1364 ; AVX1-LABEL: splat128_mem_v4f64_from_v2f64:
1366 ; AVX1-NEXT: vmovaps (%rdi), %xmm0
1367 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1370 ; AVX2-LABEL: splat128_mem_v4f64_from_v2f64:
1372 ; AVX2-NEXT: vmovaps (%rdi), %xmm0
1373 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1376 ; AVX512VL-LABEL: splat128_mem_v4f64_from_v2f64:
1378 ; AVX512VL-NEXT: vmovapd (%rdi), %xmm0
1379 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm0, %ymm0
1380 ; AVX512VL-NEXT: retq
1381 %v = load <2 x double>, <2 x double>* %ptr
1382 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1383 ret <4 x double> %shuffle
1386 define <4 x double> @bitcast_v4f64_0426(<4 x double> %a, <4 x double> %b) {
1387 ; AVX1-LABEL: bitcast_v4f64_0426:
1389 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1392 ; AVX2-LABEL: bitcast_v4f64_0426:
1394 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1397 ; AVX512VL-LABEL: bitcast_v4f64_0426:
1399 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1400 ; AVX512VL-NEXT: retq
1401 %shuffle64 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
1402 %bitcast32 = bitcast <4 x double> %shuffle64 to <8 x float>
1403 %shuffle32 = shufflevector <8 x float> %bitcast32, <8 x float> undef, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4>
1404 %bitcast16 = bitcast <8 x float> %shuffle32 to <16 x i16>
1405 %shuffle16 = shufflevector <16 x i16> %bitcast16, <16 x i16> undef, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5, i32 10, i32 11, i32 8, i32 9, i32 14, i32 15, i32 12, i32 13>
1406 %bitcast64 = bitcast <16 x i16> %shuffle16 to <4 x double>
1407 ret <4 x double> %bitcast64
1410 define <4 x i64> @concat_v4i64_0167(<4 x i64> %a0, <4 x i64> %a1) {
1411 ; AVX1-LABEL: concat_v4i64_0167:
1413 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1416 ; AVX2-LABEL: concat_v4i64_0167:
1418 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1421 ; AVX512VL-LABEL: concat_v4i64_0167:
1423 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1424 ; AVX512VL-NEXT: retq
1425 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1426 %a1hi = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 6, i32 7>
1427 %shuffle64 = shufflevector <2 x i64> %a0lo, <2 x i64> %a1hi, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1428 ret <4 x i64> %shuffle64
1431 define <4 x i64> @concat_v4i64_0145_bc(<4 x i64> %a0, <4 x i64> %a1) {
1432 ; AVX1-LABEL: concat_v4i64_0145_bc:
1434 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1437 ; AVX2-LABEL: concat_v4i64_0145_bc:
1439 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1442 ; AVX512VL-LABEL: concat_v4i64_0145_bc:
1444 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
1445 ; AVX512VL-NEXT: retq
1446 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1447 %a1lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 4, i32 5>
1448 %bc0lo = bitcast <2 x i64> %a0lo to <4 x i32>
1449 %bc1lo = bitcast <2 x i64> %a1lo to <4 x i32>
1450 %shuffle32 = shufflevector <4 x i32> %bc0lo, <4 x i32> %bc1lo, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
1451 %shuffle64 = bitcast <8 x i32> %shuffle32 to <4 x i64>
1452 ret <4 x i64> %shuffle64
1455 define <4 x i64> @insert_dup_mem_v4i64(i64* %ptr) {
1456 ; AVX1-LABEL: insert_dup_mem_v4i64:
1458 ; AVX1-NEXT: vbroadcastsd (%rdi), %ymm0
1461 ; AVX2-LABEL: insert_dup_mem_v4i64:
1463 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1466 ; AVX512VL-LABEL: insert_dup_mem_v4i64:
1468 ; AVX512VL-NEXT: vpbroadcastq (%rdi), %ymm0
1469 ; AVX512VL-NEXT: retq
1470 %tmp = load i64, i64* %ptr, align 1
1471 %tmp1 = insertelement <2 x i64> undef, i64 %tmp, i32 0
1472 %tmp2 = shufflevector <2 x i64> %tmp1, <2 x i64> undef, <4 x i32> zeroinitializer