1 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1
2 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --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 ; AVX-LABEL: shuffle_v4f64_0023:
175 ; AVX-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,3]
178 ; AVX512VL-LABEL: shuffle_v4f64_0023:
180 ; AVX512VL-NEXT: vpermilpd $8, %ymm0, %ymm0
181 ; AVX512VL-NEXT: retq
183 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 3>
184 ret <4 x double> %shuffle
187 define <4 x double> @shuffle_v4f64_0022(<4 x double> %a, <4 x double> %b) {
188 ; ALL-LABEL: shuffle_v4f64_0022:
190 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
192 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
193 ret <4 x double> %shuffle
196 define <4 x double> @shuffle_v4f64_1032(<4 x double> %a, <4 x double> %b) {
197 ; AVX-LABEL: shuffle_v4f64_1032:
199 ; AVX-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
202 ; AVX512VL-LABEL: shuffle_v4f64_1032:
204 ; AVX512VL-NEXT: vpermilpd $5, %ymm0, %ymm0
205 ; AVX512VL-NEXT: retq
206 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
207 ret <4 x double> %shuffle
210 define <4 x double> @shuffle_v4f64_1133(<4 x double> %a, <4 x double> %b) {
211 ; AVX-LABEL: shuffle_v4f64_1133:
213 ; AVX-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
216 ; AVX512VL-LABEL: shuffle_v4f64_1133:
218 ; AVX512VL-NEXT: vpermilpd $15, %ymm0, %ymm0
219 ; AVX512VL-NEXT: retq
220 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
221 ret <4 x double> %shuffle
224 define <4 x double> @shuffle_v4f64_1023(<4 x double> %a, <4 x double> %b) {
225 ; AVX-LABEL: shuffle_v4f64_1023:
227 ; AVX-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,3]
230 ; AVX512VL-LABEL: shuffle_v4f64_1023:
232 ; AVX512VL-NEXT: vpermilpd $9, %ymm0, %ymm0
233 ; AVX512VL-NEXT: retq
234 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 3>
235 ret <4 x double> %shuffle
238 define <4 x double> @shuffle_v4f64_1022(<4 x double> %a, <4 x double> %b) {
239 ; AVX-LABEL: shuffle_v4f64_1022:
241 ; AVX-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
244 ; AVX512VL-LABEL: shuffle_v4f64_1022:
246 ; AVX512VL-NEXT: vpermilpd $1, %ymm0, %ymm0
247 ; AVX512VL-NEXT: retq
248 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 2>
249 ret <4 x double> %shuffle
252 define <4 x double> @shuffle_v4f64_0423(<4 x double> %a, <4 x double> %b) {
253 ; AVX1-LABEL: shuffle_v4f64_0423:
255 ; AVX1-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
256 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
259 ; AVX2-LABEL: shuffle_v4f64_0423:
261 ; AVX2-NEXT: vbroadcastsd %xmm1, %ymm1
262 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
265 ; AVX512VL-LABEL: shuffle_v4f64_0423:
267 ; AVX512VL-NEXT: vbroadcastsd %xmm1, %ymm1
268 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
269 ; AVX512VL-NEXT: retq
270 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 3>
271 ret <4 x double> %shuffle
274 define <4 x double> @shuffle_v4f64_0462(<4 x double> %a, <4 x double> %b) {
275 ; ALL-LABEL: shuffle_v4f64_0462:
277 ; ALL-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
278 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
279 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
281 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 6, i32 2>
282 ret <4 x double> %shuffle
285 define <4 x double> @shuffle_v4f64_0426(<4 x double> %a, <4 x double> %b) {
286 ; AVX-LABEL: shuffle_v4f64_0426:
288 ; AVX-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
291 ; AVX512VL-LABEL: shuffle_v4f64_0426:
293 ; AVX512VL-NEXT: vunpcklpd %ymm1, %ymm0, %ymm0
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 6>
296 ret <4 x double> %shuffle
299 define <4 x double> @shuffle_v4f64_1537(<4 x double> %a, <4 x double> %b) {
300 ; AVX-LABEL: shuffle_v4f64_1537:
302 ; AVX-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
305 ; AVX512VL-LABEL: shuffle_v4f64_1537:
307 ; AVX512VL-NEXT: vunpckhpd %ymm1, %ymm0, %ymm0
308 ; AVX512VL-NEXT: retq
309 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
310 ret <4 x double> %shuffle
313 define <4 x double> @shuffle_v4f64_4062(<4 x double> %a, <4 x double> %b) {
314 ; AVX-LABEL: shuffle_v4f64_4062:
316 ; AVX-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
319 ; AVX512VL-LABEL: shuffle_v4f64_4062:
321 ; AVX512VL-NEXT: vunpcklpd %ymm0, %ymm1, %ymm0
322 ; AVX512VL-NEXT: retq
323 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
324 ret <4 x double> %shuffle
327 define <4 x double> @shuffle_v4f64_5173(<4 x double> %a, <4 x double> %b) {
328 ; AVX-LABEL: shuffle_v4f64_5173:
330 ; AVX-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
333 ; AVX512VL-LABEL: shuffle_v4f64_5173:
335 ; AVX512VL-NEXT: vunpckhpd %ymm0, %ymm1, %ymm0
336 ; AVX512VL-NEXT: retq
337 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 7, i32 3>
338 ret <4 x double> %shuffle
341 define <4 x double> @shuffle_v4f64_5163(<4 x double> %a, <4 x double> %b) {
342 ; AVX-LABEL: shuffle_v4f64_5163:
344 ; AVX-NEXT: vshufpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
347 ; AVX512VL-LABEL: shuffle_v4f64_5163:
349 ; AVX512VL-NEXT: vshufpd $11, %ymm0, %ymm1, %ymm0
350 ; AVX512VL-NEXT: retq
351 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
352 ret <4 x double> %shuffle
355 define <4 x double> @shuffle_v4f64_0527(<4 x double> %a, <4 x double> %b) {
356 ; ALL-LABEL: shuffle_v4f64_0527:
358 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
360 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
361 ret <4 x double> %shuffle
364 define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
365 ; ALL-LABEL: shuffle_v4f64_4163:
367 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
369 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
370 ret <4 x double> %shuffle
373 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
374 ; AVX-LABEL: shuffle_v4f64_0145:
376 ; AVX-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
379 ; AVX512VL-LABEL: shuffle_v4f64_0145:
381 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
382 ; AVX512VL-NEXT: retq
383 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
384 ret <4 x double> %shuffle
387 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
388 ; AVX-LABEL: shuffle_v4f64_4501:
390 ; AVX-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
393 ; AVX512VL-LABEL: shuffle_v4f64_4501:
395 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm1, %ymm0
396 ; AVX512VL-NEXT: retq
397 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
398 ret <4 x double> %shuffle
401 define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
402 ; ALL-LABEL: shuffle_v4f64_0167:
404 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
406 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
407 ret <4 x double> %shuffle
410 define <4 x double> @shuffle_v4f64_1054(<4 x double> %a, <4 x double> %b) {
411 ; AVX-LABEL: shuffle_v4f64_1054:
413 ; AVX-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
414 ; AVX-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
417 ; AVX512VL-LABEL: shuffle_v4f64_1054:
419 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
420 ; AVX512VL-NEXT: vpermilpd $5, %ymm0, %ymm0
421 ; AVX512VL-NEXT: retq
422 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
423 ret <4 x double> %shuffle
426 define <4 x double> @shuffle_v4f64_3254(<4 x double> %a, <4 x double> %b) {
427 ; AVX-LABEL: shuffle_v4f64_3254:
429 ; AVX-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
430 ; AVX-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
433 ; AVX512VL-LABEL: shuffle_v4f64_3254:
435 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
436 ; AVX512VL-NEXT: vpermilpd $5, %ymm0, %ymm0
437 ; AVX512VL-NEXT: retq
438 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
439 ret <4 x double> %shuffle
442 define <4 x double> @shuffle_v4f64_3276(<4 x double> %a, <4 x double> %b) {
443 ; AVX-LABEL: shuffle_v4f64_3276:
445 ; AVX-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
446 ; AVX-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
449 ; AVX512VL-LABEL: shuffle_v4f64_3276:
451 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
452 ; AVX512VL-NEXT: vpermilpd $5, %ymm0, %ymm0
453 ; AVX512VL-NEXT: retq
454 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
455 ret <4 x double> %shuffle
458 define <4 x double> @shuffle_v4f64_1076(<4 x double> %a, <4 x double> %b) {
459 ; AVX-LABEL: shuffle_v4f64_1076:
461 ; AVX-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
462 ; AVX-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
465 ; AVX512VL-LABEL: shuffle_v4f64_1076:
467 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
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 1, i32 0, i32 7, i32 6>
471 ret <4 x double> %shuffle
474 define <4 x double> @shuffle_v4f64_0415(<4 x double> %a, <4 x double> %b) {
475 ; AVX1-LABEL: shuffle_v4f64_0415:
477 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
478 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
479 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
482 ; AVX2-LABEL: shuffle_v4f64_0415:
484 ; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
485 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
486 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
489 ; AVX512VL-LABEL: shuffle_v4f64_0415:
491 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
492 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
493 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
494 ; AVX512VL-NEXT: retq
495 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
496 ret <4 x double> %shuffle
499 define <4 x double> @shuffle_v4f64_u062(<4 x double> %a, <4 x double> %b) {
500 ; AVX-LABEL: shuffle_v4f64_u062:
502 ; AVX-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
505 ; AVX512VL-LABEL: shuffle_v4f64_u062:
507 ; AVX512VL-NEXT: vunpcklpd %ymm0, %ymm1, %ymm0
508 ; AVX512VL-NEXT: retq
509 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 undef, i32 0, i32 6, i32 2>
510 ret <4 x double> %shuffle
513 define <4 x i64> @shuffle_v4i64_0000(<4 x i64> %a, <4 x i64> %b) {
514 ; AVX1-LABEL: shuffle_v4i64_0000:
516 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
517 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
520 ; AVX2-LABEL: shuffle_v4i64_0000:
522 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
525 ; AVX512VL-LABEL: shuffle_v4i64_0000:
527 ; AVX512VL-NEXT: vpbroadcastq %xmm0, %ymm0
528 ; AVX512VL-NEXT: retq
529 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
530 ret <4 x i64> %shuffle
533 define <4 x i64> @shuffle_v4i64_0001(<4 x i64> %a, <4 x i64> %b) {
534 ; AVX1-LABEL: shuffle_v4i64_0001:
536 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm0[0,0]
537 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
540 ; AVX2-LABEL: shuffle_v4i64_0001:
542 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
545 ; AVX512VL-LABEL: shuffle_v4i64_0001:
547 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
548 ; AVX512VL-NEXT: retq
549 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
550 ret <4 x i64> %shuffle
553 define <4 x i64> @shuffle_v4i64_0020(<4 x i64> %a, <4 x i64> %b) {
554 ; AVX1-LABEL: shuffle_v4i64_0020:
556 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
557 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
558 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
559 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
562 ; AVX2-LABEL: shuffle_v4i64_0020:
564 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
567 ; AVX512VL-LABEL: shuffle_v4i64_0020:
569 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
570 ; AVX512VL-NEXT: retq
571 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
572 ret <4 x i64> %shuffle
575 define <4 x i64> @shuffle_v4i64_0112(<4 x i64> %a, <4 x i64> %b) {
576 ; AVX1-LABEL: shuffle_v4i64_0112:
578 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
579 ; AVX1-NEXT: vshufpd {{.*#+}} xmm1 = xmm0[1],xmm1[0]
580 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
583 ; AVX2-LABEL: shuffle_v4i64_0112:
585 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
588 ; AVX512VL-LABEL: shuffle_v4i64_0112:
590 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
591 ; AVX512VL-NEXT: retq
592 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 1, i32 2>
593 ret <4 x i64> %shuffle
596 define <4 x i64> @shuffle_v4i64_0300(<4 x i64> %a, <4 x i64> %b) {
597 ; AVX1-LABEL: shuffle_v4i64_0300:
599 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
600 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
601 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
604 ; AVX2-LABEL: shuffle_v4i64_0300:
606 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
609 ; AVX512VL-LABEL: shuffle_v4i64_0300:
611 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
612 ; AVX512VL-NEXT: retq
613 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
614 ret <4 x i64> %shuffle
617 define <4 x i64> @shuffle_v4i64_1000(<4 x i64> %a, <4 x i64> %b) {
618 ; AVX1-LABEL: shuffle_v4i64_1000:
620 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
621 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
622 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
625 ; AVX2-LABEL: shuffle_v4i64_1000:
627 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
630 ; AVX512VL-LABEL: shuffle_v4i64_1000:
632 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
633 ; AVX512VL-NEXT: retq
634 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
635 ret <4 x i64> %shuffle
638 define <4 x i64> @shuffle_v4i64_2200(<4 x i64> %a, <4 x i64> %b) {
639 ; AVX1-LABEL: shuffle_v4i64_2200:
641 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
642 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
645 ; AVX2-LABEL: shuffle_v4i64_2200:
647 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
650 ; AVX512VL-LABEL: shuffle_v4i64_2200:
652 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
653 ; AVX512VL-NEXT: retq
654 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
655 ret <4 x i64> %shuffle
658 define <4 x i64> @shuffle_v4i64_3330(<4 x i64> %a, <4 x i64> %b) {
659 ; AVX1-LABEL: shuffle_v4i64_3330:
661 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
662 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3]
663 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
666 ; AVX2-LABEL: shuffle_v4i64_3330:
668 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
671 ; AVX512VL-LABEL: shuffle_v4i64_3330:
673 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
674 ; AVX512VL-NEXT: retq
675 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
676 ret <4 x i64> %shuffle
679 define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
680 ; AVX1-LABEL: shuffle_v4i64_3210:
682 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
683 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
686 ; AVX2-LABEL: shuffle_v4i64_3210:
688 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
691 ; AVX512VL-LABEL: shuffle_v4i64_3210:
693 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
694 ; AVX512VL-NEXT: retq
695 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
696 ret <4 x i64> %shuffle
699 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
700 ; AVX1-LABEL: shuffle_v4i64_0124:
702 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
703 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
704 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3]
707 ; AVX2-LABEL: shuffle_v4i64_0124:
709 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
710 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
713 ; AVX512VL-LABEL: shuffle_v4i64_0124:
715 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
716 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
717 ; AVX512VL-NEXT: retq
718 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
719 ret <4 x i64> %shuffle
722 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
723 ; AVX1-LABEL: shuffle_v4i64_0142:
725 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm1, %ymm1
726 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,1,2,2]
727 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
730 ; AVX2-LABEL: shuffle_v4i64_0142:
732 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
733 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
734 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
737 ; AVX512VL-LABEL: shuffle_v4i64_0142:
739 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm1, %ymm1
740 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
741 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
742 ; AVX512VL-NEXT: retq
743 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
744 ret <4 x i64> %shuffle
747 define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
748 ; AVX1-LABEL: shuffle_v4i64_0412:
750 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
751 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
752 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
753 ; AVX1-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
754 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
757 ; AVX2-LABEL: shuffle_v4i64_0412:
759 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
760 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
761 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
764 ; AVX512VL-LABEL: shuffle_v4i64_0412:
766 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
767 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
768 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
769 ; AVX512VL-NEXT: retq
770 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
771 ret <4 x i64> %shuffle
774 define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
775 ; AVX1-LABEL: shuffle_v4i64_4012:
777 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
778 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
779 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
780 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
781 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
784 ; AVX2-LABEL: shuffle_v4i64_4012:
786 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
787 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
790 ; AVX512VL-LABEL: shuffle_v4i64_4012:
792 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
793 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
794 ; AVX512VL-NEXT: retq
795 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
796 ret <4 x i64> %shuffle
799 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
800 ; AVX-LABEL: shuffle_v4i64_0145:
802 ; AVX-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
805 ; AVX512VL-LABEL: shuffle_v4i64_0145:
807 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
808 ; AVX512VL-NEXT: retq
809 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
810 ret <4 x i64> %shuffle
813 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
814 ; AVX1-LABEL: shuffle_v4i64_0451:
816 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm1[1],xmm0[1]
817 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
818 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
821 ; AVX2-LABEL: shuffle_v4i64_0451:
823 ; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
824 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
825 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
828 ; AVX512VL-LABEL: shuffle_v4i64_0451:
830 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm0, %ymm0
831 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
832 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
833 ; AVX512VL-NEXT: retq
834 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
835 ret <4 x i64> %shuffle
838 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
839 ; AVX-LABEL: shuffle_v4i64_4501:
841 ; AVX-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
844 ; AVX512VL-LABEL: shuffle_v4i64_4501:
846 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm1, %ymm0
847 ; AVX512VL-NEXT: retq
848 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
849 ret <4 x i64> %shuffle
852 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
853 ; AVX1-LABEL: shuffle_v4i64_4015:
855 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
856 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
857 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
860 ; AVX2-LABEL: shuffle_v4i64_4015:
862 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
863 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
864 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
867 ; AVX512VL-LABEL: shuffle_v4i64_4015:
869 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm1, %ymm1
870 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
871 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
872 ; AVX512VL-NEXT: retq
873 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
874 ret <4 x i64> %shuffle
877 define <4 x i64> @shuffle_v4i64_2u35(<4 x i64> %a, <4 x i64> %b) {
878 ; AVX1-LABEL: shuffle_v4i64_2u35:
880 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
881 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm1 = xmm0[1],xmm1[1]
882 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
885 ; AVX2-LABEL: shuffle_v4i64_2u35:
887 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
888 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
891 ; AVX512VL-LABEL: shuffle_v4i64_2u35:
893 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
894 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
895 ; AVX512VL-NEXT: retq
896 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 undef, i32 3, i32 5>
897 ret <4 x i64> %shuffle
900 define <4 x i64> @shuffle_v4i64_1251(<4 x i64> %a, <4 x i64> %b) {
901 ; AVX1-LABEL: shuffle_v4i64_1251:
903 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
904 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm2[0],ymm0[2],ymm2[3]
905 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
906 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
907 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
910 ; AVX2-LABEL: shuffle_v4i64_1251:
912 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
913 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
914 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
917 ; AVX512VL-LABEL: shuffle_v4i64_1251:
919 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
920 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
921 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
922 ; AVX512VL-NEXT: retq
923 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 5, i32 1>
924 ret <4 x i64> %shuffle
927 define <4 x i64> @shuffle_v4i64_1054(<4 x i64> %a, <4 x i64> %b) {
928 ; AVX1-LABEL: shuffle_v4i64_1054:
930 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
931 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
934 ; AVX2-LABEL: shuffle_v4i64_1054:
936 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
937 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
940 ; AVX512VL-LABEL: shuffle_v4i64_1054:
942 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
943 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
944 ; AVX512VL-NEXT: retq
945 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
946 ret <4 x i64> %shuffle
949 define <4 x i64> @shuffle_v4i64_3254(<4 x i64> %a, <4 x i64> %b) {
950 ; AVX1-LABEL: shuffle_v4i64_3254:
952 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
953 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
956 ; AVX2-LABEL: shuffle_v4i64_3254:
958 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
959 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
962 ; AVX512VL-LABEL: shuffle_v4i64_3254:
964 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
965 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
966 ; AVX512VL-NEXT: retq
967 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
968 ret <4 x i64> %shuffle
971 define <4 x i64> @shuffle_v4i64_3276(<4 x i64> %a, <4 x i64> %b) {
972 ; AVX1-LABEL: shuffle_v4i64_3276:
974 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
975 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
978 ; AVX2-LABEL: shuffle_v4i64_3276:
980 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
981 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
984 ; AVX512VL-LABEL: shuffle_v4i64_3276:
986 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
987 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
988 ; AVX512VL-NEXT: retq
989 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
990 ret <4 x i64> %shuffle
993 define <4 x i64> @shuffle_v4i64_1076(<4 x i64> %a, <4 x i64> %b) {
994 ; AVX1-LABEL: shuffle_v4i64_1076:
996 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
997 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1000 ; AVX2-LABEL: shuffle_v4i64_1076:
1002 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1003 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1006 ; AVX512VL-LABEL: shuffle_v4i64_1076:
1008 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1009 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1010 ; AVX512VL-NEXT: retq
1011 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
1012 ret <4 x i64> %shuffle
1015 define <4 x i64> @shuffle_v4i64_0415(<4 x i64> %a, <4 x i64> %b) {
1016 ; AVX1-LABEL: shuffle_v4i64_0415:
1018 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
1019 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
1020 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
1023 ; AVX2-LABEL: shuffle_v4i64_0415:
1025 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1026 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1027 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1030 ; AVX512VL-LABEL: shuffle_v4i64_0415:
1032 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1033 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1034 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1035 ; AVX512VL-NEXT: retq
1036 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
1037 ret <4 x i64> %shuffle
1040 define <4 x i64> @shuffle_v4i64_z4z6(<4 x i64> %a) {
1041 ; AVX1-LABEL: shuffle_v4i64_z4z6:
1043 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1044 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1047 ; AVX2-LABEL: shuffle_v4i64_z4z6:
1049 ; 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]
1052 ; AVX512VL-LABEL: shuffle_v4i64_z4z6:
1054 ; 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]
1055 ; AVX512VL-NEXT: retq
1056 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 0, i32 4, i32 0, i32 6>
1057 ret <4 x i64> %shuffle
1060 define <4 x i64> @shuffle_v4i64_5zuz(<4 x i64> %a) {
1061 ; AVX1-LABEL: shuffle_v4i64_5zuz:
1063 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1064 ; AVX1-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
1067 ; AVX2-LABEL: shuffle_v4i64_5zuz:
1069 ; 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
1072 ; AVX512VL-LABEL: shuffle_v4i64_5zuz:
1074 ; 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
1075 ; AVX512VL-NEXT: retq
1076 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 5, i32 0, i32 undef, i32 0>
1077 ret <4 x i64> %shuffle
1080 define <4 x i64> @shuffle_v4i64_40u2(<4 x i64> %a, <4 x i64> %b) {
1081 ; AVX1-LABEL: shuffle_v4i64_40u2:
1083 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1086 ; AVX2-LABEL: shuffle_v4i64_40u2:
1088 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1091 ; AVX512VL-LABEL: shuffle_v4i64_40u2:
1093 ; AVX512VL-NEXT: vpunpcklqdq %ymm0, %ymm1, %ymm0
1094 ; AVX512VL-NEXT: retq
1095 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 undef, i32 2>
1096 ret <4 x i64> %shuffle
1099 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
1100 ; ALL-LABEL: stress_test1:
1102 %c = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 1, i32 1, i32 0>
1103 %d = shufflevector <4 x i64> %c, <4 x i64> undef, <4 x i32> <i32 3, i32 undef, i32 2, i32 undef>
1104 %e = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 undef>
1105 %f = shufflevector <4 x i64> %d, <4 x i64> %e, <4 x i32> <i32 5, i32 1, i32 1, i32 0>
1110 define <4 x i64> @insert_reg_and_zero_v4i64(i64 %a) {
1111 ; ALL-LABEL: insert_reg_and_zero_v4i64:
1113 ; ALL-NEXT: vmovq %rdi, %xmm0
1115 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1116 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1117 ret <4 x i64> %shuffle
1120 define <4 x i64> @insert_mem_and_zero_v4i64(i64* %ptr) {
1121 ; AVX-LABEL: insert_mem_and_zero_v4i64:
1123 ; AVX-NEXT: vmovq {{.*#+}} xmm0 = mem[0],zero
1126 ; AVX512VL-LABEL: insert_mem_and_zero_v4i64:
1128 ; AVX512VL-NEXT: vmovq (%rdi), %xmm0
1129 ; AVX512VL-NEXT: retq
1130 %a = load i64, i64* %ptr
1131 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1132 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1133 ret <4 x i64> %shuffle
1136 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
1137 ; AVX-LABEL: insert_reg_and_zero_v4f64:
1139 ; AVX-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1140 ; AVX-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1143 ; AVX512VL-LABEL: insert_reg_and_zero_v4f64:
1145 ; AVX512VL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
1146 ; AVX512VL-NEXT: vmovsd %xmm0, %xmm1, %xmm0
1147 ; AVX512VL-NEXT: retq
1148 %v = insertelement <4 x double> undef, double %a, i32 0
1149 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1150 ret <4 x double> %shuffle
1153 define <4 x double> @insert_mem_and_zero_v4f64(double* %ptr) {
1154 ; AVX-LABEL: insert_mem_and_zero_v4f64:
1156 ; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1159 ; AVX512VL-LABEL: insert_mem_and_zero_v4f64:
1161 ; AVX512VL-NEXT: vmovsd (%rdi), %xmm0
1162 ; AVX512VL-NEXT: retq
1163 %a = load double, double* %ptr
1164 %v = insertelement <4 x double> undef, double %a, i32 0
1165 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1166 ret <4 x double> %shuffle
1169 define <4 x double> @splat_mem_v4f64(double* %ptr) {
1170 ; ALL-LABEL: splat_mem_v4f64:
1172 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1174 %a = load double, double* %ptr
1175 %v = insertelement <4 x double> undef, double %a, i32 0
1176 %shuffle = shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1177 ret <4 x double> %shuffle
1180 define <4 x i64> @splat_mem_v4i64(i64* %ptr) {
1181 ; AVX-LABEL: splat_mem_v4i64:
1183 ; AVX-NEXT: vbroadcastsd (%rdi), %ymm0
1186 ; AVX512VL-LABEL: splat_mem_v4i64:
1188 ; AVX512VL-NEXT: vpbroadcastq (%rdi), %ymm0
1189 ; AVX512VL-NEXT: retq
1190 %a = load i64, i64* %ptr
1191 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1192 %shuffle = shufflevector <4 x i64> %v, <4 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1193 ret <4 x i64> %shuffle
1196 define <4 x double> @splat_mem_v4f64_2(double* %p) {
1197 ; ALL-LABEL: splat_mem_v4f64_2:
1199 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1201 %1 = load double, double* %p
1202 %2 = insertelement <2 x double> undef, double %1, i32 0
1203 %3 = shufflevector <2 x double> %2, <2 x double> undef, <4 x i32> zeroinitializer
1207 define <4 x double> @splat_v4f64(<2 x double> %r) {
1208 ; AVX1-LABEL: splat_v4f64:
1210 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
1211 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1214 ; AVX2-LABEL: splat_v4f64:
1216 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
1219 ; AVX512VL-LABEL: splat_v4f64:
1221 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
1222 ; AVX512VL-NEXT: retq
1223 %1 = shufflevector <2 x double> %r, <2 x double> undef, <4 x i32> zeroinitializer
1227 define <4 x i64> @splat_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1228 ; AVX1-LABEL: splat_mem_v4i64_from_v2i64:
1230 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
1231 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1234 ; AVX2-LABEL: splat_mem_v4i64_from_v2i64:
1236 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1239 ; AVX512VL-LABEL: splat_mem_v4i64_from_v2i64:
1241 ; AVX512VL-NEXT: vmovdqa64 (%rdi), %xmm0
1242 ; AVX512VL-NEXT: vpbroadcastq %xmm0, %ymm0
1243 ; AVX512VL-NEXT: retq
1244 %v = load <2 x i64>, <2 x i64>* %ptr
1245 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1246 ret <4 x i64> %shuffle
1249 define <4 x double> @splat_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1250 ; AVX1-LABEL: splat_mem_v4f64_from_v2f64:
1252 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
1253 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1256 ; AVX2-LABEL: splat_mem_v4f64_from_v2f64:
1258 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1261 ; AVX512VL-LABEL: splat_mem_v4f64_from_v2f64:
1263 ; AVX512VL-NEXT: vbroadcastsd (%rdi), %ymm0
1264 ; AVX512VL-NEXT: retq
1265 %v = load <2 x double>, <2 x double>* %ptr
1266 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1267 ret <4 x double> %shuffle
1270 define <4 x i64> @splat128_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1271 ; AVX-LABEL: splat128_mem_v4i64_from_v2i64:
1273 ; AVX-NEXT: vmovaps (%rdi), %xmm0
1274 ; AVX-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1277 ; AVX512VL-LABEL: splat128_mem_v4i64_from_v2i64:
1279 ; AVX512VL-NEXT: vmovdqa64 (%rdi), %xmm0
1280 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm0, %ymm0
1281 ; AVX512VL-NEXT: retq
1282 %v = load <2 x i64>, <2 x i64>* %ptr
1283 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1284 ret <4 x i64> %shuffle
1287 define <4 x double> @splat128_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1288 ; AVX-LABEL: splat128_mem_v4f64_from_v2f64:
1290 ; AVX-NEXT: vmovaps (%rdi), %xmm0
1291 ; AVX-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1294 ; AVX512VL-LABEL: splat128_mem_v4f64_from_v2f64:
1296 ; AVX512VL-NEXT: vmovapd (%rdi), %xmm0
1297 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm0, %ymm0
1298 ; AVX512VL-NEXT: retq
1299 %v = load <2 x double>, <2 x double>* %ptr
1300 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1301 ret <4 x double> %shuffle
1304 define <4 x double> @bitcast_v4f64_0426(<4 x double> %a, <4 x double> %b) {
1305 ; AVX1-LABEL: bitcast_v4f64_0426:
1307 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1310 ; AVX2-LABEL: bitcast_v4f64_0426:
1312 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1315 ; AVX512VL-LABEL: bitcast_v4f64_0426:
1317 ; AVX512VL-NEXT: vpunpcklqdq %ymm1, %ymm0, %ymm0
1318 ; AVX512VL-NEXT: retq
1319 %shuffle64 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
1320 %bitcast32 = bitcast <4 x double> %shuffle64 to <8 x float>
1321 %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>
1322 %bitcast16 = bitcast <8 x float> %shuffle32 to <16 x i16>
1323 %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>
1324 %bitcast64 = bitcast <16 x i16> %shuffle16 to <4 x double>
1325 ret <4 x double> %bitcast64
1328 define <4 x i64> @concat_v4i64_0167(<4 x i64> %a0, <4 x i64> %a1) {
1329 ; AVX1-LABEL: concat_v4i64_0167:
1331 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1334 ; AVX2-LABEL: concat_v4i64_0167:
1336 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1339 ; AVX512VL-LABEL: concat_v4i64_0167:
1341 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1342 ; AVX512VL-NEXT: retq
1343 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1344 %a1hi = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 6, i32 7>
1345 %shuffle64 = shufflevector <2 x i64> %a0lo, <2 x i64> %a1hi, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1346 ret <4 x i64> %shuffle64
1349 define <4 x i64> @concat_v4i64_0145_bc(<4 x i64> %a0, <4 x i64> %a1) {
1350 ; AVX-LABEL: concat_v4i64_0145_bc:
1352 ; AVX-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1355 ; AVX512VL-LABEL: concat_v4i64_0145_bc:
1357 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
1358 ; AVX512VL-NEXT: retq
1359 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1360 %a1lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 4, i32 5>
1361 %bc0lo = bitcast <2 x i64> %a0lo to <4 x i32>
1362 %bc1lo = bitcast <2 x i64> %a1lo to <4 x i32>
1363 %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>
1364 %shuffle64 = bitcast <8 x i32> %shuffle32 to <4 x i64>
1365 ret <4 x i64> %shuffle64
1368 define <4 x i64> @insert_dup_mem_v4i64(i64* %ptr) {
1369 ; AVX-LABEL: insert_dup_mem_v4i64:
1371 ; AVX-NEXT: vbroadcastsd (%rdi), %ymm0
1374 ; AVX512VL-LABEL: insert_dup_mem_v4i64:
1376 ; AVX512VL-NEXT: vpbroadcastq (%rdi), %ymm0
1377 ; AVX512VL-NEXT: retq
1378 %tmp = load i64, i64* %ptr, align 1
1379 %tmp1 = insertelement <2 x i64> undef, i64 %tmp, i32 0
1380 %tmp2 = shufflevector <2 x i64> %tmp1, <2 x i64> undef, <4 x i32> zeroinitializer