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 ; ALL-LABEL: shuffle_v4f64_0023:
175 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,3]
178 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 3>
179 ret <4 x double> %shuffle
182 define <4 x double> @shuffle_v4f64_0022(<4 x double> %a, <4 x double> %b) {
183 ; ALL-LABEL: shuffle_v4f64_0022:
185 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
187 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
188 ret <4 x double> %shuffle
191 define <4 x double> @shuffle_v4f64mem_0022(<4 x double>* %ptr, <4 x double> %b) {
192 ; ALL-LABEL: shuffle_v4f64mem_0022:
194 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = mem[0,0,2,2]
196 %a = load <4 x double>, <4 x double>* %ptr
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 ; ALL-LABEL: shuffle_v4f64_1032:
204 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
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 ; ALL-LABEL: shuffle_v4f64_1133:
213 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
215 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
216 ret <4 x double> %shuffle
219 define <4 x double> @shuffle_v4f64_1023(<4 x double> %a, <4 x double> %b) {
220 ; ALL-LABEL: shuffle_v4f64_1023:
222 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,3]
224 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 3>
225 ret <4 x double> %shuffle
228 define <4 x double> @shuffle_v4f64_1022(<4 x double> %a, <4 x double> %b) {
229 ; ALL-LABEL: shuffle_v4f64_1022:
231 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
233 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 2>
234 ret <4 x double> %shuffle
237 define <4 x double> @shuffle_v4f64_0423(<4 x double> %a, <4 x double> %b) {
238 ; AVX1-LABEL: shuffle_v4f64_0423:
240 ; AVX1-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
241 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
244 ; AVX2-LABEL: shuffle_v4f64_0423:
246 ; AVX2-NEXT: vbroadcastsd %xmm1, %ymm1
247 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
250 ; AVX512VL-LABEL: shuffle_v4f64_0423:
252 ; AVX512VL-NEXT: vbroadcastsd %xmm1, %ymm1
253 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
254 ; AVX512VL-NEXT: retq
255 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 3>
256 ret <4 x double> %shuffle
259 define <4 x double> @shuffle_v4f64_0462(<4 x double> %a, <4 x double> %b) {
260 ; ALL-LABEL: shuffle_v4f64_0462:
262 ; ALL-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
263 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
264 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
266 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 6, i32 2>
267 ret <4 x double> %shuffle
270 define <4 x double> @shuffle_v4f64_0426(<4 x double> %a, <4 x double> %b) {
271 ; ALL-LABEL: shuffle_v4f64_0426:
273 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
275 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
276 ret <4 x double> %shuffle
279 define <4 x double> @shuffle_v4f64_1537(<4 x double> %a, <4 x double> %b) {
280 ; ALL-LABEL: shuffle_v4f64_1537:
282 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
284 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
285 ret <4 x double> %shuffle
288 define <4 x double> @shuffle_v4f64_4062(<4 x double> %a, <4 x double> %b) {
289 ; ALL-LABEL: shuffle_v4f64_4062:
291 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
293 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
294 ret <4 x double> %shuffle
297 define <4 x double> @shuffle_v4f64_5173(<4 x double> %a, <4 x double> %b) {
298 ; ALL-LABEL: shuffle_v4f64_5173:
300 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
302 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 7, i32 3>
303 ret <4 x double> %shuffle
306 define <4 x double> @shuffle_v4f64_5163(<4 x double> %a, <4 x double> %b) {
307 ; ALL-LABEL: shuffle_v4f64_5163:
309 ; ALL-NEXT: vshufpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
311 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
312 ret <4 x double> %shuffle
315 define <4 x double> @shuffle_v4f64_0527(<4 x double> %a, <4 x double> %b) {
316 ; ALL-LABEL: shuffle_v4f64_0527:
318 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
320 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
321 ret <4 x double> %shuffle
324 define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
325 ; ALL-LABEL: shuffle_v4f64_4163:
327 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
329 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
330 ret <4 x double> %shuffle
333 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
334 ; AVX1-LABEL: shuffle_v4f64_0145:
336 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
339 ; AVX2-LABEL: shuffle_v4f64_0145:
341 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
344 ; AVX512VL-LABEL: shuffle_v4f64_0145:
346 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
347 ; AVX512VL-NEXT: retq
348 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
349 ret <4 x double> %shuffle
352 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
353 ; AVX1-LABEL: shuffle_v4f64_4501:
355 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
358 ; AVX2-LABEL: shuffle_v4f64_4501:
360 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
363 ; AVX512VL-LABEL: shuffle_v4f64_4501:
365 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm1, %ymm0
366 ; AVX512VL-NEXT: retq
367 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
368 ret <4 x double> %shuffle
371 define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
372 ; ALL-LABEL: shuffle_v4f64_0167:
374 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
376 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
377 ret <4 x double> %shuffle
380 define <4 x double> @shuffle_v4f64_1054(<4 x double> %a, <4 x double> %b) {
381 ; AVX1-LABEL: shuffle_v4f64_1054:
383 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
384 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
387 ; AVX2-LABEL: shuffle_v4f64_1054:
389 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
390 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
393 ; AVX512VL-LABEL: shuffle_v4f64_1054:
395 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
396 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
397 ; AVX512VL-NEXT: retq
398 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
399 ret <4 x double> %shuffle
402 define <4 x double> @shuffle_v4f64_3254(<4 x double> %a, <4 x double> %b) {
403 ; AVX1-LABEL: shuffle_v4f64_3254:
405 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
406 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
409 ; AVX2-LABEL: shuffle_v4f64_3254:
411 ; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
412 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
415 ; AVX512VL-LABEL: shuffle_v4f64_3254:
417 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
418 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
419 ; AVX512VL-NEXT: retq
420 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
421 ret <4 x double> %shuffle
424 define <4 x double> @shuffle_v4f64_3276(<4 x double> %a, <4 x double> %b) {
425 ; AVX1-LABEL: shuffle_v4f64_3276:
427 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
428 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
431 ; AVX2-LABEL: shuffle_v4f64_3276:
433 ; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
434 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
437 ; AVX512VL-LABEL: shuffle_v4f64_3276:
439 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
440 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
441 ; AVX512VL-NEXT: retq
442 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
443 ret <4 x double> %shuffle
446 define <4 x double> @shuffle_v4f64_1076(<4 x double> %a, <4 x double> %b) {
447 ; ALL-LABEL: shuffle_v4f64_1076:
449 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
450 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
452 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
453 ret <4 x double> %shuffle
456 define <4 x double> @shuffle_v4f64_0415(<4 x double> %a, <4 x double> %b) {
457 ; AVX1-LABEL: shuffle_v4f64_0415:
459 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
460 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
461 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
464 ; AVX2-LABEL: shuffle_v4f64_0415:
466 ; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
467 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
468 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
471 ; AVX512VL-LABEL: shuffle_v4f64_0415:
473 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
474 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
475 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
476 ; AVX512VL-NEXT: retq
477 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
478 ret <4 x double> %shuffle
481 define <4 x double> @shuffle_v4f64_u062(<4 x double> %a, <4 x double> %b) {
482 ; ALL-LABEL: shuffle_v4f64_u062:
484 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
486 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 undef, i32 0, i32 6, i32 2>
487 ret <4 x double> %shuffle
490 define <4 x double> @shuffle_v4f64_11uu(<4 x double> %a, <4 x double> %b) {
491 ; ALL-LABEL: shuffle_v4f64_11uu:
493 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,2,2]
495 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 undef, i32 undef>
496 ret <4 x double> %shuffle
499 define <4 x double> @shuffle_v4f64_22uu(<4 x double> %a, <4 x double> %b) {
500 ; AVX1-LABEL: shuffle_v4f64_22uu:
502 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
503 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
506 ; AVX2-LABEL: shuffle_v4f64_22uu:
508 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,3]
511 ; AVX512VL-LABEL: shuffle_v4f64_22uu:
513 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,3]
514 ; AVX512VL-NEXT: retq
515 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 undef, i32 undef>
516 ret <4 x double> %shuffle
519 define <4 x double> @shuffle_v4f64_3333(<4 x double> %a, <4 x double> %b) {
520 ; AVX1-LABEL: shuffle_v4f64_3333:
522 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
523 ; AVX1-NEXT: vmovhlps {{.*#+}} xmm0 = xmm0[1,1]
524 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
527 ; AVX2-LABEL: shuffle_v4f64_3333:
529 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,3]
532 ; AVX512VL-LABEL: shuffle_v4f64_3333:
534 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,3]
535 ; AVX512VL-NEXT: retq
536 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
537 ret <4 x double> %shuffle
540 define <4 x i64> @shuffle_v4i64_0000(<4 x i64> %a, <4 x i64> %b) {
541 ; AVX1-LABEL: shuffle_v4i64_0000:
543 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
544 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
547 ; AVX2-LABEL: shuffle_v4i64_0000:
549 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
552 ; AVX512VL-LABEL: shuffle_v4i64_0000:
554 ; AVX512VL-NEXT: vpbroadcastq %xmm0, %ymm0
555 ; AVX512VL-NEXT: retq
556 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
557 ret <4 x i64> %shuffle
560 define <4 x i64> @shuffle_v4i64_0001(<4 x i64> %a, <4 x i64> %b) {
561 ; AVX1-LABEL: shuffle_v4i64_0001:
563 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm0[0,0]
564 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
567 ; AVX2-LABEL: shuffle_v4i64_0001:
569 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
572 ; AVX512VL-LABEL: shuffle_v4i64_0001:
574 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
575 ; AVX512VL-NEXT: retq
576 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
577 ret <4 x i64> %shuffle
580 define <4 x i64> @shuffle_v4i64_0020(<4 x i64> %a, <4 x i64> %b) {
581 ; AVX1-LABEL: shuffle_v4i64_0020:
583 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
584 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
585 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
586 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
589 ; AVX2-LABEL: shuffle_v4i64_0020:
591 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
594 ; AVX512VL-LABEL: shuffle_v4i64_0020:
596 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
597 ; AVX512VL-NEXT: retq
598 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
599 ret <4 x i64> %shuffle
602 define <4 x i64> @shuffle_v4i64_0112(<4 x i64> %a, <4 x i64> %b) {
603 ; AVX1-LABEL: shuffle_v4i64_0112:
605 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
606 ; AVX1-NEXT: vshufpd {{.*#+}} xmm1 = xmm0[1],xmm1[0]
607 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
610 ; AVX2-LABEL: shuffle_v4i64_0112:
612 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
615 ; AVX512VL-LABEL: shuffle_v4i64_0112:
617 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
618 ; AVX512VL-NEXT: retq
619 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 1, i32 2>
620 ret <4 x i64> %shuffle
623 define <4 x i64> @shuffle_v4i64_0300(<4 x i64> %a, <4 x i64> %b) {
624 ; AVX1-LABEL: shuffle_v4i64_0300:
626 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
627 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
628 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
631 ; AVX2-LABEL: shuffle_v4i64_0300:
633 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
636 ; AVX512VL-LABEL: shuffle_v4i64_0300:
638 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
639 ; AVX512VL-NEXT: retq
640 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
641 ret <4 x i64> %shuffle
644 define <4 x i64> @shuffle_v4i64_1000(<4 x i64> %a, <4 x i64> %b) {
645 ; AVX1-LABEL: shuffle_v4i64_1000:
647 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
648 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
649 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
652 ; AVX2-LABEL: shuffle_v4i64_1000:
654 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
657 ; AVX512VL-LABEL: shuffle_v4i64_1000:
659 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
660 ; AVX512VL-NEXT: retq
661 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
662 ret <4 x i64> %shuffle
665 define <4 x i64> @shuffle_v4i64_2200(<4 x i64> %a, <4 x i64> %b) {
666 ; AVX1-LABEL: shuffle_v4i64_2200:
668 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
669 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
672 ; AVX2-LABEL: shuffle_v4i64_2200:
674 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
677 ; AVX512VL-LABEL: shuffle_v4i64_2200:
679 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
680 ; AVX512VL-NEXT: retq
681 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
682 ret <4 x i64> %shuffle
685 define <4 x i64> @shuffle_v4i64_3330(<4 x i64> %a, <4 x i64> %b) {
686 ; AVX1-LABEL: shuffle_v4i64_3330:
688 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
689 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3]
690 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
693 ; AVX2-LABEL: shuffle_v4i64_3330:
695 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
698 ; AVX512VL-LABEL: shuffle_v4i64_3330:
700 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
701 ; AVX512VL-NEXT: retq
702 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
703 ret <4 x i64> %shuffle
706 define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
707 ; AVX1-LABEL: shuffle_v4i64_3210:
709 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
710 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
713 ; AVX2-LABEL: shuffle_v4i64_3210:
715 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
718 ; AVX512VL-LABEL: shuffle_v4i64_3210:
720 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
721 ; AVX512VL-NEXT: retq
722 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
723 ret <4 x i64> %shuffle
726 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
727 ; AVX1-LABEL: shuffle_v4i64_0124:
729 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
730 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
731 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3]
734 ; AVX2-LABEL: shuffle_v4i64_0124:
736 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
737 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
740 ; AVX512VL-LABEL: shuffle_v4i64_0124:
742 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
743 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
744 ; AVX512VL-NEXT: retq
745 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
746 ret <4 x i64> %shuffle
749 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
750 ; AVX1-LABEL: shuffle_v4i64_0142:
752 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm1, %ymm1
753 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,1,2,2]
754 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
757 ; AVX2-LABEL: shuffle_v4i64_0142:
759 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
760 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
761 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
764 ; AVX512VL-LABEL: shuffle_v4i64_0142:
766 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm1, %ymm1
767 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
768 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
769 ; AVX512VL-NEXT: retq
770 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
771 ret <4 x i64> %shuffle
774 define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
775 ; AVX1-LABEL: shuffle_v4i64_0412:
777 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
778 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
779 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
780 ; AVX1-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
781 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
784 ; AVX2-LABEL: shuffle_v4i64_0412:
786 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
787 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
788 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
791 ; AVX512VL-LABEL: shuffle_v4i64_0412:
793 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
794 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
795 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
796 ; AVX512VL-NEXT: retq
797 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
798 ret <4 x i64> %shuffle
801 define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
802 ; AVX1-LABEL: shuffle_v4i64_4012:
804 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
805 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
806 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
807 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
808 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
811 ; AVX2-LABEL: shuffle_v4i64_4012:
813 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
814 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
817 ; AVX512VL-LABEL: shuffle_v4i64_4012:
819 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
820 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
821 ; AVX512VL-NEXT: retq
822 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
823 ret <4 x i64> %shuffle
826 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
827 ; AVX1-LABEL: shuffle_v4i64_0145:
829 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
832 ; AVX2-LABEL: shuffle_v4i64_0145:
834 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
837 ; AVX512VL-LABEL: shuffle_v4i64_0145:
839 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
840 ; AVX512VL-NEXT: retq
841 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
842 ret <4 x i64> %shuffle
845 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
846 ; AVX1-LABEL: shuffle_v4i64_0451:
848 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm1[1],xmm0[1]
849 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
850 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
853 ; AVX2-LABEL: shuffle_v4i64_0451:
855 ; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
856 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
857 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
860 ; AVX512VL-LABEL: shuffle_v4i64_0451:
862 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm0, %ymm0
863 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
864 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
865 ; AVX512VL-NEXT: retq
866 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
867 ret <4 x i64> %shuffle
870 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
871 ; AVX1-LABEL: shuffle_v4i64_4501:
873 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
876 ; AVX2-LABEL: shuffle_v4i64_4501:
878 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
881 ; AVX512VL-LABEL: shuffle_v4i64_4501:
883 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm1, %ymm0
884 ; AVX512VL-NEXT: retq
885 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
886 ret <4 x i64> %shuffle
889 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
890 ; AVX1-LABEL: shuffle_v4i64_4015:
892 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
893 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
894 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
897 ; AVX2-LABEL: shuffle_v4i64_4015:
899 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
900 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
901 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
904 ; AVX512VL-LABEL: shuffle_v4i64_4015:
906 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm1, %ymm1
907 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
908 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
909 ; AVX512VL-NEXT: retq
910 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
911 ret <4 x i64> %shuffle
914 define <4 x i64> @shuffle_v4i64_2u35(<4 x i64> %a, <4 x i64> %b) {
915 ; AVX1-LABEL: shuffle_v4i64_2u35:
917 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
918 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm1 = xmm0[1],xmm1[1]
919 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
922 ; AVX2-LABEL: shuffle_v4i64_2u35:
924 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
925 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
928 ; AVX512VL-LABEL: shuffle_v4i64_2u35:
930 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
931 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
932 ; AVX512VL-NEXT: retq
933 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 undef, i32 3, i32 5>
934 ret <4 x i64> %shuffle
937 define <4 x i64> @shuffle_v4i64_1251(<4 x i64> %a, <4 x i64> %b) {
938 ; AVX1-LABEL: shuffle_v4i64_1251:
940 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
941 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm2[0],ymm0[2],ymm2[3]
942 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
943 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
944 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
947 ; AVX2-LABEL: shuffle_v4i64_1251:
949 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
950 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
951 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
954 ; AVX512VL-LABEL: shuffle_v4i64_1251:
956 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
957 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
958 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
959 ; AVX512VL-NEXT: retq
960 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 5, i32 1>
961 ret <4 x i64> %shuffle
964 define <4 x i64> @shuffle_v4i64_1054(<4 x i64> %a, <4 x i64> %b) {
965 ; AVX1-LABEL: shuffle_v4i64_1054:
967 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
968 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
971 ; AVX2-LABEL: shuffle_v4i64_1054:
973 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
974 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
977 ; AVX512VL-LABEL: shuffle_v4i64_1054:
979 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
980 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
981 ; AVX512VL-NEXT: retq
982 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
983 ret <4 x i64> %shuffle
986 define <4 x i64> @shuffle_v4i64_3254(<4 x i64> %a, <4 x i64> %b) {
987 ; AVX1-LABEL: shuffle_v4i64_3254:
989 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
990 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
993 ; AVX2-LABEL: shuffle_v4i64_3254:
995 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
996 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
999 ; AVX512VL-LABEL: shuffle_v4i64_3254:
1001 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
1002 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1003 ; AVX512VL-NEXT: retq
1004 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
1005 ret <4 x i64> %shuffle
1008 define <4 x i64> @shuffle_v4i64_3276(<4 x i64> %a, <4 x i64> %b) {
1009 ; AVX1-LABEL: shuffle_v4i64_3276:
1011 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1012 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1015 ; AVX2-LABEL: shuffle_v4i64_3276:
1017 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1018 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1021 ; AVX512VL-LABEL: shuffle_v4i64_3276:
1023 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1024 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1025 ; AVX512VL-NEXT: retq
1026 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
1027 ret <4 x i64> %shuffle
1030 define <4 x i64> @shuffle_v4i64_1076(<4 x i64> %a, <4 x i64> %b) {
1031 ; AVX1-LABEL: shuffle_v4i64_1076:
1033 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1034 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1037 ; AVX2-LABEL: shuffle_v4i64_1076:
1039 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1040 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1043 ; AVX512VL-LABEL: shuffle_v4i64_1076:
1045 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1046 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1047 ; AVX512VL-NEXT: retq
1048 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
1049 ret <4 x i64> %shuffle
1052 define <4 x i64> @shuffle_v4i64_0415(<4 x i64> %a, <4 x i64> %b) {
1053 ; AVX1-LABEL: shuffle_v4i64_0415:
1055 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
1056 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
1057 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
1060 ; AVX2-LABEL: shuffle_v4i64_0415:
1062 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1063 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1064 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1067 ; AVX512VL-LABEL: shuffle_v4i64_0415:
1069 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1070 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1071 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1072 ; AVX512VL-NEXT: retq
1073 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
1074 ret <4 x i64> %shuffle
1077 define <4 x i64> @shuffle_v4i64_z4z6(<4 x i64> %a) {
1078 ; AVX1-LABEL: shuffle_v4i64_z4z6:
1080 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1081 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1084 ; AVX2-LABEL: shuffle_v4i64_z4z6:
1086 ; 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]
1089 ; AVX512VL-LABEL: shuffle_v4i64_z4z6:
1091 ; 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]
1092 ; AVX512VL-NEXT: retq
1093 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 0, i32 4, i32 0, i32 6>
1094 ret <4 x i64> %shuffle
1097 define <4 x i64> @shuffle_v4i64_5zuz(<4 x i64> %a) {
1098 ; AVX1-LABEL: shuffle_v4i64_5zuz:
1100 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1101 ; AVX1-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
1104 ; AVX2-LABEL: shuffle_v4i64_5zuz:
1106 ; 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
1109 ; AVX512VL-LABEL: shuffle_v4i64_5zuz:
1111 ; 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
1112 ; AVX512VL-NEXT: retq
1113 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 5, i32 0, i32 undef, i32 0>
1114 ret <4 x i64> %shuffle
1117 define <4 x i64> @shuffle_v4i64_40u2(<4 x i64> %a, <4 x i64> %b) {
1118 ; AVX1-LABEL: shuffle_v4i64_40u2:
1120 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1123 ; AVX2-LABEL: shuffle_v4i64_40u2:
1125 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1128 ; AVX512VL-LABEL: shuffle_v4i64_40u2:
1130 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1131 ; AVX512VL-NEXT: retq
1132 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 undef, i32 2>
1133 ret <4 x i64> %shuffle
1136 define <4 x i64> @shuffle_v4i64_11uu(<4 x i64> %a, <4 x i64> %b) {
1137 ; AVX1-LABEL: shuffle_v4i64_11uu:
1139 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,2,2]
1142 ; AVX2-LABEL: shuffle_v4i64_11uu:
1144 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
1147 ; AVX512VL-LABEL: shuffle_v4i64_11uu:
1149 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
1150 ; AVX512VL-NEXT: retq
1151 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 1, i32 undef, i32 undef>
1152 ret <4 x i64> %shuffle
1155 define <4 x i64> @shuffle_v4i64_22uu(<4 x i64> %a, <4 x i64> %b) {
1156 ; AVX1-LABEL: shuffle_v4i64_22uu:
1158 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
1159 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
1162 ; AVX2-LABEL: shuffle_v4i64_22uu:
1164 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,3]
1167 ; AVX512VL-LABEL: shuffle_v4i64_22uu:
1169 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,3]
1170 ; AVX512VL-NEXT: retq
1171 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 undef, i32 undef>
1172 ret <4 x i64> %shuffle
1175 define <4 x i64> @shuffle_v4i64_3333(<4 x i64> %a, <4 x i64> %b) {
1176 ; AVX1-LABEL: shuffle_v4i64_3333:
1178 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
1179 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,1]
1180 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1183 ; AVX2-LABEL: shuffle_v4i64_3333:
1185 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,3]
1188 ; AVX512VL-LABEL: shuffle_v4i64_3333:
1190 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,3]
1191 ; AVX512VL-NEXT: retq
1192 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
1193 ret <4 x i64> %shuffle
1196 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
1197 ; ALL-LABEL: stress_test1:
1199 %c = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 1, i32 1, i32 0>
1200 %d = shufflevector <4 x i64> %c, <4 x i64> undef, <4 x i32> <i32 3, i32 undef, i32 2, i32 undef>
1201 %e = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 undef>
1202 %f = shufflevector <4 x i64> %d, <4 x i64> %e, <4 x i32> <i32 5, i32 1, i32 1, i32 0>
1207 define <4 x i64> @insert_reg_and_zero_v4i64(i64 %a) {
1208 ; ALL-LABEL: insert_reg_and_zero_v4i64:
1210 ; ALL-NEXT: vmovq %rdi, %xmm0
1212 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1213 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1214 ret <4 x i64> %shuffle
1217 define <4 x i64> @insert_mem_and_zero_v4i64(i64* %ptr) {
1218 ; AVX1-LABEL: insert_mem_and_zero_v4i64:
1220 ; AVX1-NEXT: vmovq {{.*#+}} xmm0 = mem[0],zero
1223 ; AVX2-LABEL: insert_mem_and_zero_v4i64:
1225 ; AVX2-NEXT: vmovq {{.*#+}} xmm0 = mem[0],zero
1228 ; AVX512VL-LABEL: insert_mem_and_zero_v4i64:
1230 ; AVX512VL-NEXT: vmovq (%rdi), %xmm0
1231 ; AVX512VL-NEXT: retq
1232 %a = load i64, i64* %ptr
1233 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1234 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1235 ret <4 x i64> %shuffle
1238 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
1239 ; AVX1-LABEL: insert_reg_and_zero_v4f64:
1241 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1242 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1245 ; AVX2-LABEL: insert_reg_and_zero_v4f64:
1247 ; AVX2-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1248 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1251 ; AVX512VL-LABEL: insert_reg_and_zero_v4f64:
1253 ; AVX512VL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
1254 ; AVX512VL-NEXT: vmovsd %xmm0, %xmm1, %xmm0
1255 ; AVX512VL-NEXT: retq
1256 %v = insertelement <4 x double> undef, double %a, i32 0
1257 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1258 ret <4 x double> %shuffle
1261 define <4 x double> @insert_mem_and_zero_v4f64(double* %ptr) {
1262 ; AVX1-LABEL: insert_mem_and_zero_v4f64:
1264 ; AVX1-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1267 ; AVX2-LABEL: insert_mem_and_zero_v4f64:
1269 ; AVX2-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1272 ; AVX512VL-LABEL: insert_mem_and_zero_v4f64:
1274 ; AVX512VL-NEXT: vmovsd (%rdi), %xmm0
1275 ; AVX512VL-NEXT: retq
1276 %a = load double, double* %ptr
1277 %v = insertelement <4 x double> undef, double %a, i32 0
1278 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1279 ret <4 x double> %shuffle
1282 define <4 x double> @splat_mem_v4f64(double* %ptr) {
1283 ; ALL-LABEL: splat_mem_v4f64:
1285 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1287 %a = load double, double* %ptr
1288 %v = insertelement <4 x double> undef, double %a, i32 0
1289 %shuffle = shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1290 ret <4 x double> %shuffle
1293 define <4 x i64> @splat_mem_v4i64(i64* %ptr) {
1294 ; AVX1-LABEL: splat_mem_v4i64:
1296 ; AVX1-NEXT: vbroadcastsd (%rdi), %ymm0
1299 ; AVX2-LABEL: splat_mem_v4i64:
1301 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1304 ; AVX512VL-LABEL: splat_mem_v4i64:
1306 ; AVX512VL-NEXT: vpbroadcastq (%rdi), %ymm0
1307 ; AVX512VL-NEXT: retq
1308 %a = load i64, i64* %ptr
1309 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1310 %shuffle = shufflevector <4 x i64> %v, <4 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1311 ret <4 x i64> %shuffle
1314 define <4 x double> @splat_mem_v4f64_2(double* %p) {
1315 ; ALL-LABEL: splat_mem_v4f64_2:
1317 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1319 %1 = load double, double* %p
1320 %2 = insertelement <2 x double> undef, double %1, i32 0
1321 %3 = shufflevector <2 x double> %2, <2 x double> undef, <4 x i32> zeroinitializer
1325 define <4 x double> @splat_v4f64(<2 x double> %r) {
1326 ; AVX1-LABEL: splat_v4f64:
1328 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
1329 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1332 ; AVX2-LABEL: splat_v4f64:
1334 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
1337 ; AVX512VL-LABEL: splat_v4f64:
1339 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
1340 ; AVX512VL-NEXT: retq
1341 %1 = shufflevector <2 x double> %r, <2 x double> undef, <4 x i32> zeroinitializer
1345 define <4 x i64> @splat_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1346 ; AVX1-LABEL: splat_mem_v4i64_from_v2i64:
1348 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
1349 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1352 ; AVX2-LABEL: splat_mem_v4i64_from_v2i64:
1354 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1357 ; AVX512VL-LABEL: splat_mem_v4i64_from_v2i64:
1359 ; AVX512VL-NEXT: vmovdqa64 (%rdi), %xmm0
1360 ; AVX512VL-NEXT: vpbroadcastq %xmm0, %ymm0
1361 ; AVX512VL-NEXT: retq
1362 %v = load <2 x i64>, <2 x i64>* %ptr
1363 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1364 ret <4 x i64> %shuffle
1367 define <4 x double> @splat_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1368 ; AVX1-LABEL: splat_mem_v4f64_from_v2f64:
1370 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
1371 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1374 ; AVX2-LABEL: splat_mem_v4f64_from_v2f64:
1376 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1379 ; AVX512VL-LABEL: splat_mem_v4f64_from_v2f64:
1381 ; AVX512VL-NEXT: vbroadcastsd (%rdi), %ymm0
1382 ; AVX512VL-NEXT: retq
1383 %v = load <2 x double>, <2 x double>* %ptr
1384 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1385 ret <4 x double> %shuffle
1388 define <4 x i64> @splat128_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1389 ; AVX1-LABEL: splat128_mem_v4i64_from_v2i64:
1391 ; AVX1-NEXT: vmovaps (%rdi), %xmm0
1392 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1395 ; AVX2-LABEL: splat128_mem_v4i64_from_v2i64:
1397 ; AVX2-NEXT: vmovaps (%rdi), %xmm0
1398 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1401 ; AVX512VL-LABEL: splat128_mem_v4i64_from_v2i64:
1403 ; AVX512VL-NEXT: vmovdqa64 (%rdi), %xmm0
1404 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm0, %ymm0
1405 ; AVX512VL-NEXT: retq
1406 %v = load <2 x i64>, <2 x i64>* %ptr
1407 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1408 ret <4 x i64> %shuffle
1411 define <4 x double> @splat128_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1412 ; AVX1-LABEL: splat128_mem_v4f64_from_v2f64:
1414 ; AVX1-NEXT: vmovaps (%rdi), %xmm0
1415 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1418 ; AVX2-LABEL: splat128_mem_v4f64_from_v2f64:
1420 ; AVX2-NEXT: vmovaps (%rdi), %xmm0
1421 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1424 ; AVX512VL-LABEL: splat128_mem_v4f64_from_v2f64:
1426 ; AVX512VL-NEXT: vmovapd (%rdi), %xmm0
1427 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm0, %ymm0
1428 ; AVX512VL-NEXT: retq
1429 %v = load <2 x double>, <2 x double>* %ptr
1430 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1431 ret <4 x double> %shuffle
1434 define <4 x double> @bitcast_v4f64_0426(<4 x double> %a, <4 x double> %b) {
1435 ; AVX1-LABEL: bitcast_v4f64_0426:
1437 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1440 ; AVX2-LABEL: bitcast_v4f64_0426:
1442 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1445 ; AVX512VL-LABEL: bitcast_v4f64_0426:
1447 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1448 ; AVX512VL-NEXT: retq
1449 %shuffle64 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
1450 %bitcast32 = bitcast <4 x double> %shuffle64 to <8 x float>
1451 %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>
1452 %bitcast16 = bitcast <8 x float> %shuffle32 to <16 x i16>
1453 %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>
1454 %bitcast64 = bitcast <16 x i16> %shuffle16 to <4 x double>
1455 ret <4 x double> %bitcast64
1458 define <4 x i64> @concat_v4i64_0167(<4 x i64> %a0, <4 x i64> %a1) {
1459 ; AVX1-LABEL: concat_v4i64_0167:
1461 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1464 ; AVX2-LABEL: concat_v4i64_0167:
1466 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1469 ; AVX512VL-LABEL: concat_v4i64_0167:
1471 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1472 ; AVX512VL-NEXT: retq
1473 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1474 %a1hi = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 6, i32 7>
1475 %shuffle64 = shufflevector <2 x i64> %a0lo, <2 x i64> %a1hi, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1476 ret <4 x i64> %shuffle64
1479 define <4 x i64> @concat_v4i64_0145_bc(<4 x i64> %a0, <4 x i64> %a1) {
1480 ; AVX1-LABEL: concat_v4i64_0145_bc:
1482 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1485 ; AVX2-LABEL: concat_v4i64_0145_bc:
1487 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1490 ; AVX512VL-LABEL: concat_v4i64_0145_bc:
1492 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
1493 ; AVX512VL-NEXT: retq
1494 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1495 %a1lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 4, i32 5>
1496 %bc0lo = bitcast <2 x i64> %a0lo to <4 x i32>
1497 %bc1lo = bitcast <2 x i64> %a1lo to <4 x i32>
1498 %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>
1499 %shuffle64 = bitcast <8 x i32> %shuffle32 to <4 x i64>
1500 ret <4 x i64> %shuffle64
1503 define <4 x i64> @insert_dup_mem_v4i64(i64* %ptr) {
1504 ; AVX1-LABEL: insert_dup_mem_v4i64:
1506 ; AVX1-NEXT: vbroadcastsd (%rdi), %ymm0
1509 ; AVX2-LABEL: insert_dup_mem_v4i64:
1511 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1514 ; AVX512VL-LABEL: insert_dup_mem_v4i64:
1516 ; AVX512VL-NEXT: vpbroadcastq (%rdi), %ymm0
1517 ; AVX512VL-NEXT: retq
1518 %tmp = load i64, i64* %ptr, align 1
1519 %tmp1 = insertelement <2 x i64> undef, i64 %tmp, i32 0
1520 %tmp2 = shufflevector <2 x i64> %tmp1, <2 x i64> undef, <4 x i32> zeroinitializer