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_v4f64_1032(<4 x double> %a, <4 x double> %b) {
192 ; ALL-LABEL: shuffle_v4f64_1032:
194 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
196 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
197 ret <4 x double> %shuffle
200 define <4 x double> @shuffle_v4f64_1133(<4 x double> %a, <4 x double> %b) {
201 ; ALL-LABEL: shuffle_v4f64_1133:
203 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
205 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
206 ret <4 x double> %shuffle
209 define <4 x double> @shuffle_v4f64_1023(<4 x double> %a, <4 x double> %b) {
210 ; ALL-LABEL: shuffle_v4f64_1023:
212 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,3]
214 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 3>
215 ret <4 x double> %shuffle
218 define <4 x double> @shuffle_v4f64_1022(<4 x double> %a, <4 x double> %b) {
219 ; ALL-LABEL: shuffle_v4f64_1022:
221 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
223 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 2>
224 ret <4 x double> %shuffle
227 define <4 x double> @shuffle_v4f64_0423(<4 x double> %a, <4 x double> %b) {
228 ; AVX1-LABEL: shuffle_v4f64_0423:
230 ; AVX1-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
231 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
234 ; AVX2-LABEL: shuffle_v4f64_0423:
236 ; AVX2-NEXT: vbroadcastsd %xmm1, %ymm1
237 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
240 ; AVX512VL-LABEL: shuffle_v4f64_0423:
242 ; AVX512VL-NEXT: vbroadcastsd %xmm1, %ymm1
243 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
244 ; AVX512VL-NEXT: retq
245 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 3>
246 ret <4 x double> %shuffle
249 define <4 x double> @shuffle_v4f64_0462(<4 x double> %a, <4 x double> %b) {
250 ; ALL-LABEL: shuffle_v4f64_0462:
252 ; ALL-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
253 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
254 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
256 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 6, i32 2>
257 ret <4 x double> %shuffle
260 define <4 x double> @shuffle_v4f64_0426(<4 x double> %a, <4 x double> %b) {
261 ; ALL-LABEL: shuffle_v4f64_0426:
263 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
265 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
266 ret <4 x double> %shuffle
269 define <4 x double> @shuffle_v4f64_1537(<4 x double> %a, <4 x double> %b) {
270 ; ALL-LABEL: shuffle_v4f64_1537:
272 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
274 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
275 ret <4 x double> %shuffle
278 define <4 x double> @shuffle_v4f64_4062(<4 x double> %a, <4 x double> %b) {
279 ; ALL-LABEL: shuffle_v4f64_4062:
281 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
283 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
284 ret <4 x double> %shuffle
287 define <4 x double> @shuffle_v4f64_5173(<4 x double> %a, <4 x double> %b) {
288 ; ALL-LABEL: shuffle_v4f64_5173:
290 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
292 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 7, i32 3>
293 ret <4 x double> %shuffle
296 define <4 x double> @shuffle_v4f64_5163(<4 x double> %a, <4 x double> %b) {
297 ; ALL-LABEL: shuffle_v4f64_5163:
299 ; ALL-NEXT: vshufpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
301 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
302 ret <4 x double> %shuffle
305 define <4 x double> @shuffle_v4f64_0527(<4 x double> %a, <4 x double> %b) {
306 ; ALL-LABEL: shuffle_v4f64_0527:
308 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
310 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
311 ret <4 x double> %shuffle
314 define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
315 ; ALL-LABEL: shuffle_v4f64_4163:
317 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
319 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
320 ret <4 x double> %shuffle
323 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
324 ; AVX1-LABEL: shuffle_v4f64_0145:
326 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
329 ; AVX2-LABEL: shuffle_v4f64_0145:
331 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
334 ; AVX512VL-LABEL: shuffle_v4f64_0145:
336 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
337 ; AVX512VL-NEXT: retq
338 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
339 ret <4 x double> %shuffle
342 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
343 ; AVX1-LABEL: shuffle_v4f64_4501:
345 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
348 ; AVX2-LABEL: shuffle_v4f64_4501:
350 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
353 ; AVX512VL-LABEL: shuffle_v4f64_4501:
355 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm1, %ymm0
356 ; AVX512VL-NEXT: retq
357 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
358 ret <4 x double> %shuffle
361 define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
362 ; ALL-LABEL: shuffle_v4f64_0167:
364 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
366 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
367 ret <4 x double> %shuffle
370 define <4 x double> @shuffle_v4f64_1054(<4 x double> %a, <4 x double> %b) {
371 ; AVX1-LABEL: shuffle_v4f64_1054:
373 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
374 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
377 ; AVX2-LABEL: shuffle_v4f64_1054:
379 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
380 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
383 ; AVX512VL-LABEL: shuffle_v4f64_1054:
385 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
386 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
387 ; AVX512VL-NEXT: retq
388 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
389 ret <4 x double> %shuffle
392 define <4 x double> @shuffle_v4f64_3254(<4 x double> %a, <4 x double> %b) {
393 ; AVX1-LABEL: shuffle_v4f64_3254:
395 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
396 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
399 ; AVX2-LABEL: shuffle_v4f64_3254:
401 ; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
402 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
405 ; AVX512VL-LABEL: shuffle_v4f64_3254:
407 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
408 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
409 ; AVX512VL-NEXT: retq
410 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
411 ret <4 x double> %shuffle
414 define <4 x double> @shuffle_v4f64_3276(<4 x double> %a, <4 x double> %b) {
415 ; AVX1-LABEL: shuffle_v4f64_3276:
417 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
418 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
421 ; AVX2-LABEL: shuffle_v4f64_3276:
423 ; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
424 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
427 ; AVX512VL-LABEL: shuffle_v4f64_3276:
429 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
430 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
431 ; AVX512VL-NEXT: retq
432 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
433 ret <4 x double> %shuffle
436 define <4 x double> @shuffle_v4f64_1076(<4 x double> %a, <4 x double> %b) {
437 ; ALL-LABEL: shuffle_v4f64_1076:
439 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
440 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
442 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
443 ret <4 x double> %shuffle
446 define <4 x double> @shuffle_v4f64_0415(<4 x double> %a, <4 x double> %b) {
447 ; AVX1-LABEL: shuffle_v4f64_0415:
449 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
450 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
451 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
454 ; AVX2-LABEL: shuffle_v4f64_0415:
456 ; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
457 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
458 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
461 ; AVX512VL-LABEL: shuffle_v4f64_0415:
463 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
464 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
465 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
466 ; AVX512VL-NEXT: retq
467 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
468 ret <4 x double> %shuffle
471 define <4 x double> @shuffle_v4f64_u062(<4 x double> %a, <4 x double> %b) {
472 ; ALL-LABEL: shuffle_v4f64_u062:
474 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
476 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 undef, i32 0, i32 6, i32 2>
477 ret <4 x double> %shuffle
480 define <4 x double> @shuffle_v4f64_11uu(<4 x double> %a, <4 x double> %b) {
481 ; ALL-LABEL: shuffle_v4f64_11uu:
483 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,2,2]
485 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 undef, i32 undef>
486 ret <4 x double> %shuffle
489 define <4 x double> @shuffle_v4f64_22uu(<4 x double> %a, <4 x double> %b) {
490 ; AVX1-LABEL: shuffle_v4f64_22uu:
492 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
493 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
496 ; AVX2-LABEL: shuffle_v4f64_22uu:
498 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,3]
501 ; AVX512VL-LABEL: shuffle_v4f64_22uu:
503 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,3]
504 ; AVX512VL-NEXT: retq
505 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 undef, i32 undef>
506 ret <4 x double> %shuffle
509 define <4 x double> @shuffle_v4f64_3333(<4 x double> %a, <4 x double> %b) {
510 ; AVX1-LABEL: shuffle_v4f64_3333:
512 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
513 ; AVX1-NEXT: vmovhlps {{.*#+}} xmm0 = xmm0[1,1]
514 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
517 ; AVX2-LABEL: shuffle_v4f64_3333:
519 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,3]
522 ; AVX512VL-LABEL: shuffle_v4f64_3333:
524 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,3]
525 ; AVX512VL-NEXT: retq
526 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
527 ret <4 x double> %shuffle
530 define <4 x i64> @shuffle_v4i64_0000(<4 x i64> %a, <4 x i64> %b) {
531 ; AVX1-LABEL: shuffle_v4i64_0000:
533 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
534 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
537 ; AVX2-LABEL: shuffle_v4i64_0000:
539 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
542 ; AVX512VL-LABEL: shuffle_v4i64_0000:
544 ; AVX512VL-NEXT: vpbroadcastq %xmm0, %ymm0
545 ; AVX512VL-NEXT: retq
546 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
547 ret <4 x i64> %shuffle
550 define <4 x i64> @shuffle_v4i64_0001(<4 x i64> %a, <4 x i64> %b) {
551 ; AVX1-LABEL: shuffle_v4i64_0001:
553 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm0[0,0]
554 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
557 ; AVX2-LABEL: shuffle_v4i64_0001:
559 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
562 ; AVX512VL-LABEL: shuffle_v4i64_0001:
564 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
565 ; AVX512VL-NEXT: retq
566 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
567 ret <4 x i64> %shuffle
570 define <4 x i64> @shuffle_v4i64_0020(<4 x i64> %a, <4 x i64> %b) {
571 ; AVX1-LABEL: shuffle_v4i64_0020:
573 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
574 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
575 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
576 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
579 ; AVX2-LABEL: shuffle_v4i64_0020:
581 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
584 ; AVX512VL-LABEL: shuffle_v4i64_0020:
586 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
587 ; AVX512VL-NEXT: retq
588 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
589 ret <4 x i64> %shuffle
592 define <4 x i64> @shuffle_v4i64_0112(<4 x i64> %a, <4 x i64> %b) {
593 ; AVX1-LABEL: shuffle_v4i64_0112:
595 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
596 ; AVX1-NEXT: vshufpd {{.*#+}} xmm1 = xmm0[1],xmm1[0]
597 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
600 ; AVX2-LABEL: shuffle_v4i64_0112:
602 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
605 ; AVX512VL-LABEL: shuffle_v4i64_0112:
607 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
608 ; AVX512VL-NEXT: retq
609 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 1, i32 2>
610 ret <4 x i64> %shuffle
613 define <4 x i64> @shuffle_v4i64_0300(<4 x i64> %a, <4 x i64> %b) {
614 ; AVX1-LABEL: shuffle_v4i64_0300:
616 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
617 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
618 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
621 ; AVX2-LABEL: shuffle_v4i64_0300:
623 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
626 ; AVX512VL-LABEL: shuffle_v4i64_0300:
628 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
629 ; AVX512VL-NEXT: retq
630 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
631 ret <4 x i64> %shuffle
634 define <4 x i64> @shuffle_v4i64_1000(<4 x i64> %a, <4 x i64> %b) {
635 ; AVX1-LABEL: shuffle_v4i64_1000:
637 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
638 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
639 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
642 ; AVX2-LABEL: shuffle_v4i64_1000:
644 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
647 ; AVX512VL-LABEL: shuffle_v4i64_1000:
649 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
650 ; AVX512VL-NEXT: retq
651 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
652 ret <4 x i64> %shuffle
655 define <4 x i64> @shuffle_v4i64_2200(<4 x i64> %a, <4 x i64> %b) {
656 ; AVX1-LABEL: shuffle_v4i64_2200:
658 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
659 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
662 ; AVX2-LABEL: shuffle_v4i64_2200:
664 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
667 ; AVX512VL-LABEL: shuffle_v4i64_2200:
669 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
670 ; AVX512VL-NEXT: retq
671 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
672 ret <4 x i64> %shuffle
675 define <4 x i64> @shuffle_v4i64_3330(<4 x i64> %a, <4 x i64> %b) {
676 ; AVX1-LABEL: shuffle_v4i64_3330:
678 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
679 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3]
680 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
683 ; AVX2-LABEL: shuffle_v4i64_3330:
685 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
688 ; AVX512VL-LABEL: shuffle_v4i64_3330:
690 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
691 ; AVX512VL-NEXT: retq
692 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
693 ret <4 x i64> %shuffle
696 define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
697 ; AVX1-LABEL: shuffle_v4i64_3210:
699 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
700 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
703 ; AVX2-LABEL: shuffle_v4i64_3210:
705 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
708 ; AVX512VL-LABEL: shuffle_v4i64_3210:
710 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
711 ; AVX512VL-NEXT: retq
712 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
713 ret <4 x i64> %shuffle
716 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
717 ; AVX1-LABEL: shuffle_v4i64_0124:
719 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
720 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
721 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3]
724 ; AVX2-LABEL: shuffle_v4i64_0124:
726 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
727 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
730 ; AVX512VL-LABEL: shuffle_v4i64_0124:
732 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
733 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
734 ; AVX512VL-NEXT: retq
735 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
736 ret <4 x i64> %shuffle
739 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
740 ; AVX1-LABEL: shuffle_v4i64_0142:
742 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm1, %ymm1
743 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,1,2,2]
744 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
747 ; AVX2-LABEL: shuffle_v4i64_0142:
749 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
750 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
751 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
754 ; AVX512VL-LABEL: shuffle_v4i64_0142:
756 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm1, %ymm1
757 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
758 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
759 ; AVX512VL-NEXT: retq
760 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
761 ret <4 x i64> %shuffle
764 define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
765 ; AVX1-LABEL: shuffle_v4i64_0412:
767 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
768 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
769 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
770 ; AVX1-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
771 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
774 ; AVX2-LABEL: shuffle_v4i64_0412:
776 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
777 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
778 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
781 ; AVX512VL-LABEL: shuffle_v4i64_0412:
783 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
784 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
785 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
786 ; AVX512VL-NEXT: retq
787 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
788 ret <4 x i64> %shuffle
791 define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
792 ; AVX1-LABEL: shuffle_v4i64_4012:
794 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
795 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
796 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
797 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
798 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
801 ; AVX2-LABEL: shuffle_v4i64_4012:
803 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
804 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
807 ; AVX512VL-LABEL: shuffle_v4i64_4012:
809 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
810 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
811 ; AVX512VL-NEXT: retq
812 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
813 ret <4 x i64> %shuffle
816 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
817 ; AVX1-LABEL: shuffle_v4i64_0145:
819 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
822 ; AVX2-LABEL: shuffle_v4i64_0145:
824 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
827 ; AVX512VL-LABEL: shuffle_v4i64_0145:
829 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
830 ; AVX512VL-NEXT: retq
831 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
832 ret <4 x i64> %shuffle
835 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
836 ; AVX1-LABEL: shuffle_v4i64_0451:
838 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm1[1],xmm0[1]
839 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
840 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
843 ; AVX2-LABEL: shuffle_v4i64_0451:
845 ; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
846 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
847 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
850 ; AVX512VL-LABEL: shuffle_v4i64_0451:
852 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm0, %ymm0
853 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
854 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
855 ; AVX512VL-NEXT: retq
856 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
857 ret <4 x i64> %shuffle
860 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
861 ; AVX1-LABEL: shuffle_v4i64_4501:
863 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
866 ; AVX2-LABEL: shuffle_v4i64_4501:
868 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
871 ; AVX512VL-LABEL: shuffle_v4i64_4501:
873 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm1, %ymm0
874 ; AVX512VL-NEXT: retq
875 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
876 ret <4 x i64> %shuffle
879 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
880 ; AVX1-LABEL: shuffle_v4i64_4015:
882 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
883 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
884 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
887 ; AVX2-LABEL: shuffle_v4i64_4015:
889 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
890 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
891 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
894 ; AVX512VL-LABEL: shuffle_v4i64_4015:
896 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm1, %ymm1
897 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
898 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
899 ; AVX512VL-NEXT: retq
900 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
901 ret <4 x i64> %shuffle
904 define <4 x i64> @shuffle_v4i64_2u35(<4 x i64> %a, <4 x i64> %b) {
905 ; AVX1-LABEL: shuffle_v4i64_2u35:
907 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
908 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm1 = xmm0[1],xmm1[1]
909 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
912 ; AVX2-LABEL: shuffle_v4i64_2u35:
914 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
915 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
918 ; AVX512VL-LABEL: shuffle_v4i64_2u35:
920 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
921 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
922 ; AVX512VL-NEXT: retq
923 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 undef, i32 3, i32 5>
924 ret <4 x i64> %shuffle
927 define <4 x i64> @shuffle_v4i64_1251(<4 x i64> %a, <4 x i64> %b) {
928 ; AVX1-LABEL: shuffle_v4i64_1251:
930 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
931 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm2[0],ymm0[2],ymm2[3]
932 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
933 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
934 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
937 ; AVX2-LABEL: shuffle_v4i64_1251:
939 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
940 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
941 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
944 ; AVX512VL-LABEL: shuffle_v4i64_1251:
946 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
947 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
948 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
949 ; AVX512VL-NEXT: retq
950 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 5, i32 1>
951 ret <4 x i64> %shuffle
954 define <4 x i64> @shuffle_v4i64_1054(<4 x i64> %a, <4 x i64> %b) {
955 ; AVX1-LABEL: shuffle_v4i64_1054:
957 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
958 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
961 ; AVX2-LABEL: shuffle_v4i64_1054:
963 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
964 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
967 ; AVX512VL-LABEL: shuffle_v4i64_1054:
969 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
970 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
971 ; AVX512VL-NEXT: retq
972 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
973 ret <4 x i64> %shuffle
976 define <4 x i64> @shuffle_v4i64_3254(<4 x i64> %a, <4 x i64> %b) {
977 ; AVX1-LABEL: shuffle_v4i64_3254:
979 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
980 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
983 ; AVX2-LABEL: shuffle_v4i64_3254:
985 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
986 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
989 ; AVX512VL-LABEL: shuffle_v4i64_3254:
991 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
992 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
993 ; AVX512VL-NEXT: retq
994 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
995 ret <4 x i64> %shuffle
998 define <4 x i64> @shuffle_v4i64_3276(<4 x i64> %a, <4 x i64> %b) {
999 ; AVX1-LABEL: shuffle_v4i64_3276:
1001 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1002 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1005 ; AVX2-LABEL: shuffle_v4i64_3276:
1007 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1008 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1011 ; AVX512VL-LABEL: shuffle_v4i64_3276:
1013 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1014 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1015 ; AVX512VL-NEXT: retq
1016 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
1017 ret <4 x i64> %shuffle
1020 define <4 x i64> @shuffle_v4i64_1076(<4 x i64> %a, <4 x i64> %b) {
1021 ; AVX1-LABEL: shuffle_v4i64_1076:
1023 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1024 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1027 ; AVX2-LABEL: shuffle_v4i64_1076:
1029 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1030 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1033 ; AVX512VL-LABEL: shuffle_v4i64_1076:
1035 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1036 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1037 ; AVX512VL-NEXT: retq
1038 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
1039 ret <4 x i64> %shuffle
1042 define <4 x i64> @shuffle_v4i64_0415(<4 x i64> %a, <4 x i64> %b) {
1043 ; AVX1-LABEL: shuffle_v4i64_0415:
1045 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
1046 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
1047 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
1050 ; AVX2-LABEL: shuffle_v4i64_0415:
1052 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1053 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1054 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1057 ; AVX512VL-LABEL: shuffle_v4i64_0415:
1059 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1060 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1061 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1062 ; AVX512VL-NEXT: retq
1063 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
1064 ret <4 x i64> %shuffle
1067 define <4 x i64> @shuffle_v4i64_z4z6(<4 x i64> %a) {
1068 ; AVX1-LABEL: shuffle_v4i64_z4z6:
1070 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1071 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1074 ; AVX2-LABEL: shuffle_v4i64_z4z6:
1076 ; 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]
1079 ; AVX512VL-LABEL: shuffle_v4i64_z4z6:
1081 ; 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]
1082 ; AVX512VL-NEXT: retq
1083 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 0, i32 4, i32 0, i32 6>
1084 ret <4 x i64> %shuffle
1087 define <4 x i64> @shuffle_v4i64_5zuz(<4 x i64> %a) {
1088 ; AVX1-LABEL: shuffle_v4i64_5zuz:
1090 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1091 ; AVX1-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
1094 ; AVX2-LABEL: shuffle_v4i64_5zuz:
1096 ; 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
1099 ; AVX512VL-LABEL: shuffle_v4i64_5zuz:
1101 ; 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
1102 ; AVX512VL-NEXT: retq
1103 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 5, i32 0, i32 undef, i32 0>
1104 ret <4 x i64> %shuffle
1107 define <4 x i64> @shuffle_v4i64_40u2(<4 x i64> %a, <4 x i64> %b) {
1108 ; AVX1-LABEL: shuffle_v4i64_40u2:
1110 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1113 ; AVX2-LABEL: shuffle_v4i64_40u2:
1115 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1118 ; AVX512VL-LABEL: shuffle_v4i64_40u2:
1120 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1121 ; AVX512VL-NEXT: retq
1122 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 undef, i32 2>
1123 ret <4 x i64> %shuffle
1126 define <4 x i64> @shuffle_v4i64_11uu(<4 x i64> %a, <4 x i64> %b) {
1127 ; AVX1-LABEL: shuffle_v4i64_11uu:
1129 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,2,2]
1132 ; AVX2-LABEL: shuffle_v4i64_11uu:
1134 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
1137 ; AVX512VL-LABEL: shuffle_v4i64_11uu:
1139 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
1140 ; AVX512VL-NEXT: retq
1141 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 1, i32 undef, i32 undef>
1142 ret <4 x i64> %shuffle
1145 define <4 x i64> @shuffle_v4i64_22uu(<4 x i64> %a, <4 x i64> %b) {
1146 ; AVX1-LABEL: shuffle_v4i64_22uu:
1148 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
1149 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
1152 ; AVX2-LABEL: shuffle_v4i64_22uu:
1154 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,3]
1157 ; AVX512VL-LABEL: shuffle_v4i64_22uu:
1159 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,3]
1160 ; AVX512VL-NEXT: retq
1161 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 undef, i32 undef>
1162 ret <4 x i64> %shuffle
1165 define <4 x i64> @shuffle_v4i64_3333(<4 x i64> %a, <4 x i64> %b) {
1166 ; AVX1-LABEL: shuffle_v4i64_3333:
1168 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
1169 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,1]
1170 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1173 ; AVX2-LABEL: shuffle_v4i64_3333:
1175 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,3]
1178 ; AVX512VL-LABEL: shuffle_v4i64_3333:
1180 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,3]
1181 ; AVX512VL-NEXT: retq
1182 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
1183 ret <4 x i64> %shuffle
1186 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
1187 ; ALL-LABEL: stress_test1:
1189 %c = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 1, i32 1, i32 0>
1190 %d = shufflevector <4 x i64> %c, <4 x i64> undef, <4 x i32> <i32 3, i32 undef, i32 2, i32 undef>
1191 %e = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 undef>
1192 %f = shufflevector <4 x i64> %d, <4 x i64> %e, <4 x i32> <i32 5, i32 1, i32 1, i32 0>
1197 define <4 x i64> @insert_reg_and_zero_v4i64(i64 %a) {
1198 ; ALL-LABEL: insert_reg_and_zero_v4i64:
1200 ; ALL-NEXT: vmovq %rdi, %xmm0
1202 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1203 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1204 ret <4 x i64> %shuffle
1207 define <4 x i64> @insert_mem_and_zero_v4i64(i64* %ptr) {
1208 ; AVX1-LABEL: insert_mem_and_zero_v4i64:
1210 ; AVX1-NEXT: vmovq {{.*#+}} xmm0 = mem[0],zero
1213 ; AVX2-LABEL: insert_mem_and_zero_v4i64:
1215 ; AVX2-NEXT: vmovq {{.*#+}} xmm0 = mem[0],zero
1218 ; AVX512VL-LABEL: insert_mem_and_zero_v4i64:
1220 ; AVX512VL-NEXT: vmovq (%rdi), %xmm0
1221 ; AVX512VL-NEXT: retq
1222 %a = load i64, i64* %ptr
1223 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1224 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1225 ret <4 x i64> %shuffle
1228 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
1229 ; AVX1-LABEL: insert_reg_and_zero_v4f64:
1231 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1232 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1235 ; AVX2-LABEL: insert_reg_and_zero_v4f64:
1237 ; AVX2-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1238 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1241 ; AVX512VL-LABEL: insert_reg_and_zero_v4f64:
1243 ; AVX512VL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
1244 ; AVX512VL-NEXT: vmovsd %xmm0, %xmm1, %xmm0
1245 ; AVX512VL-NEXT: retq
1246 %v = insertelement <4 x double> undef, double %a, i32 0
1247 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1248 ret <4 x double> %shuffle
1251 define <4 x double> @insert_mem_and_zero_v4f64(double* %ptr) {
1252 ; AVX1-LABEL: insert_mem_and_zero_v4f64:
1254 ; AVX1-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1257 ; AVX2-LABEL: insert_mem_and_zero_v4f64:
1259 ; AVX2-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1262 ; AVX512VL-LABEL: insert_mem_and_zero_v4f64:
1264 ; AVX512VL-NEXT: vmovsd (%rdi), %xmm0
1265 ; AVX512VL-NEXT: retq
1266 %a = load double, double* %ptr
1267 %v = insertelement <4 x double> undef, double %a, i32 0
1268 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1269 ret <4 x double> %shuffle
1272 define <4 x double> @splat_mem_v4f64(double* %ptr) {
1273 ; ALL-LABEL: splat_mem_v4f64:
1275 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1277 %a = load double, double* %ptr
1278 %v = insertelement <4 x double> undef, double %a, i32 0
1279 %shuffle = shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1280 ret <4 x double> %shuffle
1283 define <4 x i64> @splat_mem_v4i64(i64* %ptr) {
1284 ; AVX1-LABEL: splat_mem_v4i64:
1286 ; AVX1-NEXT: vbroadcastsd (%rdi), %ymm0
1289 ; AVX2-LABEL: splat_mem_v4i64:
1291 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1294 ; AVX512VL-LABEL: splat_mem_v4i64:
1296 ; AVX512VL-NEXT: vpbroadcastq (%rdi), %ymm0
1297 ; AVX512VL-NEXT: retq
1298 %a = load i64, i64* %ptr
1299 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1300 %shuffle = shufflevector <4 x i64> %v, <4 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1301 ret <4 x i64> %shuffle
1304 define <4 x double> @splat_mem_v4f64_2(double* %p) {
1305 ; ALL-LABEL: splat_mem_v4f64_2:
1307 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1309 %1 = load double, double* %p
1310 %2 = insertelement <2 x double> undef, double %1, i32 0
1311 %3 = shufflevector <2 x double> %2, <2 x double> undef, <4 x i32> zeroinitializer
1315 define <4 x double> @splat_v4f64(<2 x double> %r) {
1316 ; AVX1-LABEL: splat_v4f64:
1318 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
1319 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1322 ; AVX2-LABEL: splat_v4f64:
1324 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
1327 ; AVX512VL-LABEL: splat_v4f64:
1329 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
1330 ; AVX512VL-NEXT: retq
1331 %1 = shufflevector <2 x double> %r, <2 x double> undef, <4 x i32> zeroinitializer
1335 define <4 x i64> @splat_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1336 ; AVX1-LABEL: splat_mem_v4i64_from_v2i64:
1338 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
1339 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1342 ; AVX2-LABEL: splat_mem_v4i64_from_v2i64:
1344 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1347 ; AVX512VL-LABEL: splat_mem_v4i64_from_v2i64:
1349 ; AVX512VL-NEXT: vmovdqa64 (%rdi), %xmm0
1350 ; AVX512VL-NEXT: vpbroadcastq %xmm0, %ymm0
1351 ; AVX512VL-NEXT: retq
1352 %v = load <2 x i64>, <2 x i64>* %ptr
1353 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1354 ret <4 x i64> %shuffle
1357 define <4 x double> @splat_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1358 ; AVX1-LABEL: splat_mem_v4f64_from_v2f64:
1360 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
1361 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1364 ; AVX2-LABEL: splat_mem_v4f64_from_v2f64:
1366 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1369 ; AVX512VL-LABEL: splat_mem_v4f64_from_v2f64:
1371 ; AVX512VL-NEXT: vbroadcastsd (%rdi), %ymm0
1372 ; AVX512VL-NEXT: retq
1373 %v = load <2 x double>, <2 x double>* %ptr
1374 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1375 ret <4 x double> %shuffle
1378 define <4 x i64> @splat128_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1379 ; AVX1-LABEL: splat128_mem_v4i64_from_v2i64:
1381 ; AVX1-NEXT: vmovaps (%rdi), %xmm0
1382 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1385 ; AVX2-LABEL: splat128_mem_v4i64_from_v2i64:
1387 ; AVX2-NEXT: vmovaps (%rdi), %xmm0
1388 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1391 ; AVX512VL-LABEL: splat128_mem_v4i64_from_v2i64:
1393 ; AVX512VL-NEXT: vmovdqa64 (%rdi), %xmm0
1394 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm0, %ymm0
1395 ; AVX512VL-NEXT: retq
1396 %v = load <2 x i64>, <2 x i64>* %ptr
1397 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1398 ret <4 x i64> %shuffle
1401 define <4 x double> @splat128_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1402 ; AVX1-LABEL: splat128_mem_v4f64_from_v2f64:
1404 ; AVX1-NEXT: vmovaps (%rdi), %xmm0
1405 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1408 ; AVX2-LABEL: splat128_mem_v4f64_from_v2f64:
1410 ; AVX2-NEXT: vmovaps (%rdi), %xmm0
1411 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1414 ; AVX512VL-LABEL: splat128_mem_v4f64_from_v2f64:
1416 ; AVX512VL-NEXT: vmovapd (%rdi), %xmm0
1417 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm0, %ymm0
1418 ; AVX512VL-NEXT: retq
1419 %v = load <2 x double>, <2 x double>* %ptr
1420 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1421 ret <4 x double> %shuffle
1424 define <4 x double> @bitcast_v4f64_0426(<4 x double> %a, <4 x double> %b) {
1425 ; AVX1-LABEL: bitcast_v4f64_0426:
1427 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1430 ; AVX2-LABEL: bitcast_v4f64_0426:
1432 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1435 ; AVX512VL-LABEL: bitcast_v4f64_0426:
1437 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1438 ; AVX512VL-NEXT: retq
1439 %shuffle64 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
1440 %bitcast32 = bitcast <4 x double> %shuffle64 to <8 x float>
1441 %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>
1442 %bitcast16 = bitcast <8 x float> %shuffle32 to <16 x i16>
1443 %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>
1444 %bitcast64 = bitcast <16 x i16> %shuffle16 to <4 x double>
1445 ret <4 x double> %bitcast64
1448 define <4 x i64> @concat_v4i64_0167(<4 x i64> %a0, <4 x i64> %a1) {
1449 ; AVX1-LABEL: concat_v4i64_0167:
1451 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1454 ; AVX2-LABEL: concat_v4i64_0167:
1456 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1459 ; AVX512VL-LABEL: concat_v4i64_0167:
1461 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1462 ; AVX512VL-NEXT: retq
1463 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1464 %a1hi = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 6, i32 7>
1465 %shuffle64 = shufflevector <2 x i64> %a0lo, <2 x i64> %a1hi, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1466 ret <4 x i64> %shuffle64
1469 define <4 x i64> @concat_v4i64_0145_bc(<4 x i64> %a0, <4 x i64> %a1) {
1470 ; AVX1-LABEL: concat_v4i64_0145_bc:
1472 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1475 ; AVX2-LABEL: concat_v4i64_0145_bc:
1477 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1480 ; AVX512VL-LABEL: concat_v4i64_0145_bc:
1482 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
1483 ; AVX512VL-NEXT: retq
1484 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1485 %a1lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 4, i32 5>
1486 %bc0lo = bitcast <2 x i64> %a0lo to <4 x i32>
1487 %bc1lo = bitcast <2 x i64> %a1lo to <4 x i32>
1488 %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>
1489 %shuffle64 = bitcast <8 x i32> %shuffle32 to <4 x i64>
1490 ret <4 x i64> %shuffle64
1493 define <4 x i64> @insert_dup_mem_v4i64(i64* %ptr) {
1494 ; AVX1-LABEL: insert_dup_mem_v4i64:
1496 ; AVX1-NEXT: vbroadcastsd (%rdi), %ymm0
1499 ; AVX2-LABEL: insert_dup_mem_v4i64:
1501 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1504 ; AVX512VL-LABEL: insert_dup_mem_v4i64:
1506 ; AVX512VL-NEXT: vpbroadcastq (%rdi), %ymm0
1507 ; AVX512VL-NEXT: retq
1508 %tmp = load i64, i64* %ptr, align 1
1509 %tmp1 = insertelement <2 x i64> undef, i64 %tmp, i32 0
1510 %tmp2 = shufflevector <2 x i64> %tmp1, <2 x i64> undef, <4 x i32> zeroinitializer