1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx | FileCheck %s --check-prefix=ALL --check-prefix=AVX1
3 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX2
4 ; RUN: llc < %s -mcpu=knl -mattr=+avx512vl | FileCheck %s --check-prefix=ALL --check-prefix=AVX512VL
6 target triple = "x86_64-unknown-unknown"
8 define <4 x double> @shuffle_v4f64_0000(<4 x double> %a, <4 x double> %b) {
9 ; AVX1-LABEL: shuffle_v4f64_0000:
11 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
12 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
15 ; AVX2-LABEL: shuffle_v4f64_0000:
17 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
20 ; AVX512VL-LABEL: shuffle_v4f64_0000:
22 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
24 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
25 ret <4 x double> %shuffle
28 define <4 x double> @shuffle_v4f64_0001(<4 x double> %a, <4 x double> %b) {
29 ; AVX1-LABEL: shuffle_v4f64_0001:
31 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm0[0,0]
32 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
35 ; AVX2-LABEL: shuffle_v4f64_0001:
37 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,0,0,1]
40 ; AVX512VL-LABEL: shuffle_v4f64_0001:
42 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,0,0,1]
44 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
45 ret <4 x double> %shuffle
48 define <4 x double> @shuffle_v4f64_0020(<4 x double> %a, <4 x double> %b) {
49 ; AVX1-LABEL: shuffle_v4f64_0020:
51 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
52 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
53 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
54 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
57 ; AVX2-LABEL: shuffle_v4f64_0020:
59 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,0,2,0]
62 ; AVX512VL-LABEL: shuffle_v4f64_0020:
64 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,0,2,0]
66 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
67 ret <4 x double> %shuffle
70 define <4 x double> @shuffle_v4f64_0300(<4 x double> %a, <4 x double> %b) {
71 ; AVX1-LABEL: shuffle_v4f64_0300:
73 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
74 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
75 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
78 ; AVX2-LABEL: shuffle_v4f64_0300:
80 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,0,0]
83 ; AVX512VL-LABEL: shuffle_v4f64_0300:
85 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,0,0]
87 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
88 ret <4 x double> %shuffle
91 define <4 x double> @shuffle_v4f64_1000(<4 x double> %a, <4 x double> %b) {
92 ; AVX1-LABEL: shuffle_v4f64_1000:
94 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
95 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
96 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
99 ; AVX2-LABEL: shuffle_v4f64_1000:
101 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,0,0]
104 ; AVX512VL-LABEL: shuffle_v4f64_1000:
106 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,0,0]
107 ; AVX512VL-NEXT: retq
108 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
109 ret <4 x double> %shuffle
112 define <4 x double> @shuffle_v4f64_2200(<4 x double> %a, <4 x double> %b) {
113 ; AVX1-LABEL: shuffle_v4f64_2200:
115 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
116 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
119 ; AVX2-LABEL: shuffle_v4f64_2200:
121 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,0,0]
124 ; AVX512VL-LABEL: shuffle_v4f64_2200:
126 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,0,0]
127 ; AVX512VL-NEXT: retq
128 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
129 ret <4 x double> %shuffle
132 define <4 x double> @shuffle_v4f64_3330(<4 x double> %a, <4 x double> %b) {
133 ; AVX1-LABEL: shuffle_v4f64_3330:
135 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
136 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3]
137 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
140 ; AVX2-LABEL: shuffle_v4f64_3330:
142 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,0]
145 ; AVX512VL-LABEL: shuffle_v4f64_3330:
147 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,0]
148 ; AVX512VL-NEXT: retq
149 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
150 ret <4 x double> %shuffle
153 define <4 x double> @shuffle_v4f64_3210(<4 x double> %a, <4 x double> %b) {
154 ; AVX1-LABEL: shuffle_v4f64_3210:
156 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
157 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
160 ; AVX2-LABEL: shuffle_v4f64_3210:
162 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,1,0]
165 ; AVX512VL-LABEL: shuffle_v4f64_3210:
167 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,1,0]
168 ; AVX512VL-NEXT: retq
169 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
170 ret <4 x double> %shuffle
173 define <4 x double> @shuffle_v4f64_0023(<4 x double> %a, <4 x double> %b) {
174 ; ALL-LABEL: shuffle_v4f64_0023:
176 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,3]
179 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 3>
180 ret <4 x double> %shuffle
183 define <4 x double> @shuffle_v4f64_0022(<4 x double> %a, <4 x double> %b) {
184 ; ALL-LABEL: shuffle_v4f64_0022:
186 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
188 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
189 ret <4 x double> %shuffle
192 define <4 x double> @shuffle_v4f64mem_0022(<4 x double>* %ptr, <4 x double> %b) {
193 ; ALL-LABEL: shuffle_v4f64mem_0022:
195 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = mem[0,0,2,2]
197 %a = load <4 x double>, <4 x double>* %ptr
198 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
199 ret <4 x double> %shuffle
202 define <4 x double> @shuffle_v4f64_1032(<4 x double> %a, <4 x double> %b) {
203 ; ALL-LABEL: shuffle_v4f64_1032:
205 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
207 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
208 ret <4 x double> %shuffle
211 define <4 x double> @shuffle_v4f64_1133(<4 x double> %a, <4 x double> %b) {
212 ; ALL-LABEL: shuffle_v4f64_1133:
214 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
216 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
217 ret <4 x double> %shuffle
220 define <4 x double> @shuffle_v4f64_1023(<4 x double> %a, <4 x double> %b) {
221 ; ALL-LABEL: shuffle_v4f64_1023:
223 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,3]
225 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 3>
226 ret <4 x double> %shuffle
229 define <4 x double> @shuffle_v4f64_1022(<4 x double> %a, <4 x double> %b) {
230 ; ALL-LABEL: shuffle_v4f64_1022:
232 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
234 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 2>
235 ret <4 x double> %shuffle
238 define <4 x double> @shuffle_v4f64_0423(<4 x double> %a, <4 x double> %b) {
239 ; AVX1-LABEL: shuffle_v4f64_0423:
241 ; AVX1-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
242 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
245 ; AVX2-LABEL: shuffle_v4f64_0423:
247 ; AVX2-NEXT: vbroadcastsd %xmm1, %ymm1
248 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
251 ; AVX512VL-LABEL: shuffle_v4f64_0423:
253 ; AVX512VL-NEXT: vbroadcastsd %xmm1, %ymm1
254 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
255 ; AVX512VL-NEXT: retq
256 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 3>
257 ret <4 x double> %shuffle
260 define <4 x double> @shuffle_v4f64_0462(<4 x double> %a, <4 x double> %b) {
261 ; ALL-LABEL: shuffle_v4f64_0462:
263 ; ALL-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
264 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
265 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
267 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 6, i32 2>
268 ret <4 x double> %shuffle
271 define <4 x double> @shuffle_v4f64_0426(<4 x double> %a, <4 x double> %b) {
272 ; ALL-LABEL: shuffle_v4f64_0426:
274 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
276 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
277 ret <4 x double> %shuffle
280 define <4 x double> @shuffle_v4f64_1537(<4 x double> %a, <4 x double> %b) {
281 ; ALL-LABEL: shuffle_v4f64_1537:
283 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
285 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
286 ret <4 x double> %shuffle
289 define <4 x double> @shuffle_v4f64_4062(<4 x double> %a, <4 x double> %b) {
290 ; ALL-LABEL: shuffle_v4f64_4062:
292 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
294 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
295 ret <4 x double> %shuffle
298 define <4 x double> @shuffle_v4f64_5173(<4 x double> %a, <4 x double> %b) {
299 ; ALL-LABEL: shuffle_v4f64_5173:
301 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
303 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 7, i32 3>
304 ret <4 x double> %shuffle
307 define <4 x double> @shuffle_v4f64_5163(<4 x double> %a, <4 x double> %b) {
308 ; ALL-LABEL: shuffle_v4f64_5163:
310 ; ALL-NEXT: vshufpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
312 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
313 ret <4 x double> %shuffle
316 define <4 x double> @shuffle_v4f64_0527(<4 x double> %a, <4 x double> %b) {
317 ; ALL-LABEL: shuffle_v4f64_0527:
319 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
321 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
322 ret <4 x double> %shuffle
325 define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
326 ; ALL-LABEL: shuffle_v4f64_4163:
328 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
330 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
331 ret <4 x double> %shuffle
334 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
335 ; AVX1-LABEL: shuffle_v4f64_0145:
337 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
340 ; AVX2-LABEL: shuffle_v4f64_0145:
342 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
345 ; AVX512VL-LABEL: shuffle_v4f64_0145:
347 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
348 ; AVX512VL-NEXT: retq
349 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
350 ret <4 x double> %shuffle
353 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
354 ; AVX1-LABEL: shuffle_v4f64_4501:
356 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
359 ; AVX2-LABEL: shuffle_v4f64_4501:
361 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
364 ; AVX512VL-LABEL: shuffle_v4f64_4501:
366 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm1, %ymm0
367 ; AVX512VL-NEXT: retq
368 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
369 ret <4 x double> %shuffle
372 define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
373 ; ALL-LABEL: shuffle_v4f64_0167:
375 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
377 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
378 ret <4 x double> %shuffle
381 define <4 x double> @shuffle_v4f64_1054(<4 x double> %a, <4 x double> %b) {
382 ; AVX1-LABEL: shuffle_v4f64_1054:
384 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
385 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
388 ; AVX2-LABEL: shuffle_v4f64_1054:
390 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
391 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
394 ; AVX512VL-LABEL: shuffle_v4f64_1054:
396 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
397 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
398 ; AVX512VL-NEXT: retq
399 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
400 ret <4 x double> %shuffle
403 define <4 x double> @shuffle_v4f64_3254(<4 x double> %a, <4 x double> %b) {
404 ; AVX1-LABEL: shuffle_v4f64_3254:
406 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
407 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
410 ; AVX2-LABEL: shuffle_v4f64_3254:
412 ; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
413 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
416 ; AVX512VL-LABEL: shuffle_v4f64_3254:
418 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
419 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
420 ; AVX512VL-NEXT: retq
421 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
422 ret <4 x double> %shuffle
425 define <4 x double> @shuffle_v4f64_3276(<4 x double> %a, <4 x double> %b) {
426 ; AVX1-LABEL: shuffle_v4f64_3276:
428 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
429 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
432 ; AVX2-LABEL: shuffle_v4f64_3276:
434 ; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
435 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
438 ; AVX512VL-LABEL: shuffle_v4f64_3276:
440 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
441 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
442 ; AVX512VL-NEXT: retq
443 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
444 ret <4 x double> %shuffle
447 define <4 x double> @shuffle_v4f64_1076(<4 x double> %a, <4 x double> %b) {
448 ; ALL-LABEL: shuffle_v4f64_1076:
450 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
451 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
453 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
454 ret <4 x double> %shuffle
457 define <4 x double> @shuffle_v4f64_0415(<4 x double> %a, <4 x double> %b) {
458 ; AVX1-LABEL: shuffle_v4f64_0415:
460 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
461 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
462 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
465 ; AVX2-LABEL: shuffle_v4f64_0415:
467 ; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
468 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
469 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
472 ; AVX512VL-LABEL: shuffle_v4f64_0415:
474 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
475 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
476 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
477 ; AVX512VL-NEXT: retq
478 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
479 ret <4 x double> %shuffle
482 define <4 x double> @shuffle_v4f64_u062(<4 x double> %a, <4 x double> %b) {
483 ; ALL-LABEL: shuffle_v4f64_u062:
485 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
487 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 undef, i32 0, i32 6, i32 2>
488 ret <4 x double> %shuffle
491 define <4 x double> @shuffle_v4f64_11uu(<4 x double> %a, <4 x double> %b) {
492 ; ALL-LABEL: shuffle_v4f64_11uu:
494 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,2,2]
496 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 undef, i32 undef>
497 ret <4 x double> %shuffle
500 define <4 x double> @shuffle_v4f64_22uu(<4 x double> %a, <4 x double> %b) {
501 ; AVX1-LABEL: shuffle_v4f64_22uu:
503 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
504 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
507 ; AVX2-LABEL: shuffle_v4f64_22uu:
509 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,3]
512 ; AVX512VL-LABEL: shuffle_v4f64_22uu:
514 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,3]
515 ; AVX512VL-NEXT: retq
516 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 undef, i32 undef>
517 ret <4 x double> %shuffle
520 define <4 x double> @shuffle_v4f64_3333(<4 x double> %a, <4 x double> %b) {
521 ; AVX1-LABEL: shuffle_v4f64_3333:
523 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
524 ; AVX1-NEXT: vmovhlps {{.*#+}} xmm0 = xmm0[1,1]
525 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
528 ; AVX2-LABEL: shuffle_v4f64_3333:
530 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,3]
533 ; AVX512VL-LABEL: shuffle_v4f64_3333:
535 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,3]
536 ; AVX512VL-NEXT: retq
537 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
538 ret <4 x double> %shuffle
541 define <4 x i64> @shuffle_v4i64_0000(<4 x i64> %a, <4 x i64> %b) {
542 ; AVX1-LABEL: shuffle_v4i64_0000:
544 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
545 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
548 ; AVX2-LABEL: shuffle_v4i64_0000:
550 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
553 ; AVX512VL-LABEL: shuffle_v4i64_0000:
555 ; AVX512VL-NEXT: vpbroadcastq %xmm0, %ymm0
556 ; AVX512VL-NEXT: retq
557 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
558 ret <4 x i64> %shuffle
561 define <4 x i64> @shuffle_v4i64_0001(<4 x i64> %a, <4 x i64> %b) {
562 ; AVX1-LABEL: shuffle_v4i64_0001:
564 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm0[0,0]
565 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
568 ; AVX2-LABEL: shuffle_v4i64_0001:
570 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
573 ; AVX512VL-LABEL: shuffle_v4i64_0001:
575 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
576 ; AVX512VL-NEXT: retq
577 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
578 ret <4 x i64> %shuffle
581 define <4 x i64> @shuffle_v4i64_0020(<4 x i64> %a, <4 x i64> %b) {
582 ; AVX1-LABEL: shuffle_v4i64_0020:
584 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
585 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
586 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
587 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
590 ; AVX2-LABEL: shuffle_v4i64_0020:
592 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
595 ; AVX512VL-LABEL: shuffle_v4i64_0020:
597 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
598 ; AVX512VL-NEXT: retq
599 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
600 ret <4 x i64> %shuffle
603 define <4 x i64> @shuffle_v4i64_0112(<4 x i64> %a, <4 x i64> %b) {
604 ; AVX1-LABEL: shuffle_v4i64_0112:
606 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
607 ; AVX1-NEXT: vshufpd {{.*#+}} xmm1 = xmm0[1],xmm1[0]
608 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
611 ; AVX2-LABEL: shuffle_v4i64_0112:
613 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
616 ; AVX512VL-LABEL: shuffle_v4i64_0112:
618 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
619 ; AVX512VL-NEXT: retq
620 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 1, i32 2>
621 ret <4 x i64> %shuffle
624 define <4 x i64> @shuffle_v4i64_0300(<4 x i64> %a, <4 x i64> %b) {
625 ; AVX1-LABEL: shuffle_v4i64_0300:
627 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
628 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
629 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
632 ; AVX2-LABEL: shuffle_v4i64_0300:
634 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
637 ; AVX512VL-LABEL: shuffle_v4i64_0300:
639 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
640 ; AVX512VL-NEXT: retq
641 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
642 ret <4 x i64> %shuffle
645 define <4 x i64> @shuffle_v4i64_1000(<4 x i64> %a, <4 x i64> %b) {
646 ; AVX1-LABEL: shuffle_v4i64_1000:
648 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
649 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
650 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
653 ; AVX2-LABEL: shuffle_v4i64_1000:
655 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
658 ; AVX512VL-LABEL: shuffle_v4i64_1000:
660 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
661 ; AVX512VL-NEXT: retq
662 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
663 ret <4 x i64> %shuffle
666 define <4 x i64> @shuffle_v4i64_2200(<4 x i64> %a, <4 x i64> %b) {
667 ; AVX1-LABEL: shuffle_v4i64_2200:
669 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
670 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
673 ; AVX2-LABEL: shuffle_v4i64_2200:
675 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
678 ; AVX512VL-LABEL: shuffle_v4i64_2200:
680 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
681 ; AVX512VL-NEXT: retq
682 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
683 ret <4 x i64> %shuffle
686 define <4 x i64> @shuffle_v4i64_3330(<4 x i64> %a, <4 x i64> %b) {
687 ; AVX1-LABEL: shuffle_v4i64_3330:
689 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
690 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3]
691 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
694 ; AVX2-LABEL: shuffle_v4i64_3330:
696 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
699 ; AVX512VL-LABEL: shuffle_v4i64_3330:
701 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
702 ; AVX512VL-NEXT: retq
703 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
704 ret <4 x i64> %shuffle
707 define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
708 ; AVX1-LABEL: shuffle_v4i64_3210:
710 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
711 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
714 ; AVX2-LABEL: shuffle_v4i64_3210:
716 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
719 ; AVX512VL-LABEL: shuffle_v4i64_3210:
721 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
722 ; AVX512VL-NEXT: retq
723 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
724 ret <4 x i64> %shuffle
727 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
728 ; AVX1-LABEL: shuffle_v4i64_0124:
730 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
731 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
732 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3]
735 ; AVX2-LABEL: shuffle_v4i64_0124:
737 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
738 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
741 ; AVX512VL-LABEL: shuffle_v4i64_0124:
743 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
744 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
745 ; AVX512VL-NEXT: retq
746 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
747 ret <4 x i64> %shuffle
750 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
751 ; AVX1-LABEL: shuffle_v4i64_0142:
753 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm1, %ymm1
754 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,1,2,2]
755 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
758 ; AVX2-LABEL: shuffle_v4i64_0142:
760 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
761 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
762 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
765 ; AVX512VL-LABEL: shuffle_v4i64_0142:
767 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm1, %ymm1
768 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
769 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
770 ; AVX512VL-NEXT: retq
771 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
772 ret <4 x i64> %shuffle
775 define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
776 ; AVX1-LABEL: shuffle_v4i64_0412:
778 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
779 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
780 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
781 ; AVX1-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
782 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
785 ; AVX2-LABEL: shuffle_v4i64_0412:
787 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
788 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
789 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
792 ; AVX512VL-LABEL: shuffle_v4i64_0412:
794 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
795 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
796 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
797 ; AVX512VL-NEXT: retq
798 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
799 ret <4 x i64> %shuffle
802 define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
803 ; AVX1-LABEL: shuffle_v4i64_4012:
805 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
806 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
807 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
808 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
809 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
812 ; AVX2-LABEL: shuffle_v4i64_4012:
814 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
815 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
818 ; AVX512VL-LABEL: shuffle_v4i64_4012:
820 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
821 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
822 ; AVX512VL-NEXT: retq
823 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
824 ret <4 x i64> %shuffle
827 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
828 ; AVX1-LABEL: shuffle_v4i64_0145:
830 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
833 ; AVX2-LABEL: shuffle_v4i64_0145:
835 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
838 ; AVX512VL-LABEL: shuffle_v4i64_0145:
840 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
841 ; AVX512VL-NEXT: retq
842 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
843 ret <4 x i64> %shuffle
846 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
847 ; AVX1-LABEL: shuffle_v4i64_0451:
849 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm1[1],xmm0[1]
850 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
851 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
854 ; AVX2-LABEL: shuffle_v4i64_0451:
856 ; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
857 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
858 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
861 ; AVX512VL-LABEL: shuffle_v4i64_0451:
863 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm0, %ymm0
864 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
865 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
866 ; AVX512VL-NEXT: retq
867 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
868 ret <4 x i64> %shuffle
871 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
872 ; AVX1-LABEL: shuffle_v4i64_4501:
874 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
877 ; AVX2-LABEL: shuffle_v4i64_4501:
879 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
882 ; AVX512VL-LABEL: shuffle_v4i64_4501:
884 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm1, %ymm0
885 ; AVX512VL-NEXT: retq
886 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
887 ret <4 x i64> %shuffle
890 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
891 ; AVX1-LABEL: shuffle_v4i64_4015:
893 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
894 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
895 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
898 ; AVX2-LABEL: shuffle_v4i64_4015:
900 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
901 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
902 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
905 ; AVX512VL-LABEL: shuffle_v4i64_4015:
907 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm1, %ymm1
908 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
909 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
910 ; AVX512VL-NEXT: retq
911 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
912 ret <4 x i64> %shuffle
915 define <4 x i64> @shuffle_v4i64_2u35(<4 x i64> %a, <4 x i64> %b) {
916 ; AVX1-LABEL: shuffle_v4i64_2u35:
918 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
919 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm1 = xmm0[1],xmm1[1]
920 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
923 ; AVX2-LABEL: shuffle_v4i64_2u35:
925 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
926 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
929 ; AVX512VL-LABEL: shuffle_v4i64_2u35:
931 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
932 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
933 ; AVX512VL-NEXT: retq
934 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 undef, i32 3, i32 5>
935 ret <4 x i64> %shuffle
938 define <4 x i64> @shuffle_v4i64_1251(<4 x i64> %a, <4 x i64> %b) {
939 ; AVX1-LABEL: shuffle_v4i64_1251:
941 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
942 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm2[0],ymm0[2],ymm2[3]
943 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
944 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
945 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
948 ; AVX2-LABEL: shuffle_v4i64_1251:
950 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
951 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
952 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
955 ; AVX512VL-LABEL: shuffle_v4i64_1251:
957 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
958 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
959 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
960 ; AVX512VL-NEXT: retq
961 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 5, i32 1>
962 ret <4 x i64> %shuffle
965 define <4 x i64> @shuffle_v4i64_1054(<4 x i64> %a, <4 x i64> %b) {
966 ; AVX1-LABEL: shuffle_v4i64_1054:
968 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
969 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
972 ; AVX2-LABEL: shuffle_v4i64_1054:
974 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
975 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
978 ; AVX512VL-LABEL: shuffle_v4i64_1054:
980 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
981 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
982 ; AVX512VL-NEXT: retq
983 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
984 ret <4 x i64> %shuffle
987 define <4 x i64> @shuffle_v4i64_3254(<4 x i64> %a, <4 x i64> %b) {
988 ; AVX1-LABEL: shuffle_v4i64_3254:
990 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
991 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
994 ; AVX2-LABEL: shuffle_v4i64_3254:
996 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
997 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1000 ; AVX512VL-LABEL: shuffle_v4i64_3254:
1002 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
1003 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1004 ; AVX512VL-NEXT: retq
1005 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
1006 ret <4 x i64> %shuffle
1009 define <4 x i64> @shuffle_v4i64_3276(<4 x i64> %a, <4 x i64> %b) {
1010 ; AVX1-LABEL: shuffle_v4i64_3276:
1012 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1013 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1016 ; AVX2-LABEL: shuffle_v4i64_3276:
1018 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1019 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1022 ; AVX512VL-LABEL: shuffle_v4i64_3276:
1024 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1025 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1026 ; AVX512VL-NEXT: retq
1027 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
1028 ret <4 x i64> %shuffle
1031 define <4 x i64> @shuffle_v4i64_1076(<4 x i64> %a, <4 x i64> %b) {
1032 ; AVX1-LABEL: shuffle_v4i64_1076:
1034 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1035 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1038 ; AVX2-LABEL: shuffle_v4i64_1076:
1040 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1041 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1044 ; AVX512VL-LABEL: shuffle_v4i64_1076:
1046 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1047 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1048 ; AVX512VL-NEXT: retq
1049 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
1050 ret <4 x i64> %shuffle
1053 define <4 x i64> @shuffle_v4i64_0415(<4 x i64> %a, <4 x i64> %b) {
1054 ; AVX1-LABEL: shuffle_v4i64_0415:
1056 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
1057 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
1058 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
1061 ; AVX2-LABEL: shuffle_v4i64_0415:
1063 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1064 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1065 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1068 ; AVX512VL-LABEL: shuffle_v4i64_0415:
1070 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1071 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1072 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1073 ; AVX512VL-NEXT: retq
1074 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
1075 ret <4 x i64> %shuffle
1078 define <4 x i64> @shuffle_v4i64_z4z6(<4 x i64> %a) {
1079 ; AVX1-LABEL: shuffle_v4i64_z4z6:
1081 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1082 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1085 ; AVX2-LABEL: shuffle_v4i64_z4z6:
1087 ; 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]
1090 ; AVX512VL-LABEL: shuffle_v4i64_z4z6:
1092 ; 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]
1093 ; AVX512VL-NEXT: retq
1094 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 0, i32 4, i32 0, i32 6>
1095 ret <4 x i64> %shuffle
1098 define <4 x i64> @shuffle_v4i64_5zuz(<4 x i64> %a) {
1099 ; AVX1-LABEL: shuffle_v4i64_5zuz:
1101 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1102 ; AVX1-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
1105 ; AVX2-LABEL: shuffle_v4i64_5zuz:
1107 ; 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
1110 ; AVX512VL-LABEL: shuffle_v4i64_5zuz:
1112 ; 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
1113 ; AVX512VL-NEXT: retq
1114 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 5, i32 0, i32 undef, i32 0>
1115 ret <4 x i64> %shuffle
1118 define <4 x i64> @shuffle_v4i64_40u2(<4 x i64> %a, <4 x i64> %b) {
1119 ; AVX1-LABEL: shuffle_v4i64_40u2:
1121 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1124 ; AVX2-LABEL: shuffle_v4i64_40u2:
1126 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1129 ; AVX512VL-LABEL: shuffle_v4i64_40u2:
1131 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1132 ; AVX512VL-NEXT: retq
1133 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 undef, i32 2>
1134 ret <4 x i64> %shuffle
1137 define <4 x i64> @shuffle_v4i64_11uu(<4 x i64> %a, <4 x i64> %b) {
1138 ; AVX1-LABEL: shuffle_v4i64_11uu:
1140 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,2,2]
1143 ; AVX2-LABEL: shuffle_v4i64_11uu:
1145 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
1148 ; AVX512VL-LABEL: shuffle_v4i64_11uu:
1150 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
1151 ; AVX512VL-NEXT: retq
1152 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 1, i32 undef, i32 undef>
1153 ret <4 x i64> %shuffle
1156 define <4 x i64> @shuffle_v4i64_22uu(<4 x i64> %a, <4 x i64> %b) {
1157 ; AVX1-LABEL: shuffle_v4i64_22uu:
1159 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
1160 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
1163 ; AVX2-LABEL: shuffle_v4i64_22uu:
1165 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,3]
1168 ; AVX512VL-LABEL: shuffle_v4i64_22uu:
1170 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,3]
1171 ; AVX512VL-NEXT: retq
1172 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 undef, i32 undef>
1173 ret <4 x i64> %shuffle
1176 define <4 x i64> @shuffle_v4i64_3333(<4 x i64> %a, <4 x i64> %b) {
1177 ; AVX1-LABEL: shuffle_v4i64_3333:
1179 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
1180 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,1]
1181 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1184 ; AVX2-LABEL: shuffle_v4i64_3333:
1186 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,3]
1189 ; AVX512VL-LABEL: shuffle_v4i64_3333:
1191 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,3]
1192 ; AVX512VL-NEXT: retq
1193 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
1194 ret <4 x i64> %shuffle
1197 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
1198 ; ALL-LABEL: stress_test1:
1200 %c = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 1, i32 1, i32 0>
1201 %d = shufflevector <4 x i64> %c, <4 x i64> undef, <4 x i32> <i32 3, i32 undef, i32 2, i32 undef>
1202 %e = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 undef>
1203 %f = shufflevector <4 x i64> %d, <4 x i64> %e, <4 x i32> <i32 5, i32 1, i32 1, i32 0>
1208 define <4 x i64> @insert_reg_and_zero_v4i64(i64 %a) {
1209 ; ALL-LABEL: insert_reg_and_zero_v4i64:
1211 ; ALL-NEXT: vmovq %rdi, %xmm0
1213 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1214 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1215 ret <4 x i64> %shuffle
1218 define <4 x i64> @insert_mem_and_zero_v4i64(i64* %ptr) {
1219 ; AVX1-LABEL: insert_mem_and_zero_v4i64:
1221 ; AVX1-NEXT: vmovq {{.*#+}} xmm0 = mem[0],zero
1224 ; AVX2-LABEL: insert_mem_and_zero_v4i64:
1226 ; AVX2-NEXT: vmovq {{.*#+}} xmm0 = mem[0],zero
1229 ; AVX512VL-LABEL: insert_mem_and_zero_v4i64:
1231 ; AVX512VL-NEXT: vmovq (%rdi), %xmm0
1232 ; AVX512VL-NEXT: retq
1233 %a = load i64, i64* %ptr
1234 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1235 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1236 ret <4 x i64> %shuffle
1239 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
1240 ; AVX1-LABEL: insert_reg_and_zero_v4f64:
1242 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1243 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1246 ; AVX2-LABEL: insert_reg_and_zero_v4f64:
1248 ; AVX2-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1249 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1252 ; AVX512VL-LABEL: insert_reg_and_zero_v4f64:
1254 ; AVX512VL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
1255 ; AVX512VL-NEXT: vmovsd %xmm0, %xmm1, %xmm0
1256 ; AVX512VL-NEXT: retq
1257 %v = insertelement <4 x double> undef, double %a, i32 0
1258 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1259 ret <4 x double> %shuffle
1262 define <4 x double> @insert_mem_and_zero_v4f64(double* %ptr) {
1263 ; AVX1-LABEL: insert_mem_and_zero_v4f64:
1265 ; AVX1-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1268 ; AVX2-LABEL: insert_mem_and_zero_v4f64:
1270 ; AVX2-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1273 ; AVX512VL-LABEL: insert_mem_and_zero_v4f64:
1275 ; AVX512VL-NEXT: vmovsd (%rdi), %xmm0
1276 ; AVX512VL-NEXT: retq
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> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1280 ret <4 x double> %shuffle
1283 define <4 x double> @splat_mem_v4f64(double* %ptr) {
1284 ; ALL-LABEL: splat_mem_v4f64:
1286 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1288 %a = load double, double* %ptr
1289 %v = insertelement <4 x double> undef, double %a, i32 0
1290 %shuffle = shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1291 ret <4 x double> %shuffle
1294 define <4 x i64> @splat_mem_v4i64(i64* %ptr) {
1295 ; AVX1-LABEL: splat_mem_v4i64:
1297 ; AVX1-NEXT: vbroadcastsd (%rdi), %ymm0
1300 ; AVX2-LABEL: splat_mem_v4i64:
1302 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1305 ; AVX512VL-LABEL: splat_mem_v4i64:
1307 ; AVX512VL-NEXT: vpbroadcastq (%rdi), %ymm0
1308 ; AVX512VL-NEXT: retq
1309 %a = load i64, i64* %ptr
1310 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1311 %shuffle = shufflevector <4 x i64> %v, <4 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1312 ret <4 x i64> %shuffle
1315 define <4 x double> @splat_mem_v4f64_2(double* %p) {
1316 ; ALL-LABEL: splat_mem_v4f64_2:
1318 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1320 %1 = load double, double* %p
1321 %2 = insertelement <2 x double> undef, double %1, i32 0
1322 %3 = shufflevector <2 x double> %2, <2 x double> undef, <4 x i32> zeroinitializer
1326 define <4 x double> @splat_v4f64(<2 x double> %r) {
1327 ; AVX1-LABEL: splat_v4f64:
1329 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
1330 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1333 ; AVX2-LABEL: splat_v4f64:
1335 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
1338 ; AVX512VL-LABEL: splat_v4f64:
1340 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
1341 ; AVX512VL-NEXT: retq
1342 %1 = shufflevector <2 x double> %r, <2 x double> undef, <4 x i32> zeroinitializer
1346 define <4 x i64> @splat_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1347 ; AVX1-LABEL: splat_mem_v4i64_from_v2i64:
1349 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
1350 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1353 ; AVX2-LABEL: splat_mem_v4i64_from_v2i64:
1355 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1358 ; AVX512VL-LABEL: splat_mem_v4i64_from_v2i64:
1360 ; AVX512VL-NEXT: vmovdqa64 (%rdi), %xmm0
1361 ; AVX512VL-NEXT: vpbroadcastq %xmm0, %ymm0
1362 ; AVX512VL-NEXT: retq
1363 %v = load <2 x i64>, <2 x i64>* %ptr
1364 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1365 ret <4 x i64> %shuffle
1368 define <4 x double> @splat_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1369 ; AVX1-LABEL: splat_mem_v4f64_from_v2f64:
1371 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = mem[0,0]
1372 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1375 ; AVX2-LABEL: splat_mem_v4f64_from_v2f64:
1377 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1380 ; AVX512VL-LABEL: splat_mem_v4f64_from_v2f64:
1382 ; AVX512VL-NEXT: vbroadcastsd (%rdi), %ymm0
1383 ; AVX512VL-NEXT: retq
1384 %v = load <2 x double>, <2 x double>* %ptr
1385 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1386 ret <4 x double> %shuffle
1389 define <4 x i64> @splat128_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1390 ; AVX1-LABEL: splat128_mem_v4i64_from_v2i64:
1392 ; AVX1-NEXT: vmovaps (%rdi), %xmm0
1393 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1396 ; AVX2-LABEL: splat128_mem_v4i64_from_v2i64:
1398 ; AVX2-NEXT: vmovaps (%rdi), %xmm0
1399 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1402 ; AVX512VL-LABEL: splat128_mem_v4i64_from_v2i64:
1404 ; AVX512VL-NEXT: vmovdqa64 (%rdi), %xmm0
1405 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm0, %ymm0
1406 ; AVX512VL-NEXT: retq
1407 %v = load <2 x i64>, <2 x i64>* %ptr
1408 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1409 ret <4 x i64> %shuffle
1412 define <4 x double> @splat128_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1413 ; AVX1-LABEL: splat128_mem_v4f64_from_v2f64:
1415 ; AVX1-NEXT: vmovaps (%rdi), %xmm0
1416 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1419 ; AVX2-LABEL: splat128_mem_v4f64_from_v2f64:
1421 ; AVX2-NEXT: vmovaps (%rdi), %xmm0
1422 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1425 ; AVX512VL-LABEL: splat128_mem_v4f64_from_v2f64:
1427 ; AVX512VL-NEXT: vmovapd (%rdi), %xmm0
1428 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm0, %ymm0
1429 ; AVX512VL-NEXT: retq
1430 %v = load <2 x double>, <2 x double>* %ptr
1431 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1432 ret <4 x double> %shuffle
1435 define <4 x double> @bitcast_v4f64_0426(<4 x double> %a, <4 x double> %b) {
1436 ; AVX1-LABEL: bitcast_v4f64_0426:
1438 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1441 ; AVX2-LABEL: bitcast_v4f64_0426:
1443 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1446 ; AVX512VL-LABEL: bitcast_v4f64_0426:
1448 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1449 ; AVX512VL-NEXT: retq
1450 %shuffle64 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
1451 %bitcast32 = bitcast <4 x double> %shuffle64 to <8 x float>
1452 %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>
1453 %bitcast16 = bitcast <8 x float> %shuffle32 to <16 x i16>
1454 %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>
1455 %bitcast64 = bitcast <16 x i16> %shuffle16 to <4 x double>
1456 ret <4 x double> %bitcast64
1459 define <4 x i64> @concat_v4i64_0167(<4 x i64> %a0, <4 x i64> %a1) {
1460 ; AVX1-LABEL: concat_v4i64_0167:
1462 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1465 ; AVX2-LABEL: concat_v4i64_0167:
1467 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1470 ; AVX512VL-LABEL: concat_v4i64_0167:
1472 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1473 ; AVX512VL-NEXT: retq
1474 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1475 %a1hi = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 6, i32 7>
1476 %shuffle64 = shufflevector <2 x i64> %a0lo, <2 x i64> %a1hi, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1477 ret <4 x i64> %shuffle64
1480 define <4 x i64> @concat_v4i64_0145_bc(<4 x i64> %a0, <4 x i64> %a1) {
1481 ; AVX1-LABEL: concat_v4i64_0145_bc:
1483 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1486 ; AVX2-LABEL: concat_v4i64_0145_bc:
1488 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1491 ; AVX512VL-LABEL: concat_v4i64_0145_bc:
1493 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
1494 ; AVX512VL-NEXT: retq
1495 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1496 %a1lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 4, i32 5>
1497 %bc0lo = bitcast <2 x i64> %a0lo to <4 x i32>
1498 %bc1lo = bitcast <2 x i64> %a1lo to <4 x i32>
1499 %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>
1500 %shuffle64 = bitcast <8 x i32> %shuffle32 to <4 x i64>
1501 ret <4 x i64> %shuffle64
1504 define <4 x i64> @insert_dup_mem_v4i64(i64* %ptr) {
1505 ; AVX1-LABEL: insert_dup_mem_v4i64:
1507 ; AVX1-NEXT: vbroadcastsd (%rdi), %ymm0
1510 ; AVX2-LABEL: insert_dup_mem_v4i64:
1512 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
1515 ; AVX512VL-LABEL: insert_dup_mem_v4i64:
1517 ; AVX512VL-NEXT: vpbroadcastq (%rdi), %ymm0
1518 ; AVX512VL-NEXT: retq
1519 %tmp = load i64, i64* %ptr, align 1
1520 %tmp1 = insertelement <2 x i64> undef, i64 %tmp, i32 0
1521 %tmp2 = shufflevector <2 x i64> %tmp1, <2 x i64> undef, <4 x i32> zeroinitializer