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: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
116 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
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_2222(<4 x double> %a, <4 x double> %b) {
133 ; AVX1-LABEL: shuffle_v4f64_2222:
135 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
136 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
139 ; AVX2-LABEL: shuffle_v4f64_2222:
141 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,2]
144 ; AVX512VL-LABEL: shuffle_v4f64_2222:
146 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,2]
147 ; AVX512VL-NEXT: retq
148 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 2, i32 2>
149 ret <4 x double> %shuffle
152 define <4 x double> @shuffle_v4f64_2222_bc(<4 x i64> %a, <4 x i64> %b) {
153 ; AVX1-LABEL: shuffle_v4f64_2222_bc:
155 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
156 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
159 ; AVX2-LABEL: shuffle_v4f64_2222_bc:
161 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,2]
164 ; AVX512VL-LABEL: shuffle_v4f64_2222_bc:
166 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,2]
167 ; AVX512VL-NEXT: retq
168 %tmp0 = bitcast <4 x i64> %a to <4 x double>
169 %tmp1 = bitcast <4 x i64> %b to <4 x double>
170 %shuffle = shufflevector <4 x double> %tmp0, <4 x double> %tmp1, <4 x i32> <i32 2, i32 2, i32 2, i32 2>
171 ret <4 x double> %shuffle
174 define <4 x double> @shuffle_v4f64_3330(<4 x double> %a, <4 x double> %b) {
175 ; AVX1-LABEL: shuffle_v4f64_3330:
177 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
178 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3]
179 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
182 ; AVX2-LABEL: shuffle_v4f64_3330:
184 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,0]
187 ; AVX512VL-LABEL: shuffle_v4f64_3330:
189 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,0]
190 ; AVX512VL-NEXT: retq
191 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
192 ret <4 x double> %shuffle
195 define <4 x double> @shuffle_v4f64_3210(<4 x double> %a, <4 x double> %b) {
196 ; AVX1-LABEL: shuffle_v4f64_3210:
198 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
199 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
202 ; AVX2-LABEL: shuffle_v4f64_3210:
204 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,1,0]
207 ; AVX512VL-LABEL: shuffle_v4f64_3210:
209 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,1,0]
210 ; AVX512VL-NEXT: retq
211 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
212 ret <4 x double> %shuffle
215 define <4 x double> @shuffle_v4f64_0023(<4 x double> %a, <4 x double> %b) {
216 ; ALL-LABEL: shuffle_v4f64_0023:
218 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,3]
221 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 3>
222 ret <4 x double> %shuffle
225 define <4 x double> @shuffle_v4f64_0022(<4 x double> %a, <4 x double> %b) {
226 ; ALL-LABEL: shuffle_v4f64_0022:
228 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
230 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
231 ret <4 x double> %shuffle
234 define <4 x double> @shuffle_v4f64mem_0022(<4 x double>* %ptr, <4 x double> %b) {
235 ; ALL-LABEL: shuffle_v4f64mem_0022:
237 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = mem[0,0,2,2]
239 %a = load <4 x double>, <4 x double>* %ptr
240 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
241 ret <4 x double> %shuffle
244 define <4 x double> @shuffle_v4f64_1032(<4 x double> %a, <4 x double> %b) {
245 ; ALL-LABEL: shuffle_v4f64_1032:
247 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
249 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
250 ret <4 x double> %shuffle
253 define <4 x double> @shuffle_v4f64_1133(<4 x double> %a, <4 x double> %b) {
254 ; ALL-LABEL: shuffle_v4f64_1133:
256 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
258 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
259 ret <4 x double> %shuffle
262 define <4 x double> @shuffle_v4f64_1023(<4 x double> %a, <4 x double> %b) {
263 ; ALL-LABEL: shuffle_v4f64_1023:
265 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,3]
267 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 3>
268 ret <4 x double> %shuffle
271 define <4 x double> @shuffle_v4f64_1022(<4 x double> %a, <4 x double> %b) {
272 ; ALL-LABEL: shuffle_v4f64_1022:
274 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
276 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 2>
277 ret <4 x double> %shuffle
280 define <4 x double> @shuffle_v4f64_0423(<4 x double> %a, <4 x double> %b) {
281 ; ALL-LABEL: shuffle_v4f64_0423:
283 ; ALL-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
284 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
286 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 3>
287 ret <4 x double> %shuffle
290 define <4 x double> @shuffle_v4f64_0462(<4 x double> %a, <4 x double> %b) {
291 ; ALL-LABEL: shuffle_v4f64_0462:
293 ; ALL-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
294 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
295 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
297 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 6, i32 2>
298 ret <4 x double> %shuffle
301 define <4 x double> @shuffle_v4f64_0426(<4 x double> %a, <4 x double> %b) {
302 ; ALL-LABEL: shuffle_v4f64_0426:
304 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
306 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
307 ret <4 x double> %shuffle
310 define <4 x double> @shuffle_v4f64_1537(<4 x double> %a, <4 x double> %b) {
311 ; ALL-LABEL: shuffle_v4f64_1537:
313 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
315 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
316 ret <4 x double> %shuffle
319 define <4 x double> @shuffle_v4f64_4062(<4 x double> %a, <4 x double> %b) {
320 ; ALL-LABEL: shuffle_v4f64_4062:
322 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
324 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
325 ret <4 x double> %shuffle
328 define <4 x double> @shuffle_v4f64_5173(<4 x double> %a, <4 x double> %b) {
329 ; ALL-LABEL: shuffle_v4f64_5173:
331 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
333 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 7, i32 3>
334 ret <4 x double> %shuffle
337 define <4 x double> @shuffle_v4f64_5163(<4 x double> %a, <4 x double> %b) {
338 ; ALL-LABEL: shuffle_v4f64_5163:
340 ; ALL-NEXT: vshufpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
342 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
343 ret <4 x double> %shuffle
346 define <4 x double> @shuffle_v4f64_0527(<4 x double> %a, <4 x double> %b) {
347 ; ALL-LABEL: shuffle_v4f64_0527:
349 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
351 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
352 ret <4 x double> %shuffle
355 define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
356 ; ALL-LABEL: shuffle_v4f64_4163:
358 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
360 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
361 ret <4 x double> %shuffle
364 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
365 ; ALL-LABEL: shuffle_v4f64_0145:
367 ; ALL-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
369 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
370 ret <4 x double> %shuffle
373 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
374 ; ALL-LABEL: shuffle_v4f64_4501:
376 ; ALL-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
378 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
379 ret <4 x double> %shuffle
382 define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
383 ; ALL-LABEL: shuffle_v4f64_0167:
385 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
387 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
388 ret <4 x double> %shuffle
391 define <4 x double> @shuffle_v4f64_1054(<4 x double> %a, <4 x double> %b) {
392 ; ALL-LABEL: shuffle_v4f64_1054:
394 ; ALL-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
395 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
397 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
398 ret <4 x double> %shuffle
401 define <4 x double> @shuffle_v4f64_3254(<4 x double> %a, <4 x double> %b) {
402 ; ALL-LABEL: shuffle_v4f64_3254:
404 ; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
405 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
407 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
408 ret <4 x double> %shuffle
411 define <4 x double> @shuffle_v4f64_3276(<4 x double> %a, <4 x double> %b) {
412 ; ALL-LABEL: shuffle_v4f64_3276:
414 ; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
415 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
417 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
418 ret <4 x double> %shuffle
421 define <4 x double> @shuffle_v4f64_1076(<4 x double> %a, <4 x double> %b) {
422 ; ALL-LABEL: shuffle_v4f64_1076:
424 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
425 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
427 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
428 ret <4 x double> %shuffle
431 define <4 x double> @shuffle_v4f64_0415(<4 x double> %a, <4 x double> %b) {
432 ; AVX1-LABEL: shuffle_v4f64_0415:
434 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
435 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
436 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
439 ; AVX2-LABEL: shuffle_v4f64_0415:
441 ; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
442 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
443 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
446 ; AVX512VL-LABEL: shuffle_v4f64_0415:
448 ; AVX512VL-NEXT: vmovapd {{.*#+}} ymm2 = [0,4,1,5]
449 ; AVX512VL-NEXT: vpermt2pd %ymm1, %ymm2, %ymm0
450 ; AVX512VL-NEXT: retq
451 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
452 ret <4 x double> %shuffle
455 define <4 x double> @shuffle_v4f64_u062(<4 x double> %a, <4 x double> %b) {
456 ; ALL-LABEL: shuffle_v4f64_u062:
458 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
460 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 undef, i32 0, i32 6, i32 2>
461 ret <4 x double> %shuffle
464 define <4 x double> @shuffle_v4f64_15uu(<4 x double> %a, <4 x double> %b) {
465 ; ALL-LABEL: shuffle_v4f64_15uu:
467 ; ALL-NEXT: vunpckhpd {{.*#+}} xmm0 = xmm0[1],xmm1[1]
469 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 undef, i32 undef>
470 ret <4 x double> %shuffle
473 define <4 x double> @shuffle_v4f64_11uu(<4 x double> %a, <4 x double> %b) {
474 ; ALL-LABEL: shuffle_v4f64_11uu:
476 ; ALL-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,1]
478 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 undef, i32 undef>
479 ret <4 x double> %shuffle
482 define <4 x double> @shuffle_v4f64_22uu(<4 x double> %a, <4 x double> %b) {
483 ; AVX1-LABEL: shuffle_v4f64_22uu:
485 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
486 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
489 ; AVX2-LABEL: shuffle_v4f64_22uu:
491 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,3]
494 ; AVX512VL-LABEL: shuffle_v4f64_22uu:
496 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,3]
497 ; AVX512VL-NEXT: retq
498 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 undef, i32 undef>
499 ret <4 x double> %shuffle
502 define <4 x double> @shuffle_v4f64_3333(<4 x double> %a, <4 x double> %b) {
503 ; AVX1-LABEL: shuffle_v4f64_3333:
505 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
506 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
509 ; AVX2-LABEL: shuffle_v4f64_3333:
511 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,3]
514 ; AVX512VL-LABEL: shuffle_v4f64_3333:
516 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,3]
517 ; AVX512VL-NEXT: retq
518 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
519 ret <4 x double> %shuffle
522 define <4 x double> @shuffle_v4f64_0z3z(<4 x double> %a, <4 x double> %b) {
523 ; AVX1-LABEL: shuffle_v4f64_0z3z:
525 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,3,2]
526 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
527 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
530 ; AVX2-LABEL: shuffle_v4f64_0z3z:
532 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,3,2]
533 ; AVX2-NEXT: vxorpd %ymm1, %ymm1, %ymm1
534 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
537 ; AVX512VL-LABEL: shuffle_v4f64_0z3z:
539 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,3,2]
540 ; AVX512VL-NEXT: vpxor %ymm1, %ymm1, %ymm1
541 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
542 ; AVX512VL-NEXT: retq
543 %shuffle = shufflevector <4 x double> %a, <4 x double> <double 0.000000e+00, double undef, double undef, double undef>, <4 x i32> <i32 0, i32 4, i32 3, i32 4>
544 ret <4 x double> %shuffle
547 define <4 x double> @shuffle_v4f64_1z2z(<4 x double> %a, <4 x double> %b) {
548 ; AVX1-LABEL: shuffle_v4f64_1z2z:
550 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
551 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
552 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
553 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
554 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
555 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
558 ; AVX2-LABEL: shuffle_v4f64_1z2z:
560 ; AVX2-NEXT: vxorpd %ymm1, %ymm1, %ymm1
561 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
562 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,2,0]
565 ; AVX512VL-LABEL: shuffle_v4f64_1z2z:
567 ; AVX512VL-NEXT: vpxor %ymm1, %ymm1, %ymm1
568 ; AVX512VL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
569 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,2,0]
570 ; AVX512VL-NEXT: retq
571 %1 = shufflevector <4 x double> %a, <4 x double> <double 0.000000e+00, double undef, double undef, double undef>, <4 x i32> <i32 1, i32 4, i32 2, i32 4>
575 define <4 x i64> @shuffle_v4i64_0000(<4 x i64> %a, <4 x i64> %b) {
576 ; AVX1-LABEL: shuffle_v4i64_0000:
578 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
579 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
582 ; AVX2-LABEL: shuffle_v4i64_0000:
584 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
587 ; AVX512VL-LABEL: shuffle_v4i64_0000:
589 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
590 ; AVX512VL-NEXT: retq
591 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
592 ret <4 x i64> %shuffle
595 define <4 x i64> @shuffle_v4i64_0001(<4 x i64> %a, <4 x i64> %b) {
596 ; AVX1-LABEL: shuffle_v4i64_0001:
598 ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[0,1,0,1]
599 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
602 ; AVX2-LABEL: shuffle_v4i64_0001:
604 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
607 ; AVX512VL-LABEL: shuffle_v4i64_0001:
609 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
610 ; AVX512VL-NEXT: retq
611 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
612 ret <4 x i64> %shuffle
615 define <4 x i64> @shuffle_v4i64_0020(<4 x i64> %a, <4 x i64> %b) {
616 ; AVX1-LABEL: shuffle_v4i64_0020:
618 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
619 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
620 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
621 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
624 ; AVX2-LABEL: shuffle_v4i64_0020:
626 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
629 ; AVX512VL-LABEL: shuffle_v4i64_0020:
631 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
632 ; AVX512VL-NEXT: retq
633 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
634 ret <4 x i64> %shuffle
637 define <4 x i64> @shuffle_v4i64_0112(<4 x i64> %a, <4 x i64> %b) {
638 ; AVX1-LABEL: shuffle_v4i64_0112:
640 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
641 ; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm0[8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7]
642 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
645 ; AVX2-LABEL: shuffle_v4i64_0112:
647 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
650 ; AVX512VL-LABEL: shuffle_v4i64_0112:
652 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
653 ; AVX512VL-NEXT: retq
654 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 1, i32 2>
655 ret <4 x i64> %shuffle
658 define <4 x i64> @shuffle_v4i64_0300(<4 x i64> %a, <4 x i64> %b) {
659 ; AVX1-LABEL: shuffle_v4i64_0300:
661 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
662 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
663 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
666 ; AVX2-LABEL: shuffle_v4i64_0300:
668 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
671 ; AVX512VL-LABEL: shuffle_v4i64_0300:
673 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
674 ; AVX512VL-NEXT: retq
675 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
676 ret <4 x i64> %shuffle
679 define <4 x i64> @shuffle_v4i64_1000(<4 x i64> %a, <4 x i64> %b) {
680 ; AVX1-LABEL: shuffle_v4i64_1000:
682 ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
683 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
684 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
687 ; AVX2-LABEL: shuffle_v4i64_1000:
689 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
692 ; AVX512VL-LABEL: shuffle_v4i64_1000:
694 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
695 ; AVX512VL-NEXT: retq
696 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
697 ret <4 x i64> %shuffle
700 define <4 x i64> @shuffle_v4i64_2200(<4 x i64> %a, <4 x i64> %b) {
701 ; AVX1-LABEL: shuffle_v4i64_2200:
703 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
704 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
707 ; AVX2-LABEL: shuffle_v4i64_2200:
709 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
712 ; AVX512VL-LABEL: shuffle_v4i64_2200:
714 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
715 ; AVX512VL-NEXT: retq
716 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
717 ret <4 x i64> %shuffle
720 define <4 x i64> @shuffle_v4i64_3330(<4 x i64> %a, <4 x i64> %b) {
721 ; AVX1-LABEL: shuffle_v4i64_3330:
723 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
724 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3]
725 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
728 ; AVX2-LABEL: shuffle_v4i64_3330:
730 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
733 ; AVX512VL-LABEL: shuffle_v4i64_3330:
735 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
736 ; AVX512VL-NEXT: retq
737 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
738 ret <4 x i64> %shuffle
741 define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
742 ; AVX1-LABEL: shuffle_v4i64_3210:
744 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
745 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
748 ; AVX2-LABEL: shuffle_v4i64_3210:
750 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
753 ; AVX512VL-LABEL: shuffle_v4i64_3210:
755 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
756 ; AVX512VL-NEXT: retq
757 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
758 ret <4 x i64> %shuffle
761 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
762 ; AVX1-LABEL: shuffle_v4i64_0124:
764 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
765 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
766 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3]
769 ; AVX2-LABEL: shuffle_v4i64_0124:
771 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
772 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
775 ; AVX512VL-LABEL: shuffle_v4i64_0124:
777 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
778 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
779 ; AVX512VL-NEXT: retq
780 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
781 ret <4 x i64> %shuffle
784 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
785 ; AVX1-LABEL: shuffle_v4i64_0142:
787 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
788 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,1,2,2]
789 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
792 ; AVX2-LABEL: shuffle_v4i64_0142:
794 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1
795 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
796 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
799 ; AVX512VL-LABEL: shuffle_v4i64_0142:
801 ; AVX512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1
802 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
803 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
804 ; AVX512VL-NEXT: retq
805 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
806 ret <4 x i64> %shuffle
809 define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
810 ; AVX1-LABEL: shuffle_v4i64_0412:
812 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
813 ; AVX1-NEXT: vpalignr {{.*#+}} xmm2 = xmm0[8,9,10,11,12,13,14,15],xmm2[0,1,2,3,4,5,6,7]
814 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
815 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
816 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
819 ; AVX2-LABEL: shuffle_v4i64_0412:
821 ; AVX2-NEXT: vpbroadcastq %xmm1, %xmm1
822 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
823 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
826 ; AVX512VL-LABEL: shuffle_v4i64_0412:
828 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %xmm1
829 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
830 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
831 ; AVX512VL-NEXT: retq
832 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
833 ret <4 x i64> %shuffle
836 define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
837 ; AVX1-LABEL: shuffle_v4i64_4012:
839 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
840 ; AVX1-NEXT: vpalignr {{.*#+}} xmm2 = xmm0[8,9,10,11,12,13,14,15],xmm2[0,1,2,3,4,5,6,7]
841 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
842 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
843 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
846 ; AVX2-LABEL: shuffle_v4i64_4012:
848 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
849 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
852 ; AVX512VL-LABEL: shuffle_v4i64_4012:
854 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
855 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
856 ; AVX512VL-NEXT: retq
857 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
858 ret <4 x i64> %shuffle
861 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
862 ; AVX1-LABEL: shuffle_v4i64_0145:
864 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
867 ; AVX2-LABEL: shuffle_v4i64_0145:
869 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
872 ; AVX512VL-LABEL: shuffle_v4i64_0145:
874 ; AVX512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
875 ; AVX512VL-NEXT: retq
876 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
877 ret <4 x i64> %shuffle
880 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
881 ; AVX1-LABEL: shuffle_v4i64_0451:
883 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm1[1],xmm0[1]
884 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
885 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
888 ; AVX2-LABEL: shuffle_v4i64_0451:
890 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
891 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
892 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
895 ; AVX512VL-LABEL: shuffle_v4i64_0451:
897 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
898 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
899 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
900 ; AVX512VL-NEXT: retq
901 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
902 ret <4 x i64> %shuffle
905 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
906 ; AVX1-LABEL: shuffle_v4i64_4501:
908 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
911 ; AVX2-LABEL: shuffle_v4i64_4501:
913 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
916 ; AVX512VL-LABEL: shuffle_v4i64_4501:
918 ; AVX512VL-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
919 ; AVX512VL-NEXT: retq
920 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
921 ret <4 x i64> %shuffle
924 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
925 ; AVX1-LABEL: shuffle_v4i64_4015:
927 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
928 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
929 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
932 ; AVX2-LABEL: shuffle_v4i64_4015:
934 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,2,1]
935 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
936 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
939 ; AVX512VL-LABEL: shuffle_v4i64_4015:
941 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,2,1]
942 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
943 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
944 ; AVX512VL-NEXT: retq
945 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
946 ret <4 x i64> %shuffle
949 define <4 x i64> @shuffle_v4i64_2u35(<4 x i64> %a, <4 x i64> %b) {
950 ; AVX1-LABEL: shuffle_v4i64_2u35:
952 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
953 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm1 = xmm0[1],xmm1[1]
954 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
957 ; AVX2-LABEL: shuffle_v4i64_2u35:
959 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
960 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
963 ; AVX512VL-LABEL: shuffle_v4i64_2u35:
965 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
966 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
967 ; AVX512VL-NEXT: retq
968 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 undef, i32 3, i32 5>
969 ret <4 x i64> %shuffle
972 define <4 x i64> @shuffle_v4i64_1251(<4 x i64> %a, <4 x i64> %b) {
973 ; AVX1-LABEL: shuffle_v4i64_1251:
975 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
976 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm2[0],ymm0[2],ymm2[3]
977 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
978 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
979 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
982 ; AVX2-LABEL: shuffle_v4i64_1251:
984 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
985 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
986 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
989 ; AVX512VL-LABEL: shuffle_v4i64_1251:
991 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
992 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
993 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
994 ; AVX512VL-NEXT: retq
995 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 5, i32 1>
996 ret <4 x i64> %shuffle
999 define <4 x i64> @shuffle_v4i64_1054(<4 x i64> %a, <4 x i64> %b) {
1000 ; AVX1-LABEL: shuffle_v4i64_1054:
1002 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1003 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1006 ; AVX2-LABEL: shuffle_v4i64_1054:
1008 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
1009 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1012 ; AVX512VL-LABEL: shuffle_v4i64_1054:
1014 ; AVX512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
1015 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1016 ; AVX512VL-NEXT: retq
1017 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
1018 ret <4 x i64> %shuffle
1021 define <4 x i64> @shuffle_v4i64_3254(<4 x i64> %a, <4 x i64> %b) {
1022 ; AVX1-LABEL: shuffle_v4i64_3254:
1024 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
1025 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1028 ; AVX2-LABEL: shuffle_v4i64_3254:
1030 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
1031 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1034 ; AVX512VL-LABEL: shuffle_v4i64_3254:
1036 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
1037 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1038 ; AVX512VL-NEXT: retq
1039 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
1040 ret <4 x i64> %shuffle
1043 define <4 x i64> @shuffle_v4i64_3276(<4 x i64> %a, <4 x i64> %b) {
1044 ; AVX1-LABEL: shuffle_v4i64_3276:
1046 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1047 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1050 ; AVX2-LABEL: shuffle_v4i64_3276:
1052 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1053 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1056 ; AVX512VL-LABEL: shuffle_v4i64_3276:
1058 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
1059 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1060 ; AVX512VL-NEXT: retq
1061 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
1062 ret <4 x i64> %shuffle
1065 define <4 x i64> @shuffle_v4i64_1076(<4 x i64> %a, <4 x i64> %b) {
1066 ; AVX1-LABEL: shuffle_v4i64_1076:
1068 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1069 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1072 ; AVX2-LABEL: shuffle_v4i64_1076:
1074 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1075 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1078 ; AVX512VL-LABEL: shuffle_v4i64_1076:
1080 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1081 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1082 ; AVX512VL-NEXT: retq
1083 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
1084 ret <4 x i64> %shuffle
1087 define <4 x i64> @shuffle_v4i64_0415(<4 x i64> %a, <4 x i64> %b) {
1088 ; AVX1-LABEL: shuffle_v4i64_0415:
1090 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
1091 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
1092 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
1095 ; AVX2-LABEL: shuffle_v4i64_0415:
1097 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1098 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1099 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1102 ; AVX512VL-LABEL: shuffle_v4i64_0415:
1104 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1105 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1106 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1107 ; AVX512VL-NEXT: retq
1108 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
1109 ret <4 x i64> %shuffle
1112 define <4 x i64> @shuffle_v4i64_z4z6(<4 x i64> %a) {
1113 ; AVX1-LABEL: shuffle_v4i64_z4z6:
1115 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1116 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1119 ; AVX2-LABEL: shuffle_v4i64_z4z6:
1121 ; 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]
1124 ; AVX512VL-LABEL: shuffle_v4i64_z4z6:
1126 ; 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]
1127 ; AVX512VL-NEXT: retq
1128 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 0, i32 4, i32 0, i32 6>
1129 ret <4 x i64> %shuffle
1132 define <4 x i64> @shuffle_v4i64_5zuz(<4 x i64> %a) {
1133 ; AVX1-LABEL: shuffle_v4i64_5zuz:
1135 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1136 ; AVX1-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
1139 ; AVX2-LABEL: shuffle_v4i64_5zuz:
1141 ; 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
1144 ; AVX512VL-LABEL: shuffle_v4i64_5zuz:
1146 ; 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
1147 ; AVX512VL-NEXT: retq
1148 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 5, i32 0, i32 undef, i32 0>
1149 ret <4 x i64> %shuffle
1152 define <4 x i64> @shuffle_v4i64_40u2(<4 x i64> %a, <4 x i64> %b) {
1153 ; AVX1-LABEL: shuffle_v4i64_40u2:
1155 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1158 ; AVX2-LABEL: shuffle_v4i64_40u2:
1160 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1163 ; AVX512VL-LABEL: shuffle_v4i64_40u2:
1165 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1166 ; AVX512VL-NEXT: retq
1167 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 undef, i32 2>
1168 ret <4 x i64> %shuffle
1171 define <4 x i64> @shuffle_v4i64_15uu(<4 x i64> %a, <4 x i64> %b) {
1172 ; ALL-LABEL: shuffle_v4i64_15uu:
1174 ; ALL-NEXT: vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
1176 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 5, i32 undef, i32 undef>
1177 ret <4 x i64> %shuffle
1180 define <4 x i64> @shuffle_v4i64_11uu(<4 x i64> %a, <4 x i64> %b) {
1181 ; ALL-LABEL: shuffle_v4i64_11uu:
1183 ; ALL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
1185 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 1, i32 undef, i32 undef>
1186 ret <4 x i64> %shuffle
1189 define <4 x i64> @shuffle_v4i64_22uu(<4 x i64> %a, <4 x i64> %b) {
1190 ; AVX1-LABEL: shuffle_v4i64_22uu:
1192 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
1193 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
1196 ; AVX2-LABEL: shuffle_v4i64_22uu:
1198 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,3]
1201 ; AVX512VL-LABEL: shuffle_v4i64_22uu:
1203 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,3]
1204 ; AVX512VL-NEXT: retq
1205 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 undef, i32 undef>
1206 ret <4 x i64> %shuffle
1209 define <4 x i64> @shuffle_v4i64_3333(<4 x i64> %a, <4 x i64> %b) {
1210 ; AVX1-LABEL: shuffle_v4i64_3333:
1212 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
1213 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
1216 ; AVX2-LABEL: shuffle_v4i64_3333:
1218 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,3]
1221 ; AVX512VL-LABEL: shuffle_v4i64_3333:
1223 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,3]
1224 ; AVX512VL-NEXT: retq
1225 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
1226 ret <4 x i64> %shuffle
1229 define <4 x i64> @shuffle_v4i64_1z3z(<4 x i64> %a, <4 x i64> %b) {
1230 ; AVX1-LABEL: shuffle_v4i64_1z3z:
1232 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1233 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
1234 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
1235 ; AVX1-NEXT: vshufpd {{.*#+}} xmm1 = xmm1[1],xmm0[0]
1236 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
1237 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1240 ; AVX2-LABEL: shuffle_v4i64_1z3z:
1242 ; 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
1245 ; AVX512VL-LABEL: shuffle_v4i64_1z3z:
1247 ; 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
1248 ; AVX512VL-NEXT: retq
1249 %shuffle = shufflevector <4 x i64> %a, <4 x i64> <i64 0, i64 undef, i64 undef, i64 undef>, <4 x i32> <i32 1, i32 4, i32 3, i32 4>
1250 ret <4 x i64> %shuffle
1253 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
1254 ; ALL-LABEL: stress_test1:
1256 %c = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 1, i32 1, i32 0>
1257 %d = shufflevector <4 x i64> %c, <4 x i64> undef, <4 x i32> <i32 3, i32 undef, i32 2, i32 undef>
1258 %e = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 undef>
1259 %f = shufflevector <4 x i64> %d, <4 x i64> %e, <4 x i32> <i32 5, i32 1, i32 1, i32 0>
1264 define <4 x i64> @insert_reg_and_zero_v4i64(i64 %a) {
1265 ; ALL-LABEL: insert_reg_and_zero_v4i64:
1267 ; ALL-NEXT: vmovq %rdi, %xmm0
1269 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1270 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1271 ret <4 x i64> %shuffle
1274 define <4 x i64> @insert_mem_and_zero_v4i64(i64* %ptr) {
1275 ; ALL-LABEL: insert_mem_and_zero_v4i64:
1277 ; ALL-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1279 %a = load i64, i64* %ptr
1280 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1281 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1282 ret <4 x i64> %shuffle
1285 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
1286 ; AVX1-LABEL: insert_reg_and_zero_v4f64:
1288 ; AVX1-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def>
1289 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1290 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1293 ; AVX2-LABEL: insert_reg_and_zero_v4f64:
1295 ; AVX2-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def>
1296 ; AVX2-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1297 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1300 ; AVX512VL-LABEL: insert_reg_and_zero_v4f64:
1302 ; AVX512VL-NEXT: vpxor %xmm1, %xmm1, %xmm1
1303 ; AVX512VL-NEXT: vmovsd {{.*#+}} xmm0 = xmm0[0],xmm1[1]
1304 ; AVX512VL-NEXT: retq
1305 %v = insertelement <4 x double> undef, double %a, i32 0
1306 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1307 ret <4 x double> %shuffle
1310 define <4 x double> @insert_mem_and_zero_v4f64(double* %ptr) {
1311 ; ALL-LABEL: insert_mem_and_zero_v4f64:
1313 ; ALL-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1315 %a = load double, double* %ptr
1316 %v = insertelement <4 x double> undef, double %a, i32 0
1317 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1318 ret <4 x double> %shuffle
1321 define <4 x double> @splat_mem_v4f64(double* %ptr) {
1322 ; ALL-LABEL: splat_mem_v4f64:
1324 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1326 %a = load double, double* %ptr
1327 %v = insertelement <4 x double> undef, double %a, i32 0
1328 %shuffle = shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1329 ret <4 x double> %shuffle
1332 define <4 x i64> @splat_mem_v4i64(i64* %ptr) {
1333 ; ALL-LABEL: splat_mem_v4i64:
1335 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1337 %a = load i64, i64* %ptr
1338 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1339 %shuffle = shufflevector <4 x i64> %v, <4 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1340 ret <4 x i64> %shuffle
1343 define <4 x double> @splat_mem_v4f64_2(double* %p) {
1344 ; ALL-LABEL: splat_mem_v4f64_2:
1346 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1348 %1 = load double, double* %p
1349 %2 = insertelement <2 x double> undef, double %1, i32 0
1350 %3 = shufflevector <2 x double> %2, <2 x double> undef, <4 x i32> zeroinitializer
1354 define <4 x double> @splat_v4f64(<2 x double> %r) {
1355 ; AVX1-LABEL: splat_v4f64:
1357 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
1358 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1361 ; AVX2-LABEL: splat_v4f64:
1363 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
1366 ; AVX512VL-LABEL: splat_v4f64:
1368 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
1369 ; AVX512VL-NEXT: retq
1370 %1 = shufflevector <2 x double> %r, <2 x double> undef, <4 x i32> zeroinitializer
1374 define <4 x i64> @splat_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1375 ; ALL-LABEL: splat_mem_v4i64_from_v2i64:
1377 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1379 %v = load <2 x i64>, <2 x i64>* %ptr
1380 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1381 ret <4 x i64> %shuffle
1384 define <4 x double> @splat_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1385 ; ALL-LABEL: splat_mem_v4f64_from_v2f64:
1387 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1389 %v = load <2 x double>, <2 x double>* %ptr
1390 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1391 ret <4 x double> %shuffle
1394 define <4 x i64> @splat128_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1395 ; AVX1-LABEL: splat128_mem_v4i64_from_v2i64:
1397 ; AVX1-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
1400 ; AVX2-LABEL: splat128_mem_v4i64_from_v2i64:
1402 ; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
1405 ; AVX512VL-LABEL: splat128_mem_v4i64_from_v2i64:
1407 ; AVX512VL-NEXT: vbroadcasti128 {{.*#+}} ymm0 = mem[0,1,0,1]
1408 ; AVX512VL-NEXT: retq
1409 %v = load <2 x i64>, <2 x i64>* %ptr
1410 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1411 ret <4 x i64> %shuffle
1414 define <4 x double> @splat128_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1415 ; ALL-LABEL: splat128_mem_v4f64_from_v2f64:
1417 ; ALL-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
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> @broadcast_v4f64_0000_from_v2i64(<2 x i64> %a0) {
1425 ; AVX1-LABEL: broadcast_v4f64_0000_from_v2i64:
1427 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
1428 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1431 ; AVX2-LABEL: broadcast_v4f64_0000_from_v2i64:
1433 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
1436 ; AVX512VL-LABEL: broadcast_v4f64_0000_from_v2i64:
1438 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
1439 ; AVX512VL-NEXT: retq
1440 %1 = shufflevector <2 x i64> %a0, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1441 %2 = bitcast <4 x i64> %1 to <4 x double>
1442 %3 = shufflevector <4 x double> %2, <4 x double> undef, <4 x i32> zeroinitializer
1446 define <4 x double> @bitcast_v4f64_0426(<4 x double> %a, <4 x double> %b) {
1447 ; AVX1-LABEL: bitcast_v4f64_0426:
1449 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1452 ; AVX2-LABEL: bitcast_v4f64_0426:
1454 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1457 ; AVX512VL-LABEL: bitcast_v4f64_0426:
1459 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1460 ; AVX512VL-NEXT: retq
1461 %shuffle64 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
1462 %bitcast32 = bitcast <4 x double> %shuffle64 to <8 x float>
1463 %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>
1464 %bitcast16 = bitcast <8 x float> %shuffle32 to <16 x i16>
1465 %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>
1466 %bitcast64 = bitcast <16 x i16> %shuffle16 to <4 x double>
1467 ret <4 x double> %bitcast64
1470 define <4 x i64> @concat_v4i64_0167(<4 x i64> %a0, <4 x i64> %a1) {
1471 ; AVX1-LABEL: concat_v4i64_0167:
1473 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1476 ; AVX2-LABEL: concat_v4i64_0167:
1478 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1481 ; AVX512VL-LABEL: concat_v4i64_0167:
1483 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1484 ; AVX512VL-NEXT: retq
1485 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1486 %a1hi = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 6, i32 7>
1487 %shuffle64 = shufflevector <2 x i64> %a0lo, <2 x i64> %a1hi, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1488 ret <4 x i64> %shuffle64
1491 define <4 x i64> @concat_v4i64_0145_bc(<4 x i64> %a0, <4 x i64> %a1) {
1492 ; AVX1-LABEL: concat_v4i64_0145_bc:
1494 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1497 ; AVX2-LABEL: concat_v4i64_0145_bc:
1499 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1502 ; AVX512VL-LABEL: concat_v4i64_0145_bc:
1504 ; AVX512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
1505 ; AVX512VL-NEXT: retq
1506 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1507 %a1lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 4, i32 5>
1508 %bc0lo = bitcast <2 x i64> %a0lo to <4 x i32>
1509 %bc1lo = bitcast <2 x i64> %a1lo to <4 x i32>
1510 %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>
1511 %shuffle64 = bitcast <8 x i32> %shuffle32 to <4 x i64>
1512 ret <4 x i64> %shuffle64
1515 define <4 x i64> @insert_dup_mem_v4i64(i64* %ptr) {
1516 ; ALL-LABEL: insert_dup_mem_v4i64:
1518 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1520 %tmp = load i64, i64* %ptr, align 1
1521 %tmp1 = insertelement <2 x i64> undef, i64 %tmp, i32 0
1522 %tmp2 = shufflevector <2 x i64> %tmp1, <2 x i64> undef, <4 x i32> zeroinitializer
1526 define <4 x i64> @shuffle_v4i64_1234(<4 x i64> %a, <4 x i64> %b) {
1527 ; AVX1-LABEL: shuffle_v4i64_1234:
1529 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
1530 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
1531 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm1[0],ymm0[3],ymm1[2]
1534 ; AVX2-LABEL: shuffle_v4i64_1234:
1536 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
1537 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,3,0]
1540 ; AVX512VL-LABEL: shuffle_v4i64_1234:
1542 ; AVX512VL-NEXT: valignq {{.*#+}} ymm0 = ymm0[1,2,3],ymm1[0]
1543 ; AVX512VL-NEXT: retq
1544 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 3, i32 4>
1545 ret <4 x i64> %shuffle
1548 define <4 x i64> @shuffle_v4i64_1230(<4 x i64> %a) {
1549 ; AVX1-LABEL: shuffle_v4i64_1230:
1551 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
1552 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm1[0],ymm0[3],ymm1[2]
1555 ; AVX2-LABEL: shuffle_v4i64_1230:
1557 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,3,0]
1560 ; AVX512VL-LABEL: shuffle_v4i64_1230:
1562 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,3,0]
1563 ; AVX512VL-NEXT: retq
1564 %shuffle = shufflevector <4 x i64> %a, <4 x i64> undef, <4 x i32> <i32 1, i32 2, i32 3, i32 0>
1565 ret <4 x i64> %shuffle
1568 define <4 x i64> @shuffle_v4i64_z0z3(<4 x i64> %a, <4 x i64> %b) {
1569 ; AVX1-LABEL: shuffle_v4i64_z0z3:
1571 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,3]
1572 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1573 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
1576 ; AVX2-LABEL: shuffle_v4i64_z0z3:
1578 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,3]
1579 ; AVX2-NEXT: vpxor %ymm1, %ymm1, %ymm1
1580 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
1583 ; AVX512VL-LABEL: shuffle_v4i64_z0z3:
1585 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,3]
1586 ; AVX512VL-NEXT: vpxor %ymm1, %ymm1, %ymm1
1587 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
1588 ; AVX512VL-NEXT: retq
1589 %1 = shufflevector <4 x i64> %a, <4 x i64> <i64 0, i64 undef, i64 undef, i64 undef>, <4 x i32> <i32 4, i32 0, i32 4, i32 3>
1593 define <4 x i64> @shuffle_v4i64_1z2z(<4 x i64> %a, <4 x i64> %b) {
1594 ; AVX1-LABEL: shuffle_v4i64_1z2z:
1596 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1597 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
1598 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
1599 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
1600 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
1601 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1604 ; AVX2-LABEL: shuffle_v4i64_1z2z:
1606 ; AVX2-NEXT: vpxor %ymm1, %ymm1, %ymm1
1607 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
1608 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,2,0]
1611 ; AVX512VL-LABEL: shuffle_v4i64_1z2z:
1613 ; AVX512VL-NEXT: vpxor %ymm1, %ymm1, %ymm1
1614 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
1615 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,2,0]
1616 ; AVX512VL-NEXT: retq
1617 %1 = shufflevector <4 x i64> %a, <4 x i64> <i64 0, i64 undef, i64 undef, i64 undef>, <4 x i32> <i32 1, i32 4, i32 2, i32 4>