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_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: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
157 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
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 ; ALL-LABEL: shuffle_v4f64_0423:
241 ; ALL-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
242 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
244 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 3>
245 ret <4 x double> %shuffle
248 define <4 x double> @shuffle_v4f64_0462(<4 x double> %a, <4 x double> %b) {
249 ; ALL-LABEL: shuffle_v4f64_0462:
251 ; ALL-NEXT: vmovddup {{.*#+}} ymm1 = ymm1[0,0,2,2]
252 ; ALL-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
253 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
255 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 6, i32 2>
256 ret <4 x double> %shuffle
259 define <4 x double> @shuffle_v4f64_0426(<4 x double> %a, <4 x double> %b) {
260 ; ALL-LABEL: shuffle_v4f64_0426:
262 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
264 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
265 ret <4 x double> %shuffle
268 define <4 x double> @shuffle_v4f64_1537(<4 x double> %a, <4 x double> %b) {
269 ; ALL-LABEL: shuffle_v4f64_1537:
271 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
273 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
274 ret <4 x double> %shuffle
277 define <4 x double> @shuffle_v4f64_4062(<4 x double> %a, <4 x double> %b) {
278 ; ALL-LABEL: shuffle_v4f64_4062:
280 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
282 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
283 ret <4 x double> %shuffle
286 define <4 x double> @shuffle_v4f64_5173(<4 x double> %a, <4 x double> %b) {
287 ; ALL-LABEL: shuffle_v4f64_5173:
289 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
291 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 7, i32 3>
292 ret <4 x double> %shuffle
295 define <4 x double> @shuffle_v4f64_5163(<4 x double> %a, <4 x double> %b) {
296 ; ALL-LABEL: shuffle_v4f64_5163:
298 ; ALL-NEXT: vshufpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
300 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
301 ret <4 x double> %shuffle
304 define <4 x double> @shuffle_v4f64_0527(<4 x double> %a, <4 x double> %b) {
305 ; ALL-LABEL: shuffle_v4f64_0527:
307 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
309 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
310 ret <4 x double> %shuffle
313 define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
314 ; ALL-LABEL: shuffle_v4f64_4163:
316 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
318 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
319 ret <4 x double> %shuffle
322 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
323 ; AVX1-LABEL: shuffle_v4f64_0145:
325 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
328 ; AVX2-LABEL: shuffle_v4f64_0145:
330 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
333 ; AVX512VL-LABEL: shuffle_v4f64_0145:
335 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
336 ; AVX512VL-NEXT: retq
337 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
338 ret <4 x double> %shuffle
341 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
342 ; AVX1-LABEL: shuffle_v4f64_4501:
344 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
347 ; AVX2-LABEL: shuffle_v4f64_4501:
349 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
352 ; AVX512VL-LABEL: shuffle_v4f64_4501:
354 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm0, %ymm1, %ymm0
355 ; AVX512VL-NEXT: retq
356 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
357 ret <4 x double> %shuffle
360 define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
361 ; ALL-LABEL: shuffle_v4f64_0167:
363 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
365 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
366 ret <4 x double> %shuffle
369 define <4 x double> @shuffle_v4f64_1054(<4 x double> %a, <4 x double> %b) {
370 ; AVX1-LABEL: shuffle_v4f64_1054:
372 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
373 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
376 ; AVX2-LABEL: shuffle_v4f64_1054:
378 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
379 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
382 ; AVX512VL-LABEL: shuffle_v4f64_1054:
384 ; AVX512VL-NEXT: vinsertf32x4 $1, %xmm1, %ymm0, %ymm0
385 ; AVX512VL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
386 ; AVX512VL-NEXT: retq
387 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
388 ret <4 x double> %shuffle
391 define <4 x double> @shuffle_v4f64_3254(<4 x double> %a, <4 x double> %b) {
392 ; ALL-LABEL: shuffle_v4f64_3254:
394 ; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
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 3, i32 2, i32 5, i32 4>
398 ret <4 x double> %shuffle
401 define <4 x double> @shuffle_v4f64_3276(<4 x double> %a, <4 x double> %b) {
402 ; ALL-LABEL: shuffle_v4f64_3276:
404 ; ALL-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
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 7, i32 6>
408 ret <4 x double> %shuffle
411 define <4 x double> @shuffle_v4f64_1076(<4 x double> %a, <4 x double> %b) {
412 ; ALL-LABEL: shuffle_v4f64_1076:
414 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],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 1, i32 0, i32 7, i32 6>
418 ret <4 x double> %shuffle
421 define <4 x double> @shuffle_v4f64_0415(<4 x double> %a, <4 x double> %b) {
422 ; AVX1-LABEL: shuffle_v4f64_0415:
424 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
425 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
426 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
429 ; AVX2-LABEL: shuffle_v4f64_0415:
431 ; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,0,2,1]
432 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
433 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
436 ; AVX512VL-LABEL: shuffle_v4f64_0415:
438 ; AVX512VL-NEXT: vmovapd {{.*#+}} ymm2 = [0,4,1,5]
439 ; AVX512VL-NEXT: vpermt2pd %ymm1, %ymm2, %ymm0
440 ; AVX512VL-NEXT: retq
441 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
442 ret <4 x double> %shuffle
445 define <4 x double> @shuffle_v4f64_u062(<4 x double> %a, <4 x double> %b) {
446 ; ALL-LABEL: shuffle_v4f64_u062:
448 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
450 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 undef, i32 0, i32 6, i32 2>
451 ret <4 x double> %shuffle
454 define <4 x double> @shuffle_v4f64_15uu(<4 x double> %a, <4 x double> %b) {
455 ; ALL-LABEL: shuffle_v4f64_15uu:
457 ; ALL-NEXT: vunpckhpd {{.*#+}} xmm0 = xmm0[1],xmm1[1]
459 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 undef, i32 undef>
460 ret <4 x double> %shuffle
463 define <4 x double> @shuffle_v4f64_11uu(<4 x double> %a, <4 x double> %b) {
464 ; ALL-LABEL: shuffle_v4f64_11uu:
466 ; ALL-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,1]
468 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 undef, i32 undef>
469 ret <4 x double> %shuffle
472 define <4 x double> @shuffle_v4f64_22uu(<4 x double> %a, <4 x double> %b) {
473 ; AVX1-LABEL: shuffle_v4f64_22uu:
475 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
476 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
479 ; AVX2-LABEL: shuffle_v4f64_22uu:
481 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,3]
484 ; AVX512VL-LABEL: shuffle_v4f64_22uu:
486 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,2,3]
487 ; AVX512VL-NEXT: retq
488 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 undef, i32 undef>
489 ret <4 x double> %shuffle
492 define <4 x double> @shuffle_v4f64_3333(<4 x double> %a, <4 x double> %b) {
493 ; AVX1-LABEL: shuffle_v4f64_3333:
495 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
496 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
499 ; AVX2-LABEL: shuffle_v4f64_3333:
501 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,3]
504 ; AVX512VL-LABEL: shuffle_v4f64_3333:
506 ; AVX512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,3]
507 ; AVX512VL-NEXT: retq
508 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
509 ret <4 x double> %shuffle
512 define <4 x i64> @shuffle_v4i64_0000(<4 x i64> %a, <4 x i64> %b) {
513 ; AVX1-LABEL: shuffle_v4i64_0000:
515 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
516 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
519 ; AVX2-LABEL: shuffle_v4i64_0000:
521 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
524 ; AVX512VL-LABEL: shuffle_v4i64_0000:
526 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
527 ; AVX512VL-NEXT: retq
528 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
529 ret <4 x i64> %shuffle
532 define <4 x i64> @shuffle_v4i64_0001(<4 x i64> %a, <4 x i64> %b) {
533 ; AVX1-LABEL: shuffle_v4i64_0001:
535 ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[0,1,0,1]
536 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
539 ; AVX2-LABEL: shuffle_v4i64_0001:
541 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
544 ; AVX512VL-LABEL: shuffle_v4i64_0001:
546 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
547 ; AVX512VL-NEXT: retq
548 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
549 ret <4 x i64> %shuffle
552 define <4 x i64> @shuffle_v4i64_0020(<4 x i64> %a, <4 x i64> %b) {
553 ; AVX1-LABEL: shuffle_v4i64_0020:
555 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
556 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
557 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
558 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
561 ; AVX2-LABEL: shuffle_v4i64_0020:
563 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
566 ; AVX512VL-LABEL: shuffle_v4i64_0020:
568 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
569 ; AVX512VL-NEXT: retq
570 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
571 ret <4 x i64> %shuffle
574 define <4 x i64> @shuffle_v4i64_0112(<4 x i64> %a, <4 x i64> %b) {
575 ; AVX1-LABEL: shuffle_v4i64_0112:
577 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
578 ; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm0[8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7]
579 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
582 ; AVX2-LABEL: shuffle_v4i64_0112:
584 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
587 ; AVX512VL-LABEL: shuffle_v4i64_0112:
589 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
590 ; AVX512VL-NEXT: retq
591 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 1, i32 2>
592 ret <4 x i64> %shuffle
595 define <4 x i64> @shuffle_v4i64_0300(<4 x i64> %a, <4 x i64> %b) {
596 ; AVX1-LABEL: shuffle_v4i64_0300:
598 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
599 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
600 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
603 ; AVX2-LABEL: shuffle_v4i64_0300:
605 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
608 ; AVX512VL-LABEL: shuffle_v4i64_0300:
610 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
611 ; AVX512VL-NEXT: retq
612 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
613 ret <4 x i64> %shuffle
616 define <4 x i64> @shuffle_v4i64_1000(<4 x i64> %a, <4 x i64> %b) {
617 ; AVX1-LABEL: shuffle_v4i64_1000:
619 ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
620 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
621 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
624 ; AVX2-LABEL: shuffle_v4i64_1000:
626 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
629 ; AVX512VL-LABEL: shuffle_v4i64_1000:
631 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
632 ; AVX512VL-NEXT: retq
633 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
634 ret <4 x i64> %shuffle
637 define <4 x i64> @shuffle_v4i64_2200(<4 x i64> %a, <4 x i64> %b) {
638 ; AVX1-LABEL: shuffle_v4i64_2200:
640 ; AVX1-NEXT: vmovddup {{.*#+}} ymm0 = ymm0[0,0,2,2]
641 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
644 ; AVX2-LABEL: shuffle_v4i64_2200:
646 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
649 ; AVX512VL-LABEL: shuffle_v4i64_2200:
651 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
652 ; AVX512VL-NEXT: retq
653 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
654 ret <4 x i64> %shuffle
657 define <4 x i64> @shuffle_v4i64_3330(<4 x i64> %a, <4 x i64> %b) {
658 ; AVX1-LABEL: shuffle_v4i64_3330:
660 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
661 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3]
662 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
665 ; AVX2-LABEL: shuffle_v4i64_3330:
667 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
670 ; AVX512VL-LABEL: shuffle_v4i64_3330:
672 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
673 ; AVX512VL-NEXT: retq
674 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
675 ret <4 x i64> %shuffle
678 define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
679 ; AVX1-LABEL: shuffle_v4i64_3210:
681 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
682 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
685 ; AVX2-LABEL: shuffle_v4i64_3210:
687 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
690 ; AVX512VL-LABEL: shuffle_v4i64_3210:
692 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
693 ; AVX512VL-NEXT: retq
694 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
695 ret <4 x i64> %shuffle
698 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
699 ; AVX1-LABEL: shuffle_v4i64_0124:
701 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
702 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
703 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3]
706 ; AVX2-LABEL: shuffle_v4i64_0124:
708 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
709 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
712 ; AVX512VL-LABEL: shuffle_v4i64_0124:
714 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %ymm1
715 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
716 ; AVX512VL-NEXT: retq
717 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
718 ret <4 x i64> %shuffle
721 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
722 ; AVX1-LABEL: shuffle_v4i64_0142:
724 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
725 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,1,2,2]
726 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
729 ; AVX2-LABEL: shuffle_v4i64_0142:
731 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1
732 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
733 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
736 ; AVX512VL-LABEL: shuffle_v4i64_0142:
738 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm1
739 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
740 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
741 ; AVX512VL-NEXT: retq
742 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
743 ret <4 x i64> %shuffle
746 define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
747 ; AVX1-LABEL: shuffle_v4i64_0412:
749 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
750 ; AVX1-NEXT: vpalignr {{.*#+}} xmm2 = xmm0[8,9,10,11,12,13,14,15],xmm2[0,1,2,3,4,5,6,7]
751 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
752 ; AVX1-NEXT: vmovddup {{.*#+}} xmm1 = xmm1[0,0]
753 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
756 ; AVX2-LABEL: shuffle_v4i64_0412:
758 ; AVX2-NEXT: vpbroadcastq %xmm1, %xmm1
759 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
760 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
763 ; AVX512VL-LABEL: shuffle_v4i64_0412:
765 ; AVX512VL-NEXT: vpbroadcastq %xmm1, %xmm1
766 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
767 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
768 ; AVX512VL-NEXT: retq
769 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
770 ret <4 x i64> %shuffle
773 define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
774 ; AVX1-LABEL: shuffle_v4i64_4012:
776 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
777 ; AVX1-NEXT: vpalignr {{.*#+}} xmm2 = xmm0[8,9,10,11,12,13,14,15],xmm2[0,1,2,3,4,5,6,7]
778 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
779 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
780 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
783 ; AVX2-LABEL: shuffle_v4i64_4012:
785 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
786 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
789 ; AVX512VL-LABEL: shuffle_v4i64_4012:
791 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
792 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
793 ; AVX512VL-NEXT: retq
794 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
795 ret <4 x i64> %shuffle
798 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
799 ; AVX1-LABEL: shuffle_v4i64_0145:
801 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
804 ; AVX2-LABEL: shuffle_v4i64_0145:
806 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
809 ; AVX512VL-LABEL: shuffle_v4i64_0145:
811 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
812 ; AVX512VL-NEXT: retq
813 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
814 ret <4 x i64> %shuffle
817 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
818 ; AVX1-LABEL: shuffle_v4i64_0451:
820 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm1[1],xmm0[1]
821 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
822 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
825 ; AVX2-LABEL: shuffle_v4i64_0451:
827 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
828 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
829 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
832 ; AVX512VL-LABEL: shuffle_v4i64_0451:
834 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
835 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
836 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
837 ; AVX512VL-NEXT: retq
838 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
839 ret <4 x i64> %shuffle
842 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
843 ; AVX1-LABEL: shuffle_v4i64_4501:
845 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
848 ; AVX2-LABEL: shuffle_v4i64_4501:
850 ; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
853 ; AVX512VL-LABEL: shuffle_v4i64_4501:
855 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm0, %ymm1, %ymm0
856 ; AVX512VL-NEXT: retq
857 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
858 ret <4 x i64> %shuffle
861 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
862 ; AVX1-LABEL: shuffle_v4i64_4015:
864 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
865 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
866 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
869 ; AVX2-LABEL: shuffle_v4i64_4015:
871 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,2,1]
872 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
873 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
876 ; AVX512VL-LABEL: shuffle_v4i64_4015:
878 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,2,1]
879 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
880 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
881 ; AVX512VL-NEXT: retq
882 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
883 ret <4 x i64> %shuffle
886 define <4 x i64> @shuffle_v4i64_2u35(<4 x i64> %a, <4 x i64> %b) {
887 ; AVX1-LABEL: shuffle_v4i64_2u35:
889 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
890 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm1 = xmm0[1],xmm1[1]
891 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
894 ; AVX2-LABEL: shuffle_v4i64_2u35:
896 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
897 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
900 ; AVX512VL-LABEL: shuffle_v4i64_2u35:
902 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
903 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,1]
904 ; AVX512VL-NEXT: retq
905 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 undef, i32 3, i32 5>
906 ret <4 x i64> %shuffle
909 define <4 x i64> @shuffle_v4i64_1251(<4 x i64> %a, <4 x i64> %b) {
910 ; AVX1-LABEL: shuffle_v4i64_1251:
912 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
913 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm2[0],ymm0[2],ymm2[3]
914 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
915 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
916 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
919 ; AVX2-LABEL: shuffle_v4i64_1251:
921 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
922 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
923 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
926 ; AVX512VL-LABEL: shuffle_v4i64_1251:
928 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
929 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
930 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
931 ; AVX512VL-NEXT: retq
932 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 5, i32 1>
933 ret <4 x i64> %shuffle
936 define <4 x i64> @shuffle_v4i64_1054(<4 x i64> %a, <4 x i64> %b) {
937 ; AVX1-LABEL: shuffle_v4i64_1054:
939 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
940 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
943 ; AVX2-LABEL: shuffle_v4i64_1054:
945 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
946 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
949 ; AVX512VL-LABEL: shuffle_v4i64_1054:
951 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
952 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
953 ; AVX512VL-NEXT: retq
954 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
955 ret <4 x i64> %shuffle
958 define <4 x i64> @shuffle_v4i64_3254(<4 x i64> %a, <4 x i64> %b) {
959 ; AVX1-LABEL: shuffle_v4i64_3254:
961 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
962 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
965 ; AVX2-LABEL: shuffle_v4i64_3254:
967 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
968 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
971 ; AVX512VL-LABEL: shuffle_v4i64_3254:
973 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
974 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
975 ; AVX512VL-NEXT: retq
976 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
977 ret <4 x i64> %shuffle
980 define <4 x i64> @shuffle_v4i64_3276(<4 x i64> %a, <4 x i64> %b) {
981 ; AVX1-LABEL: shuffle_v4i64_3276:
983 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
984 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
987 ; AVX2-LABEL: shuffle_v4i64_3276:
989 ; AVX2-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
990 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
993 ; AVX512VL-LABEL: shuffle_v4i64_3276:
995 ; AVX512VL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
996 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
997 ; AVX512VL-NEXT: retq
998 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
999 ret <4 x i64> %shuffle
1002 define <4 x i64> @shuffle_v4i64_1076(<4 x i64> %a, <4 x i64> %b) {
1003 ; AVX1-LABEL: shuffle_v4i64_1076:
1005 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1006 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
1009 ; AVX2-LABEL: shuffle_v4i64_1076:
1011 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1012 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1015 ; AVX512VL-LABEL: shuffle_v4i64_1076:
1017 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1018 ; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
1019 ; AVX512VL-NEXT: retq
1020 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
1021 ret <4 x i64> %shuffle
1024 define <4 x i64> @shuffle_v4i64_0415(<4 x i64> %a, <4 x i64> %b) {
1025 ; AVX1-LABEL: shuffle_v4i64_0415:
1027 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
1028 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
1029 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
1032 ; AVX2-LABEL: shuffle_v4i64_0415:
1034 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1035 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1036 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1039 ; AVX512VL-LABEL: shuffle_v4i64_0415:
1041 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,2,1]
1042 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
1043 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
1044 ; AVX512VL-NEXT: retq
1045 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
1046 ret <4 x i64> %shuffle
1049 define <4 x i64> @shuffle_v4i64_z4z6(<4 x i64> %a) {
1050 ; AVX1-LABEL: shuffle_v4i64_z4z6:
1052 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1053 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1056 ; AVX2-LABEL: shuffle_v4i64_z4z6:
1058 ; 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]
1061 ; AVX512VL-LABEL: shuffle_v4i64_z4z6:
1063 ; 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]
1064 ; AVX512VL-NEXT: retq
1065 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 0, i32 4, i32 0, i32 6>
1066 ret <4 x i64> %shuffle
1069 define <4 x i64> @shuffle_v4i64_5zuz(<4 x i64> %a) {
1070 ; AVX1-LABEL: shuffle_v4i64_5zuz:
1072 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1073 ; AVX1-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
1076 ; AVX2-LABEL: shuffle_v4i64_5zuz:
1078 ; 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
1081 ; AVX512VL-LABEL: shuffle_v4i64_5zuz:
1083 ; 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
1084 ; AVX512VL-NEXT: retq
1085 %shuffle = shufflevector <4 x i64> zeroinitializer, <4 x i64> %a, <4 x i32> <i32 5, i32 0, i32 undef, i32 0>
1086 ret <4 x i64> %shuffle
1089 define <4 x i64> @shuffle_v4i64_40u2(<4 x i64> %a, <4 x i64> %b) {
1090 ; AVX1-LABEL: shuffle_v4i64_40u2:
1092 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1095 ; AVX2-LABEL: shuffle_v4i64_40u2:
1097 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1100 ; AVX512VL-LABEL: shuffle_v4i64_40u2:
1102 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
1103 ; AVX512VL-NEXT: retq
1104 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 undef, i32 2>
1105 ret <4 x i64> %shuffle
1108 define <4 x i64> @shuffle_v4i64_15uu(<4 x i64> %a, <4 x i64> %b) {
1109 ; ALL-LABEL: shuffle_v4i64_15uu:
1111 ; ALL-NEXT: vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
1113 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 5, i32 undef, i32 undef>
1114 ret <4 x i64> %shuffle
1117 define <4 x i64> @shuffle_v4i64_11uu(<4 x i64> %a, <4 x i64> %b) {
1118 ; ALL-LABEL: shuffle_v4i64_11uu:
1120 ; ALL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
1122 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 1, i32 undef, i32 undef>
1123 ret <4 x i64> %shuffle
1126 define <4 x i64> @shuffle_v4i64_22uu(<4 x i64> %a, <4 x i64> %b) {
1127 ; AVX1-LABEL: shuffle_v4i64_22uu:
1129 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
1130 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
1133 ; AVX2-LABEL: shuffle_v4i64_22uu:
1135 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,3]
1138 ; AVX512VL-LABEL: shuffle_v4i64_22uu:
1140 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,3]
1141 ; AVX512VL-NEXT: retq
1142 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 undef, i32 undef>
1143 ret <4 x i64> %shuffle
1146 define <4 x i64> @shuffle_v4i64_3333(<4 x i64> %a, <4 x i64> %b) {
1147 ; AVX1-LABEL: shuffle_v4i64_3333:
1149 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
1150 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
1153 ; AVX2-LABEL: shuffle_v4i64_3333:
1155 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,3]
1158 ; AVX512VL-LABEL: shuffle_v4i64_3333:
1160 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,3]
1161 ; AVX512VL-NEXT: retq
1162 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
1163 ret <4 x i64> %shuffle
1166 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
1167 ; ALL-LABEL: stress_test1:
1169 %c = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 1, i32 1, i32 0>
1170 %d = shufflevector <4 x i64> %c, <4 x i64> undef, <4 x i32> <i32 3, i32 undef, i32 2, i32 undef>
1171 %e = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 undef>
1172 %f = shufflevector <4 x i64> %d, <4 x i64> %e, <4 x i32> <i32 5, i32 1, i32 1, i32 0>
1177 define <4 x i64> @insert_reg_and_zero_v4i64(i64 %a) {
1178 ; ALL-LABEL: insert_reg_and_zero_v4i64:
1180 ; ALL-NEXT: vmovq %rdi, %xmm0
1182 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1183 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1184 ret <4 x i64> %shuffle
1187 define <4 x i64> @insert_mem_and_zero_v4i64(i64* %ptr) {
1188 ; ALL-LABEL: insert_mem_and_zero_v4i64:
1190 ; ALL-NEXT: vmovq {{.*#+}} xmm0 = mem[0],zero
1192 %a = load i64, i64* %ptr
1193 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1194 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1195 ret <4 x i64> %shuffle
1198 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
1199 ; AVX1-LABEL: insert_reg_and_zero_v4f64:
1201 ; AVX1-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def>
1202 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1203 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1206 ; AVX2-LABEL: insert_reg_and_zero_v4f64:
1208 ; AVX2-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def>
1209 ; AVX2-NEXT: vxorpd %ymm1, %ymm1, %ymm1
1210 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
1213 ; AVX512VL-LABEL: insert_reg_and_zero_v4f64:
1215 ; AVX512VL-NEXT: vxorpd %xmm1, %xmm1, %xmm1
1216 ; AVX512VL-NEXT: vmovsd {{.*#+}} xmm0 = xmm0[0],xmm1[1]
1217 ; AVX512VL-NEXT: retq
1218 %v = insertelement <4 x double> undef, double %a, i32 0
1219 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1220 ret <4 x double> %shuffle
1223 define <4 x double> @insert_mem_and_zero_v4f64(double* %ptr) {
1224 ; ALL-LABEL: insert_mem_and_zero_v4f64:
1226 ; ALL-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
1228 %a = load double, double* %ptr
1229 %v = insertelement <4 x double> undef, double %a, i32 0
1230 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
1231 ret <4 x double> %shuffle
1234 define <4 x double> @splat_mem_v4f64(double* %ptr) {
1235 ; ALL-LABEL: splat_mem_v4f64:
1237 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1239 %a = load double, double* %ptr
1240 %v = insertelement <4 x double> undef, double %a, i32 0
1241 %shuffle = shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1242 ret <4 x double> %shuffle
1245 define <4 x i64> @splat_mem_v4i64(i64* %ptr) {
1246 ; ALL-LABEL: splat_mem_v4i64:
1248 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1250 %a = load i64, i64* %ptr
1251 %v = insertelement <4 x i64> undef, i64 %a, i64 0
1252 %shuffle = shufflevector <4 x i64> %v, <4 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1253 ret <4 x i64> %shuffle
1256 define <4 x double> @splat_mem_v4f64_2(double* %p) {
1257 ; ALL-LABEL: splat_mem_v4f64_2:
1259 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1261 %1 = load double, double* %p
1262 %2 = insertelement <2 x double> undef, double %1, i32 0
1263 %3 = shufflevector <2 x double> %2, <2 x double> undef, <4 x i32> zeroinitializer
1267 define <4 x double> @splat_v4f64(<2 x double> %r) {
1268 ; AVX1-LABEL: splat_v4f64:
1270 ; AVX1-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
1271 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1274 ; AVX2-LABEL: splat_v4f64:
1276 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
1279 ; AVX512VL-LABEL: splat_v4f64:
1281 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
1282 ; AVX512VL-NEXT: retq
1283 %1 = shufflevector <2 x double> %r, <2 x double> undef, <4 x i32> zeroinitializer
1287 define <4 x i64> @splat_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1288 ; ALL-LABEL: splat_mem_v4i64_from_v2i64:
1290 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1292 %v = load <2 x i64>, <2 x i64>* %ptr
1293 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1294 ret <4 x i64> %shuffle
1297 define <4 x double> @splat_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1298 ; ALL-LABEL: splat_mem_v4f64_from_v2f64:
1300 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1302 %v = load <2 x double>, <2 x double>* %ptr
1303 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
1304 ret <4 x double> %shuffle
1307 define <4 x i64> @splat128_mem_v4i64_from_v2i64(<2 x i64>* %ptr) {
1308 ; AVX1-LABEL: splat128_mem_v4i64_from_v2i64:
1310 ; AVX1-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
1313 ; AVX2-LABEL: splat128_mem_v4i64_from_v2i64:
1315 ; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
1318 ; AVX512VL-LABEL: splat128_mem_v4i64_from_v2i64:
1320 ; AVX512VL-NEXT: vbroadcasti32x4 {{.*#+}} ymm0 = mem[0,1,2,3,0,1,2,3]
1321 ; AVX512VL-NEXT: retq
1322 %v = load <2 x i64>, <2 x i64>* %ptr
1323 %shuffle = shufflevector <2 x i64> %v, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1324 ret <4 x i64> %shuffle
1327 define <4 x double> @splat128_mem_v4f64_from_v2f64(<2 x double>* %ptr) {
1328 ; AVX1-LABEL: splat128_mem_v4f64_from_v2f64:
1330 ; AVX1-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
1333 ; AVX2-LABEL: splat128_mem_v4f64_from_v2f64:
1335 ; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
1338 ; AVX512VL-LABEL: splat128_mem_v4f64_from_v2f64:
1340 ; AVX512VL-NEXT: vbroadcastf32x4 {{.*#+}} ymm0 = mem[0,1,2,3,0,1,2,3]
1341 ; AVX512VL-NEXT: retq
1342 %v = load <2 x double>, <2 x double>* %ptr
1343 %shuffle = shufflevector <2 x double> %v, <2 x double> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
1344 ret <4 x double> %shuffle
1347 define <4 x double> @broadcast_v4f64_0000_from_v2i64(<2 x i64> %a0) {
1348 ; AVX1-LABEL: broadcast_v4f64_0000_from_v2i64:
1350 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
1351 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
1354 ; AVX2-LABEL: broadcast_v4f64_0000_from_v2i64:
1356 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
1359 ; AVX512VL-LABEL: broadcast_v4f64_0000_from_v2i64:
1361 ; AVX512VL-NEXT: vbroadcastsd %xmm0, %ymm0
1362 ; AVX512VL-NEXT: retq
1363 %1 = shufflevector <2 x i64> %a0, <2 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1364 %2 = bitcast <4 x i64> %1 to <4 x double>
1365 %3 = shufflevector <4 x double> %2, <4 x double> undef, <4 x i32> zeroinitializer
1369 define <4 x double> @bitcast_v4f64_0426(<4 x double> %a, <4 x double> %b) {
1370 ; AVX1-LABEL: bitcast_v4f64_0426:
1372 ; AVX1-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1375 ; AVX2-LABEL: bitcast_v4f64_0426:
1377 ; AVX2-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1380 ; AVX512VL-LABEL: bitcast_v4f64_0426:
1382 ; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
1383 ; AVX512VL-NEXT: retq
1384 %shuffle64 = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
1385 %bitcast32 = bitcast <4 x double> %shuffle64 to <8 x float>
1386 %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>
1387 %bitcast16 = bitcast <8 x float> %shuffle32 to <16 x i16>
1388 %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>
1389 %bitcast64 = bitcast <16 x i16> %shuffle16 to <4 x double>
1390 ret <4 x double> %bitcast64
1393 define <4 x i64> @concat_v4i64_0167(<4 x i64> %a0, <4 x i64> %a1) {
1394 ; AVX1-LABEL: concat_v4i64_0167:
1396 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
1399 ; AVX2-LABEL: concat_v4i64_0167:
1401 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1404 ; AVX512VL-LABEL: concat_v4i64_0167:
1406 ; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
1407 ; AVX512VL-NEXT: retq
1408 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1409 %a1hi = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 6, i32 7>
1410 %shuffle64 = shufflevector <2 x i64> %a0lo, <2 x i64> %a1hi, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1411 ret <4 x i64> %shuffle64
1414 define <4 x i64> @concat_v4i64_0145_bc(<4 x i64> %a0, <4 x i64> %a1) {
1415 ; AVX1-LABEL: concat_v4i64_0145_bc:
1417 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1420 ; AVX2-LABEL: concat_v4i64_0145_bc:
1422 ; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
1425 ; AVX512VL-LABEL: concat_v4i64_0145_bc:
1427 ; AVX512VL-NEXT: vinserti32x4 $1, %xmm1, %ymm0, %ymm0
1428 ; AVX512VL-NEXT: retq
1429 %a0lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 0, i32 1>
1430 %a1lo = shufflevector <4 x i64> %a0, <4 x i64> %a1, <2 x i32> <i32 4, i32 5>
1431 %bc0lo = bitcast <2 x i64> %a0lo to <4 x i32>
1432 %bc1lo = bitcast <2 x i64> %a1lo to <4 x i32>
1433 %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>
1434 %shuffle64 = bitcast <8 x i32> %shuffle32 to <4 x i64>
1435 ret <4 x i64> %shuffle64
1438 define <4 x i64> @insert_dup_mem_v4i64(i64* %ptr) {
1439 ; ALL-LABEL: insert_dup_mem_v4i64:
1441 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
1443 %tmp = load i64, i64* %ptr, align 1
1444 %tmp1 = insertelement <2 x i64> undef, i64 %tmp, i32 0
1445 %tmp2 = shufflevector <2 x i64> %tmp1, <2 x i64> undef, <4 x i32> zeroinitializer
1449 define <4 x i64> @shuffle_v4i64_1234(<4 x i64> %a, <4 x i64> %b) {
1450 ; AVX1-LABEL: shuffle_v4i64_1234:
1452 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
1453 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
1454 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm1[0],ymm0[3],ymm1[2]
1457 ; AVX2-LABEL: shuffle_v4i64_1234:
1459 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
1460 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,3,0]
1463 ; AVX512VL-LABEL: shuffle_v4i64_1234:
1465 ; AVX512VL-NEXT: valignq {{.*#+}} ymm0 = ymm0[1,2,3],ymm1[0]
1466 ; AVX512VL-NEXT: retq
1467 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 3, i32 4>
1468 ret <4 x i64> %shuffle
1471 define <4 x i64> @shuffle_v4i64_1230(<4 x i64> %a) {
1472 ; AVX1-LABEL: shuffle_v4i64_1230:
1474 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
1475 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm1[0],ymm0[3],ymm1[2]
1478 ; AVX2-LABEL: shuffle_v4i64_1230:
1480 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,3,0]
1483 ; AVX512VL-LABEL: shuffle_v4i64_1230:
1485 ; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,3,0]
1486 ; AVX512VL-NEXT: retq
1487 %shuffle = shufflevector <4 x i64> %a, <4 x i64> undef, <4 x i32> <i32 1, i32 2, i32 3, i32 0>
1488 ret <4 x i64> %shuffle