1 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx -x86-experimental-vector-shuffle-lowering | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1
2 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx2 -x86-experimental-vector-shuffle-lowering | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
4 target triple = "x86_64-unknown-unknown"
6 define <4 x double> @shuffle_v4f64_0000(<4 x double> %a, <4 x double> %b) {
7 ; AVX1-LABEL: shuffle_v4f64_0000:
9 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
10 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
13 ; AVX2-LABEL: shuffle_v4f64_0000:
15 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
17 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
18 ret <4 x double> %shuffle
21 define <4 x double> @shuffle_v4f64_0001(<4 x double> %a, <4 x double> %b) {
22 ; AVX1-LABEL: shuffle_v4f64_0001:
24 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm0[0,0]
25 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
28 ; AVX2-LABEL: shuffle_v4f64_0001:
30 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,0,0,1]
32 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
33 ret <4 x double> %shuffle
36 define <4 x double> @shuffle_v4f64_0020(<4 x double> %a, <4 x double> %b) {
37 ; AVX1-LABEL: shuffle_v4f64_0020:
39 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
40 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
41 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
42 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
45 ; AVX2-LABEL: shuffle_v4f64_0020:
47 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,0,2,0]
49 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
50 ret <4 x double> %shuffle
53 define <4 x double> @shuffle_v4f64_0300(<4 x double> %a, <4 x double> %b) {
54 ; AVX1-LABEL: shuffle_v4f64_0300:
56 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
57 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
58 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
61 ; AVX2-LABEL: shuffle_v4f64_0300:
63 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,0,0]
65 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
66 ret <4 x double> %shuffle
69 define <4 x double> @shuffle_v4f64_1000(<4 x double> %a, <4 x double> %b) {
70 ; AVX1-LABEL: shuffle_v4f64_1000:
72 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
73 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
74 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
77 ; AVX2-LABEL: shuffle_v4f64_1000:
79 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,0,0]
81 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
82 ret <4 x double> %shuffle
85 define <4 x double> @shuffle_v4f64_2200(<4 x double> %a, <4 x double> %b) {
86 ; AVX1-LABEL: shuffle_v4f64_2200:
88 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
89 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,2]
92 ; AVX2-LABEL: shuffle_v4f64_2200:
94 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,2,0,0]
96 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
97 ret <4 x double> %shuffle
100 define <4 x double> @shuffle_v4f64_3330(<4 x double> %a, <4 x double> %b) {
101 ; AVX1-LABEL: shuffle_v4f64_3330:
103 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
104 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[1,1,2,2]
105 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,3,2]
106 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2],ymm1[3]
109 ; AVX2-LABEL: shuffle_v4f64_3330:
111 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,3,3,0]
113 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
114 ret <4 x double> %shuffle
117 define <4 x double> @shuffle_v4f64_3210(<4 x double> %a, <4 x double> %b) {
118 ; AVX1-LABEL: shuffle_v4f64_3210:
120 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
121 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
124 ; AVX2-LABEL: shuffle_v4f64_3210:
126 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,1,0]
128 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
129 ret <4 x double> %shuffle
132 define <4 x double> @shuffle_v4f64_0023(<4 x double> %a, <4 x double> %b) {
133 ; ALL-LABEL: shuffle_v4f64_0023:
135 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,3]
137 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 3>
138 ret <4 x double> %shuffle
141 define <4 x double> @shuffle_v4f64_0022(<4 x double> %a, <4 x double> %b) {
142 ; ALL-LABEL: shuffle_v4f64_0022:
144 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,2]
146 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
147 ret <4 x double> %shuffle
150 define <4 x double> @shuffle_v4f64_1032(<4 x double> %a, <4 x double> %b) {
151 ; ALL-LABEL: shuffle_v4f64_1032:
153 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
155 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
156 ret <4 x double> %shuffle
159 define <4 x double> @shuffle_v4f64_1133(<4 x double> %a, <4 x double> %b) {
160 ; ALL-LABEL: shuffle_v4f64_1133:
162 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,3]
164 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
165 ret <4 x double> %shuffle
168 define <4 x double> @shuffle_v4f64_1023(<4 x double> %a, <4 x double> %b) {
169 ; ALL-LABEL: shuffle_v4f64_1023:
171 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,3]
173 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 3>
174 ret <4 x double> %shuffle
177 define <4 x double> @shuffle_v4f64_1022(<4 x double> %a, <4 x double> %b) {
178 ; ALL-LABEL: shuffle_v4f64_1022:
180 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
182 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 2>
183 ret <4 x double> %shuffle
186 define <4 x double> @shuffle_v4f64_0423(<4 x double> %a, <4 x double> %b) {
187 ; AVX1-LABEL: shuffle_v4f64_0423:
189 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,0,2,2]
190 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
193 ; AVX2-LABEL: shuffle_v4f64_0423:
195 ; AVX2-NEXT: vbroadcastsd %xmm1, %ymm1
196 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
198 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 3>
199 ret <4 x double> %shuffle
202 define <4 x double> @shuffle_v4f64_0462(<4 x double> %a, <4 x double> %b) {
203 ; ALL-LABEL: shuffle_v4f64_0462:
205 ; ALL-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,0,2,2]
206 ; ALL-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,2]
207 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
209 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 6, i32 2>
210 ret <4 x double> %shuffle
213 define <4 x double> @shuffle_v4f64_0426(<4 x double> %a, <4 x double> %b) {
214 ; ALL-LABEL: shuffle_v4f64_0426:
216 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
218 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
219 ret <4 x double> %shuffle
222 define <4 x double> @shuffle_v4f64_1537(<4 x double> %a, <4 x double> %b) {
223 ; ALL-LABEL: shuffle_v4f64_1537:
225 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
227 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
228 ret <4 x double> %shuffle
231 define <4 x double> @shuffle_v4f64_4062(<4 x double> %a, <4 x double> %b) {
232 ; ALL-LABEL: shuffle_v4f64_4062:
234 ; ALL-NEXT: vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
236 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
237 ret <4 x double> %shuffle
240 define <4 x double> @shuffle_v4f64_5173(<4 x double> %a, <4 x double> %b) {
241 ; ALL-LABEL: shuffle_v4f64_5173:
243 ; ALL-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
245 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 7, i32 3>
246 ret <4 x double> %shuffle
249 define <4 x double> @shuffle_v4f64_5163(<4 x double> %a, <4 x double> %b) {
250 ; ALL-LABEL: shuffle_v4f64_5163:
252 ; ALL-NEXT: vshufpd {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
254 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
255 ret <4 x double> %shuffle
258 define <4 x double> @shuffle_v4f64_0527(<4 x double> %a, <4 x double> %b) {
259 ; ALL-LABEL: shuffle_v4f64_0527:
261 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
263 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
264 ret <4 x double> %shuffle
267 define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
268 ; ALL-LABEL: shuffle_v4f64_4163:
270 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
272 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
273 ret <4 x double> %shuffle
276 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
277 ; ALL-LABEL: shuffle_v4f64_0145:
279 ; ALL-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
281 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
282 ret <4 x double> %shuffle
285 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
286 ; ALL-LABEL: shuffle_v4f64_4501:
288 ; ALL-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
290 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
291 ret <4 x double> %shuffle
294 define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
295 ; ALL-LABEL: shuffle_v4f64_0167:
297 ; ALL-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
299 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
300 ret <4 x double> %shuffle
303 define <4 x double> @shuffle_v4f64_1054(<4 x double> %a, <4 x double> %b) {
304 ; AVX1-LABEL: shuffle_v4f64_1054:
306 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
307 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
308 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
311 ; AVX2-LABEL: shuffle_v4f64_1054:
313 ; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,1,0]
314 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
315 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
317 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
318 ret <4 x double> %shuffle
321 define <4 x double> @shuffle_v4f64_3254(<4 x double> %a, <4 x double> %b) {
322 ; AVX1-LABEL: shuffle_v4f64_3254:
324 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
325 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
326 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
327 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
330 ; AVX2-LABEL: shuffle_v4f64_3254:
332 ; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,1,0]
333 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,2,3]
334 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
336 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
337 ret <4 x double> %shuffle
340 define <4 x double> @shuffle_v4f64_3276(<4 x double> %a, <4 x double> %b) {
341 ; AVX1-LABEL: shuffle_v4f64_3276:
343 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
344 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
345 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm1
346 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
347 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
350 ; AVX2-LABEL: shuffle_v4f64_3276:
352 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,0,3,2]
353 ; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,2,3]
354 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
356 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
357 ret <4 x double> %shuffle
360 define <4 x double> @shuffle_v4f64_1076(<4 x double> %a, <4 x double> %b) {
361 ; AVX1-LABEL: shuffle_v4f64_1076:
363 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
364 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm1
365 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
366 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
369 ; AVX2-LABEL: shuffle_v4f64_1076:
371 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,0,3,2]
372 ; AVX2-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
373 ; AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
375 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
376 ret <4 x double> %shuffle
379 define <4 x i64> @shuffle_v4i64_0000(<4 x i64> %a, <4 x i64> %b) {
380 ; AVX1-LABEL: shuffle_v4i64_0000:
382 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
383 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
386 ; AVX2-LABEL: shuffle_v4i64_0000:
388 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
390 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
391 ret <4 x i64> %shuffle
394 define <4 x i64> @shuffle_v4i64_0001(<4 x i64> %a, <4 x i64> %b) {
395 ; AVX1-LABEL: shuffle_v4i64_0001:
397 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm0[0,0]
398 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
401 ; AVX2-LABEL: shuffle_v4i64_0001:
403 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
405 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
406 ret <4 x i64> %shuffle
409 define <4 x i64> @shuffle_v4i64_0020(<4 x i64> %a, <4 x i64> %b) {
410 ; AVX1-LABEL: shuffle_v4i64_0020:
412 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
413 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
414 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
415 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
418 ; AVX2-LABEL: shuffle_v4i64_0020:
420 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,2,0]
422 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
423 ret <4 x i64> %shuffle
426 define <4 x i64> @shuffle_v4i64_0112(<4 x i64> %a, <4 x i64> %b) {
427 ; AVX1-LABEL: shuffle_v4i64_0112:
429 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
430 ; AVX1-NEXT: vshufpd {{.*#+}} xmm1 = xmm0[1],xmm1[0]
431 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
434 ; AVX2-LABEL: shuffle_v4i64_0112:
436 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
438 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 1, i32 2>
439 ret <4 x i64> %shuffle
442 define <4 x i64> @shuffle_v4i64_0300(<4 x i64> %a, <4 x i64> %b) {
443 ; AVX1-LABEL: shuffle_v4i64_0300:
445 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
446 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,1,2,2]
447 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
450 ; AVX2-LABEL: shuffle_v4i64_0300:
452 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,3,0,0]
454 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
455 ret <4 x i64> %shuffle
458 define <4 x i64> @shuffle_v4i64_1000(<4 x i64> %a, <4 x i64> %b) {
459 ; AVX1-LABEL: shuffle_v4i64_1000:
461 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
462 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
463 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
466 ; AVX2-LABEL: shuffle_v4i64_1000:
468 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,0,0,0]
470 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
471 ret <4 x i64> %shuffle
474 define <4 x i64> @shuffle_v4i64_2200(<4 x i64> %a, <4 x i64> %b) {
475 ; AVX1-LABEL: shuffle_v4i64_2200:
477 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
478 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,2,2]
481 ; AVX2-LABEL: shuffle_v4i64_2200:
483 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,0,0]
485 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
486 ret <4 x i64> %shuffle
489 define <4 x i64> @shuffle_v4i64_3330(<4 x i64> %a, <4 x i64> %b) {
490 ; AVX1-LABEL: shuffle_v4i64_3330:
492 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
493 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[1,1,2,2]
494 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,0,3,2]
495 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2],ymm1[3]
498 ; AVX2-LABEL: shuffle_v4i64_3330:
500 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,3,3,0]
502 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
503 ret <4 x i64> %shuffle
506 define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
507 ; AVX1-LABEL: shuffle_v4i64_3210:
509 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
510 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,0,3,2]
513 ; AVX2-LABEL: shuffle_v4i64_3210:
515 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,1,0]
517 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
518 ret <4 x i64> %shuffle
521 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
522 ; AVX1-LABEL: shuffle_v4i64_0124:
524 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm1[0,0]
525 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
526 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3]
529 ; AVX2-LABEL: shuffle_v4i64_0124:
531 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
532 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
534 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
535 ret <4 x i64> %shuffle
538 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
539 ; AVX1-LABEL: shuffle_v4i64_0142:
541 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm1, %ymm1
542 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[0,1,2,2]
543 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
546 ; AVX2-LABEL: shuffle_v4i64_0142:
548 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
549 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,2]
550 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
552 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
553 ret <4 x i64> %shuffle
556 define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
557 ; AVX1-LABEL: shuffle_v4i64_0412:
559 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
560 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
561 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
562 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm1 = ymm1[0,0,2,2]
563 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
566 ; AVX2-LABEL: shuffle_v4i64_0412:
568 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,2]
569 ; AVX2-NEXT: vpbroadcastq %xmm1, %ymm1
570 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
572 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
573 ret <4 x i64> %shuffle
576 define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
577 ; AVX1-LABEL: shuffle_v4i64_4012:
579 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
580 ; AVX1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1],xmm2[0]
581 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
582 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
583 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3]
586 ; AVX2-LABEL: shuffle_v4i64_4012:
588 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,2]
589 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
591 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
592 ret <4 x i64> %shuffle
595 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
596 ; ALL-LABEL: shuffle_v4i64_0145:
598 ; ALL-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
600 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
601 ret <4 x i64> %shuffle
604 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
605 ; AVX1-LABEL: shuffle_v4i64_0451:
607 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm1[1],xmm0[1]
608 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
609 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
612 ; AVX2-LABEL: shuffle_v4i64_0451:
614 ; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
615 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,0,1,3]
616 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
618 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
619 ret <4 x i64> %shuffle
622 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
623 ; ALL-LABEL: shuffle_v4i64_4501:
625 ; ALL-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
627 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
628 ret <4 x i64> %shuffle
631 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
632 ; AVX1-LABEL: shuffle_v4i64_4015:
634 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm2 = xmm0[1],xmm1[1]
635 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
636 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
639 ; AVX2-LABEL: shuffle_v4i64_4015:
641 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
642 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,0,1,3]
643 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
645 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
646 ret <4 x i64> %shuffle
649 define <4 x i64> @shuffle_v4i64_2u35(<4 x i64> %a, <4 x i64> %b) {
650 ; AVX1-LABEL: shuffle_v4i64_2u35:
652 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
653 ; AVX1-NEXT: vunpckhpd {{.*#+}} xmm1 = xmm0[1],xmm1[1]
654 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
657 ; AVX2-LABEL: shuffle_v4i64_2u35:
659 ; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm1, %ymm1
660 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,3,3]
661 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
663 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 undef, i32 3, i32 5>
664 ret <4 x i64> %shuffle
667 define <4 x i64> @shuffle_v4i64_1251(<4 x i64> %a, <4 x i64> %b) {
668 ; AVX1-LABEL: shuffle_v4i64_1251:
670 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
671 ; AVX1-NEXT: vshufpd {{.*#+}} ymm0 = ymm0[1],ymm2[0],ymm0[2],ymm2[3]
672 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
673 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
674 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
677 ; AVX2-LABEL: shuffle_v4i64_1251:
679 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,3]
680 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[1,2,2,1]
681 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
683 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 5, i32 1>
684 ret <4 x i64> %shuffle
687 define <4 x i64> @shuffle_v4i64_1054(<4 x i64> %a, <4 x i64> %b) {
688 ; AVX1-LABEL: shuffle_v4i64_1054:
690 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
691 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
692 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
695 ; AVX2-LABEL: shuffle_v4i64_1054:
697 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,0]
698 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
699 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
701 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
702 ret <4 x i64> %shuffle
705 define <4 x i64> @shuffle_v4i64_3254(<4 x i64> %a, <4 x i64> %b) {
706 ; AVX1-LABEL: shuffle_v4i64_3254:
708 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
709 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
710 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
711 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
714 ; AVX2-LABEL: shuffle_v4i64_3254:
716 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,1,0]
717 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,2,3]
718 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
720 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
721 ret <4 x i64> %shuffle
724 define <4 x i64> @shuffle_v4i64_3276(<4 x i64> %a, <4 x i64> %b) {
725 ; AVX1-LABEL: shuffle_v4i64_3276:
727 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
728 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
729 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm1
730 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
731 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
734 ; AVX2-LABEL: shuffle_v4i64_3276:
736 ; AVX2-NEXT: vpshufd {{.*#+}} ymm1 = ymm1[2,3,0,1,6,7,4,5]
737 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[3,2,2,3]
738 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
740 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
741 ret <4 x i64> %shuffle
744 define <4 x i64> @shuffle_v4i64_1076(<4 x i64> %a, <4 x i64> %b) {
745 ; AVX1-LABEL: shuffle_v4i64_1076:
747 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
748 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm1
749 ; AVX1-NEXT: vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
750 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
753 ; AVX2-LABEL: shuffle_v4i64_1076:
755 ; AVX2-NEXT: vpshufd {{.*#+}} ymm1 = ymm1[2,3,0,1,6,7,4,5]
756 ; AVX2-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
757 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
759 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
760 ret <4 x i64> %shuffle
763 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
764 ; AVX1-LABEL: stress_test1:
766 ; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm1[2,3,0,1]
767 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm2 = ymm0[1,0,3,2]
768 ; AVX1-NEXT: vblendpd {{.*#+}} ymm1 = ymm2[0],ymm1[1],ymm2[2,3]
769 ; AVX1-NEXT: vpermilpd {{.*#+}} ymm0 = ymm0[1,1,3,2]
770 ; AVX1-NEXT: vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
773 ; AVX2-LABEL: stress_test1:
775 ; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm1[3,1,1,0]
776 ; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[3,3,1,3]
777 ; AVX2-NEXT: vpunpckhqdq {{.*#+}} ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
779 %c = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 1, i32 1, i32 0>
780 %d = shufflevector <4 x i64> %c, <4 x i64> undef, <4 x i32> <i32 3, i32 undef, i32 2, i32 undef>
781 %e = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 undef>
782 %f = shufflevector <4 x i64> %d, <4 x i64> %e, <4 x i32> <i32 5, i32 1, i32 1, i32 0>
787 define <4 x i64> @insert_reg_and_zero_v4i64(i64 %a) {
788 ; AVX1-LABEL: insert_reg_and_zero_v4i64:
790 ; AVX1-NEXT: vmovq %rdi, %xmm0
791 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
792 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
795 ; AVX2-LABEL: insert_reg_and_zero_v4i64:
797 ; AVX2-NEXT: vmovq %rdi, %xmm0
798 ; AVX2-NEXT: vpxor %ymm1, %ymm1, %ymm1
799 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
801 %v = insertelement <4 x i64> undef, i64 %a, i64 0
802 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
803 ret <4 x i64> %shuffle
806 define <4 x i64> @insert_mem_and_zero_v4i64(i64* %ptr) {
807 ; AVX1-LABEL: insert_mem_and_zero_v4i64:
809 ; AVX1-NEXT: vmovq (%rdi), %xmm0
810 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
811 ; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
814 ; AVX2-LABEL: insert_mem_and_zero_v4i64:
816 ; AVX2-NEXT: vmovq (%rdi), %xmm0
817 ; AVX2-NEXT: vpxor %ymm1, %ymm1, %ymm1
818 ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
821 %v = insertelement <4 x i64> undef, i64 %a, i64 0
822 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
823 ret <4 x i64> %shuffle
826 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
827 ; ALL-LABEL: insert_reg_and_zero_v4f64:
829 ; ALL-NEXT: vxorps %xmm1, %xmm1, %xmm1
830 ; ALL-NEXT: vmovsd %xmm0, %xmm1, %xmm0
832 %v = insertelement <4 x double> undef, double %a, i32 0
833 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
834 ret <4 x double> %shuffle
837 define <4 x double> @insert_mem_and_zero_v4f64(double* %ptr) {
838 ; ALL-LABEL: insert_mem_and_zero_v4f64:
840 ; ALL-NEXT: vmovsd (%rdi), %xmm0
842 %a = load double* %ptr
843 %v = insertelement <4 x double> undef, double %a, i32 0
844 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
845 ret <4 x double> %shuffle
848 define <4 x double> @splat_mem_v4f64(double* %ptr) {
849 ; ALL-LABEL: splat_mem_v4f64:
851 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
853 %a = load double* %ptr
854 %v = insertelement <4 x double> undef, double %a, i32 0
855 %shuffle = shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
856 ret <4 x double> %shuffle
859 define <4 x i64> @splat_mem_v4i64(i64* %ptr) {
860 ; AVX1-LABEL: splat_mem_v4i64:
862 ; AVX1-NEXT: vmovddup (%rdi), %xmm0
863 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
866 ; AVX2-LABEL: splat_mem_v4i64:
868 ; AVX2-NEXT: vbroadcastsd (%rdi), %ymm0
871 %v = insertelement <4 x i64> undef, i64 %a, i64 0
872 %shuffle = shufflevector <4 x i64> %v, <4 x i64> undef, <4 x i32> <i32 0, i32 0, i32 0, i32 0>
873 ret <4 x i64> %shuffle
876 define <4 x double> @splat_mem_v4f64_2(double* %p) {
877 ; ALL-LABEL: splat_mem_v4f64_2:
879 ; ALL-NEXT: vbroadcastsd (%rdi), %ymm0
882 %2 = insertelement <2 x double> undef, double %1, i32 0
883 %3 = shufflevector <2 x double> %2, <2 x double> undef, <4 x i32> zeroinitializer
887 define <4 x double> @splat_v4f64(<2 x double> %r) {
888 ; AVX1-LABEL: splat_v4f64:
890 ; AVX1-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
891 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
894 ; AVX2-LABEL: splat_v4f64:
896 ; AVX2-NEXT: vbroadcastsd %xmm0, %ymm0
898 %1 = shufflevector <2 x double> %r, <2 x double> undef, <4 x i32> zeroinitializer