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_0001(<4 x double> %a, <4 x double> %b) {
7 ; AVX1-LABEL: @shuffle_v4f64_0001
9 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm1 = xmm0[0,0]
10 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
13 ; AVX2-LABEL: @shuffle_v4f64_0001
15 ; AVX2-NEXT: vpermpd {{.*}} # ymm0 = ymm0[0,0,0,1]
17 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
18 ret <4 x double> %shuffle
21 define <4 x double> @shuffle_v4f64_0020(<4 x double> %a, <4 x double> %b) {
22 ; AVX1-LABEL: @shuffle_v4f64_0020
24 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
25 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm1 = xmm1[0],xmm0[0]
26 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
27 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
30 ; AVX2-LABEL: @shuffle_v4f64_0020
32 ; AVX2-NEXT: vpermpd {{.*}} # ymm0 = ymm0[0,0,2,0]
34 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
35 ret <4 x double> %shuffle
38 define <4 x double> @shuffle_v4f64_0300(<4 x double> %a, <4 x double> %b) {
39 ; AVX1-LABEL: @shuffle_v4f64_0300
41 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
42 ; AVX1-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[0,1,2,2]
43 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
46 ; AVX2-LABEL: @shuffle_v4f64_0300
48 ; AVX2-NEXT: vpermpd {{.*}} # ymm0 = ymm0[0,3,0,0]
50 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
51 ret <4 x double> %shuffle
54 define <4 x double> @shuffle_v4f64_1000(<4 x double> %a, <4 x double> %b) {
55 ; AVX1-LABEL: @shuffle_v4f64_1000
57 ; AVX1-NEXT: vpermilpd {{.*}} # xmm1 = xmm0[1,0]
58 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
59 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
62 ; AVX2-LABEL: @shuffle_v4f64_1000
64 ; AVX2-NEXT: vpermpd {{.*}} # ymm0 = ymm0[1,0,0,0]
66 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
67 ret <4 x double> %shuffle
70 define <4 x double> @shuffle_v4f64_2200(<4 x double> %a, <4 x double> %b) {
71 ; AVX1-LABEL: @shuffle_v4f64_2200
73 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
74 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,2]
77 ; AVX2-LABEL: @shuffle_v4f64_2200
79 ; AVX2-NEXT: vpermpd {{.*}} # ymm0 = ymm0[2,2,0,0]
81 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
82 ret <4 x double> %shuffle
85 define <4 x double> @shuffle_v4f64_3330(<4 x double> %a, <4 x double> %b) {
86 ; AVX1-LABEL: @shuffle_v4f64_3330
88 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
89 ; AVX1-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[1,1,2,2]
90 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,3,2]
91 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2],ymm1[3]
94 ; AVX2-LABEL: @shuffle_v4f64_3330
96 ; AVX2-NEXT: vpermpd {{.*}} # ymm0 = ymm0[3,3,3,0]
98 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
99 ret <4 x double> %shuffle
102 define <4 x double> @shuffle_v4f64_3210(<4 x double> %a, <4 x double> %b) {
103 ; AVX1-LABEL: @shuffle_v4f64_3210
105 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
106 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,3,2]
109 ; AVX2-LABEL: @shuffle_v4f64_3210
111 ; AVX2-NEXT: vpermpd {{.*}} # ymm0 = ymm0[3,2,1,0]
113 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
114 ret <4 x double> %shuffle
117 define <4 x double> @shuffle_v4f64_0023(<4 x double> %a, <4 x double> %b) {
118 ; ALL-LABEL: @shuffle_v4f64_0023
120 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,3]
122 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 3>
123 ret <4 x double> %shuffle
126 define <4 x double> @shuffle_v4f64_0022(<4 x double> %a, <4 x double> %b) {
127 ; ALL-LABEL: @shuffle_v4f64_0022
129 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,2]
131 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
132 ret <4 x double> %shuffle
135 define <4 x double> @shuffle_v4f64_1032(<4 x double> %a, <4 x double> %b) {
136 ; ALL-LABEL: @shuffle_v4f64_1032
138 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,3,2]
140 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
141 ret <4 x double> %shuffle
144 define <4 x double> @shuffle_v4f64_1133(<4 x double> %a, <4 x double> %b) {
145 ; ALL-LABEL: @shuffle_v4f64_1133
147 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,1,3,3]
149 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
150 ret <4 x double> %shuffle
153 define <4 x double> @shuffle_v4f64_1023(<4 x double> %a, <4 x double> %b) {
154 ; ALL-LABEL: @shuffle_v4f64_1023
156 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,2,3]
158 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 3>
159 ret <4 x double> %shuffle
162 define <4 x double> @shuffle_v4f64_1022(<4 x double> %a, <4 x double> %b) {
163 ; ALL-LABEL: @shuffle_v4f64_1022
165 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,2,2]
167 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 2, i32 2>
168 ret <4 x double> %shuffle
171 define <4 x double> @shuffle_v4f64_0423(<4 x double> %a, <4 x double> %b) {
172 ; ALL-LABEL: @shuffle_v4f64_0423
174 ; ALL-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[0,0,2,2]
175 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
177 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 3>
178 ret <4 x double> %shuffle
181 define <4 x double> @shuffle_v4f64_0462(<4 x double> %a, <4 x double> %b) {
182 ; ALL-LABEL: @shuffle_v4f64_0462
184 ; ALL-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[0,0,2,2]
185 ; ALL-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,2]
186 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
188 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 6, i32 2>
189 ret <4 x double> %shuffle
192 define <4 x double> @shuffle_v4f64_0426(<4 x double> %a, <4 x double> %b) {
193 ; ALL-LABEL: @shuffle_v4f64_0426
195 ; ALL-NEXT: vunpcklpd {{.*}} # ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
197 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
198 ret <4 x double> %shuffle
201 define <4 x double> @shuffle_v4f64_1537(<4 x double> %a, <4 x double> %b) {
202 ; ALL-LABEL: @shuffle_v4f64_1537
204 ; ALL-NEXT: vunpckhpd {{.*}} # ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
206 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
207 ret <4 x double> %shuffle
210 define <4 x double> @shuffle_v4f64_4062(<4 x double> %a, <4 x double> %b) {
211 ; ALL-LABEL: @shuffle_v4f64_4062
213 ; ALL-NEXT: vunpcklpd {{.*}} # ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
215 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
216 ret <4 x double> %shuffle
219 define <4 x double> @shuffle_v4f64_5173(<4 x double> %a, <4 x double> %b) {
220 ; ALL-LABEL: @shuffle_v4f64_5173
222 ; ALL-NEXT: vunpckhpd {{.*}} # ymm0 = ymm1[1],ymm0[1],ymm1[3],ymm0[3]
224 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 7, i32 3>
225 ret <4 x double> %shuffle
228 define <4 x double> @shuffle_v4f64_5163(<4 x double> %a, <4 x double> %b) {
229 ; ALL-LABEL: @shuffle_v4f64_5163
231 ; ALL-NEXT: vshufpd {{.*}} # ymm0 = ymm1[1],ymm0[1],ymm1[2],ymm0[3]
233 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
234 ret <4 x double> %shuffle
237 define <4 x double> @shuffle_v4f64_0527(<4 x double> %a, <4 x double> %b) {
238 ; ALL-LABEL: @shuffle_v4f64_0527
240 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3]
242 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
243 ret <4 x double> %shuffle
246 define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
247 ; ALL-LABEL: @shuffle_v4f64_4163
249 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3]
251 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
252 ret <4 x double> %shuffle
255 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
256 ; AVX1-LABEL: @shuffle_v4f64_0145
258 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
259 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3]
262 ; AVX2-LABEL: @shuffle_v4f64_0145
264 ; AVX2-NEXT: vpermpd {{.*}} # ymm1 = ymm1[0,1,0,1]
265 ; AVX2-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3]
267 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
268 ret <4 x double> %shuffle
271 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
272 ; AVX1-LABEL: @shuffle_v4f64_4501
274 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
275 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3]
278 ; AVX2-LABEL: @shuffle_v4f64_4501
280 ; AVX2-NEXT: vpermpd {{.*}} # ymm0 = ymm0[0,1,0,1]
281 ; AVX2-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3]
283 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
284 ret <4 x double> %shuffle
287 define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
288 ; ALL-LABEL: @shuffle_v4f64_0167
290 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3]
292 %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
293 ret <4 x double> %shuffle
296 define <4 x i64> @shuffle_v4i64_0001(<4 x i64> %a, <4 x i64> %b) {
297 ; AVX1-LABEL: @shuffle_v4i64_0001
299 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm1 = xmm0[0,0]
300 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
303 ; AVX2-LABEL: @shuffle_v4i64_0001
305 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,0,0,1]
307 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
308 ret <4 x i64> %shuffle
311 define <4 x i64> @shuffle_v4i64_0020(<4 x i64> %a, <4 x i64> %b) {
312 ; AVX1-LABEL: @shuffle_v4i64_0020
314 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
315 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm1 = xmm1[0],xmm0[0]
316 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
317 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
320 ; AVX2-LABEL: @shuffle_v4i64_0020
322 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,0,2,0]
324 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
325 ret <4 x i64> %shuffle
328 define <4 x i64> @shuffle_v4i64_0112(<4 x i64> %a, <4 x i64> %b) {
329 ; AVX1-LABEL: @shuffle_v4i64_0112
331 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
332 ; AVX1-NEXT: vshufpd {{.*}} # xmm1 = xmm0[1],xmm1[0]
333 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
336 ; AVX2-LABEL: @shuffle_v4i64_0112
338 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,1,2]
340 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 1, i32 2>
341 ret <4 x i64> %shuffle
344 define <4 x i64> @shuffle_v4i64_0300(<4 x i64> %a, <4 x i64> %b) {
345 ; AVX1-LABEL: @shuffle_v4i64_0300
347 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
348 ; AVX1-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[0,1,2,2]
349 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
352 ; AVX2-LABEL: @shuffle_v4i64_0300
354 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,3,0,0]
356 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
357 ret <4 x i64> %shuffle
360 define <4 x i64> @shuffle_v4i64_1000(<4 x i64> %a, <4 x i64> %b) {
361 ; AVX1-LABEL: @shuffle_v4i64_1000
363 ; AVX1-NEXT: vpermilpd {{.*}} # xmm1 = xmm0[1,0]
364 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
365 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
368 ; AVX2-LABEL: @shuffle_v4i64_1000
370 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[1,0,0,0]
372 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
373 ret <4 x i64> %shuffle
376 define <4 x i64> @shuffle_v4i64_2200(<4 x i64> %a, <4 x i64> %b) {
377 ; AVX1-LABEL: @shuffle_v4i64_2200
379 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
380 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,2]
383 ; AVX2-LABEL: @shuffle_v4i64_2200
385 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[2,2,0,0]
387 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
388 ret <4 x i64> %shuffle
391 define <4 x i64> @shuffle_v4i64_3330(<4 x i64> %a, <4 x i64> %b) {
392 ; AVX1-LABEL: @shuffle_v4i64_3330
394 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
395 ; AVX1-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[1,1,2,2]
396 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,0,3,2]
397 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2],ymm1[3]
400 ; AVX2-LABEL: @shuffle_v4i64_3330
402 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[3,3,3,0]
404 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
405 ret <4 x i64> %shuffle
408 define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
409 ; AVX1-LABEL: @shuffle_v4i64_3210
411 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
412 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,3,2]
415 ; AVX2-LABEL: @shuffle_v4i64_3210
417 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[3,2,1,0]
419 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
420 ret <4 x i64> %shuffle
423 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
424 ; AVX1-LABEL: @shuffle_v4i64_0124
426 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm1 = xmm1[0,0]
427 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
428 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0,1,2],ymm1[3]
431 ; AVX2-LABEL: @shuffle_v4i64_0124
433 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,1,2,0]
434 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
436 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
437 ret <4 x i64> %shuffle
440 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
441 ; AVX1-LABEL: @shuffle_v4i64_0142
443 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
444 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[0,1,2,2]
445 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
448 ; AVX2-LABEL: @shuffle_v4i64_0142
450 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,1,0,3]
451 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,2,2]
452 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
454 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
455 ret <4 x i64> %shuffle
458 define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
459 ; AVX1-LABEL: @shuffle_v4i64_0412
461 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
462 ; AVX1-NEXT: vshufpd {{.*}} # xmm2 = xmm0[1],xmm2[0]
463 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
464 ; AVX1-NEXT: vpermilpd {{.*}} # ymm1 = ymm1[0,0,2,2]
465 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
468 ; AVX2-LABEL: @shuffle_v4i64_0412
470 ; AVX2-NEXT: vpshufd {{.*}} # ymm1 = ymm1[0,1,0,1,4,5,4,5]
471 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,1,2]
472 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
474 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
475 ret <4 x i64> %shuffle
478 define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
479 ; AVX1-LABEL: @shuffle_v4i64_4012
481 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
482 ; AVX1-NEXT: vshufpd {{.*}} # xmm2 = xmm0[1],xmm2[0]
483 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
484 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
485 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0],ymm0[1,2,3]
488 ; AVX2-LABEL: @shuffle_v4i64_4012
490 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,0,1,2]
491 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
493 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
494 ret <4 x i64> %shuffle
497 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
498 ; AVX1-LABEL: @shuffle_v4i64_0145
500 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
501 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3]
504 ; AVX2-LABEL: @shuffle_v4i64_0145
506 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,1,0,1]
507 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
509 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
510 ret <4 x i64> %shuffle
513 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
514 ; AVX1-LABEL: @shuffle_v4i64_0451
516 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
517 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm2 = xmm1[0,0]
518 ; AVX1-NEXT: vpermilpd {{.*}} # xmm1 = xmm1[1,0]
519 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm2, %ymm1
520 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
523 ; AVX2-LABEL: @shuffle_v4i64_0451
525 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,0,1,3]
526 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,2,1]
527 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3,4,5],ymm0[6,7]
529 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
530 ret <4 x i64> %shuffle
533 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
534 ; AVX1-LABEL: @shuffle_v4i64_4501
536 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
537 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3]
540 ; AVX2-LABEL: @shuffle_v4i64_4501
542 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,0,1]
543 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
545 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
546 ret <4 x i64> %shuffle
549 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
550 ; AVX1-LABEL: @shuffle_v4i64_4015
552 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm1, %ymm1
553 ; AVX1-NEXT: vunpcklpd {{.*}} # xmm2 = xmm0[0,0]
554 ; AVX1-NEXT: vpermilpd {{.*}} # xmm0 = xmm0[1,0]
555 ; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm2, %ymm0
556 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm1[0],ymm0[1,2],ymm1[3]
559 ; AVX2-LABEL: @shuffle_v4i64_4015
561 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,1,2,1]
562 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,0,1,3]
563 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
565 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
566 ret <4 x i64> %shuffle
569 define <4 x i64> @shuffle_v4i64_2u35(<4 x i64> %a, <4 x i64> %b) {
570 ; AVX1-LABEL: @shuffle_v4i64_2u35
572 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
573 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
574 ; AVX1-NEXT: vpermilpd {{.*}} # xmm2 = xmm0[1,0]
575 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
576 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0,1,2],ymm1[3]
579 ; AVX2-LABEL: @shuffle_v4i64_2u35
581 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,1,2,1]
582 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[2,1,3,3]
583 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
585 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 2, i32 undef, i32 3, i32 5>
586 ret <4 x i64> %shuffle
589 define <4 x i64> @shuffle_v4i64_1251(<4 x i64> %a, <4 x i64> %b) {
590 ; AVX1-LABEL: @shuffle_v4i64_1251
592 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm2 = ymm0[2,3,0,1]
593 ; AVX1-NEXT: vshufpd {{.*}} # ymm0 = ymm0[1],ymm2[0],ymm0[2],ymm2[3]
594 ; AVX1-NEXT: vpermilpd {{.*}} # xmm1 = xmm1[1,0]
595 ; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
596 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
599 ; AVX2-LABEL: @shuffle_v4i64_1251
601 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[0,1,1,3]
602 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[1,2,2,1]
603 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
605 %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 5, i32 1>
606 ret <4 x i64> %shuffle
609 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
610 ; AVX1-LABEL: @stress_test1
612 ; AVX1-NEXT: vperm2f128 {{.*}} # ymm0 = ymm1[2,3,0,1]
613 ; AVX1-NEXT: vpermilpd {{.*}} # ymm2 = ymm0[1,0,3,2]
614 ; AVX1-NEXT: vblendpd {{.*}} # ymm1 = ymm2[0],ymm1[1],ymm2[2,3]
615 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm1
616 ; AVX1-NEXT: vpermilpd {{.*}} # xmm1 = xmm1[1,0]
617 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,1,3,2]
618 ; AVX1-NEXT: vpermilpd {{.*}} # xmm2 = xmm1[1,0]
619 ; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm1, %ymm1
620 ; AVX1-NEXT: vpermilpd {{.*}} # ymm0 = ymm0[1,0,2,2]
621 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
624 ; AVX2-LABEL: @stress_test1
626 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm1[3,1,1,0]
627 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[3,1,2,3]
628 ; AVX2-NEXT: vpermq {{.*}} # ymm1 = ymm1[3,3,1,3]
629 ; AVX2-NEXT: vpshufd {{.*}} # ymm1 = ymm1[2,3,2,3,6,7,6,7]
630 ; AVX2-NEXT: vpermq {{.*}} # ymm0 = ymm0[0,1,1,0]
631 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
633 %c = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 1, i32 1, i32 0>
634 %d = shufflevector <4 x i64> %c, <4 x i64> undef, <4 x i32> <i32 3, i32 undef, i32 2, i32 undef>
635 %e = shufflevector <4 x i64> %b, <4 x i64> undef, <4 x i32> <i32 3, i32 3, i32 1, i32 undef>
636 %f = shufflevector <4 x i64> %d, <4 x i64> %e, <4 x i32> <i32 5, i32 1, i32 1, i32 0>
641 define <4 x i64> @insert_reg_and_zero_v4i64(i64 %a) {
642 ; AVX1-LABEL: @insert_reg_and_zero_v4i64
644 ; AVX1-NEXT: vmovq %rdi, %xmm0
645 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
646 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
649 ; AVX2-LABEL: @insert_reg_and_zero_v4i64
651 ; AVX2-NEXT: vmovq %rdi, %xmm0
652 ; AVX2-NEXT: vpxor %ymm1, %ymm1, %ymm1
653 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
655 %v = insertelement <4 x i64> undef, i64 %a, i64 0
656 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
657 ret <4 x i64> %shuffle
660 define <4 x i64> @insert_mem_and_zero_v4i64(i64* %ptr) {
661 ; AVX1-LABEL: @insert_mem_and_zero_v4i64
663 ; AVX1-NEXT: vmovq (%rdi), %xmm0
664 ; AVX1-NEXT: vxorpd %ymm1, %ymm1, %ymm1
665 ; AVX1-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
668 ; AVX2-LABEL: @insert_mem_and_zero_v4i64
670 ; AVX2-NEXT: vmovq (%rdi), %xmm0
671 ; AVX2-NEXT: vpxor %ymm1, %ymm1, %ymm1
672 ; AVX2-NEXT: vpblendd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
675 %v = insertelement <4 x i64> undef, i64 %a, i64 0
676 %shuffle = shufflevector <4 x i64> %v, <4 x i64> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
677 ret <4 x i64> %shuffle
680 define <4 x double> @insert_reg_and_zero_v4f64(double %a) {
681 ; ALL-LABEL: @insert_reg_and_zero_v4f64
683 ; ALL: vxorpd %ymm1, %ymm1, %ymm1
684 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
686 %v = insertelement <4 x double> undef, double %a, i32 0
687 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
688 ret <4 x double> %shuffle
691 define <4 x double> @insert_mem_and_zero_v4f64(double* %ptr) {
692 ; ALL-LABEL: @insert_mem_and_zero_v4f64
694 ; ALL-NEXT: vmovsd (%rdi), %xmm0
695 ; ALL-NEXT: vxorpd %ymm1, %ymm1, %ymm1
696 ; ALL-NEXT: vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
698 %a = load double* %ptr
699 %v = insertelement <4 x double> undef, double %a, i32 0
700 %shuffle = shufflevector <4 x double> %v, <4 x double> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 6, i32 7>
701 ret <4 x double> %shuffle