1 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -x86-experimental-vector-shuffle-lowering | FileCheck %s --check-prefix=ALL --check-prefix=SSE2
2 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+ssse3 -x86-experimental-vector-shuffle-lowering | FileCheck %s --check-prefix=ALL --check-prefix=SSSE3
4 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
5 target triple = "x86_64-unknown-unknown"
7 define <8 x i16> @shuffle_v8i16_01012323(<8 x i16> %a, <8 x i16> %b) {
8 ; ALL-LABEL: @shuffle_v8i16_01012323
10 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,0,1,1]
12 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 2, i32 3, i32 2, i32 3>
13 ret <8 x i16> %shuffle
15 define <8 x i16> @shuffle_v8i16_67452301(<8 x i16> %a, <8 x i16> %b) {
16 ; ALL-LABEL: @shuffle_v8i16_67452301
18 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,2,1,0]
20 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 6, i32 7, i32 4, i32 5, i32 2, i32 3, i32 0, i32 1>
21 ret <8 x i16> %shuffle
23 define <8 x i16> @shuffle_v8i16_456789AB(<8 x i16> %a, <8 x i16> %b) {
24 ; ALL-LABEL: @shuffle_v8i16_456789AB
26 ; ALL: shufpd {{.*}} # xmm0 = xmm0[1],xmm1[0]
28 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
29 ret <8 x i16> %shuffle
32 define <8 x i16> @shuffle_v8i16_00000000(<8 x i16> %a, <8 x i16> %b) {
33 ; SSE2-LABEL: @shuffle_v8i16_00000000
35 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,0,3]
36 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,0,0,4,5,6,7]
37 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
40 ; SSSE3-LABEL: @shuffle_v8i16_00000000
42 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
44 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
45 ret <8 x i16> %shuffle
47 define <8 x i16> @shuffle_v8i16_00004444(<8 x i16> %a, <8 x i16> %b) {
48 ; ALL-LABEL: @shuffle_v8i16_00004444
50 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,0,0,4,5,6,7]
51 ; ALL-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
53 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 4, i32 4, i32 4, i32 4>
54 ret <8 x i16> %shuffle
56 define <8 x i16> @shuffle_v8i16_31206745(<8 x i16> %a, <8 x i16> %b) {
57 ; ALL-LABEL: @shuffle_v8i16_31206745
59 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[3,1,2,0,4,5,6,7]
60 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,3,2]
62 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 3, i32 1, i32 2, i32 0, i32 6, i32 7, i32 4, i32 5>
63 ret <8 x i16> %shuffle
65 define <8 x i16> @shuffle_v8i16_44440000(<8 x i16> %a, <8 x i16> %b) {
66 ; SSE2-LABEL: @shuffle_v8i16_44440000
68 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,1,0,3]
69 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,0,0,4,5,6,7]
70 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
73 ; SSSE3-LABEL: @shuffle_v8i16_44440000
75 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,8,9,8,9,8,9,0,1,0,1,0,1,0,1]
77 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 4, i32 4, i32 4, i32 4, i32 0, i32 0, i32 0, i32 0>
78 ret <8 x i16> %shuffle
80 define <8 x i16> @shuffle_v8i16_75643120(<8 x i16> %a, <8 x i16> %b) {
81 ; SSE2-LABEL: @shuffle_v8i16_75643120
83 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,0,1]
84 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[3,1,2,0,4,5,6,7]
85 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,5,6,4]
88 ; SSSE3-LABEL: @shuffle_v8i16_75643120
90 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[14,15,10,11,12,13,8,9,6,7,2,3,4,5,0,1]
92 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 7, i32 5, i32 6, i32 4, i32 3, i32 1, i32 2, i32 0>
93 ret <8 x i16> %shuffle
96 define <8 x i16> @shuffle_v8i16_10545410(<8 x i16> %a, <8 x i16> %b) {
97 ; SSE2-LABEL: @shuffle_v8i16_10545410
99 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,0]
100 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,0,3,2,4,5,6,7]
101 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,5,4,7,6]
104 ; SSSE3-LABEL: @shuffle_v8i16_10545410
106 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[2,3,0,1,10,11,8,9,10,11,8,9,2,3,0,1]
108 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 1, i32 0, i32 5, i32 4, i32 5, i32 4, i32 1, i32 0>
109 ret <8 x i16> %shuffle
111 define <8 x i16> @shuffle_v8i16_54105410(<8 x i16> %a, <8 x i16> %b) {
112 ; SSE2-LABEL: @shuffle_v8i16_54105410
114 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,0]
115 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[3,2,1,0,4,5,6,7]
116 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,5,4,7,6]
119 ; SSSE3-LABEL: @shuffle_v8i16_54105410
121 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[10,11,8,9,2,3,0,1,10,11,8,9,2,3,0,1]
123 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 5, i32 4, i32 1, i32 0, i32 5, i32 4, i32 1, i32 0>
124 ret <8 x i16> %shuffle
126 define <8 x i16> @shuffle_v8i16_54101054(<8 x i16> %a, <8 x i16> %b) {
127 ; SSE2-LABEL: @shuffle_v8i16_54101054
129 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,0]
130 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[3,2,1,0,4,5,6,7]
131 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,6,5,4]
134 ; SSSE3-LABEL: @shuffle_v8i16_54101054
136 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[10,11,8,9,2,3,0,1,2,3,0,1,10,11,8,9]
138 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 5, i32 4, i32 1, i32 0, i32 1, i32 0, i32 5, i32 4>
139 ret <8 x i16> %shuffle
141 define <8 x i16> @shuffle_v8i16_04400440(<8 x i16> %a, <8 x i16> %b) {
142 ; SSE2-LABEL: @shuffle_v8i16_04400440
144 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,0]
145 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
146 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,6,4,4,6]
149 ; SSSE3-LABEL: @shuffle_v8i16_04400440
151 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,8,9,8,9,0,1]
153 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 4, i32 4, i32 0, i32 0, i32 4, i32 4, i32 0>
154 ret <8 x i16> %shuffle
156 define <8 x i16> @shuffle_v8i16_40044004(<8 x i16> %a, <8 x i16> %b) {
157 ; SSE2-LABEL: @shuffle_v8i16_40044004
159 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,0]
160 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,0,0,2,4,5,6,7]
161 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,4]
164 ; SSSE3-LABEL: @shuffle_v8i16_40044004
166 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,0,1,0,1,8,9,8,9,0,1,0,1,8,9]
168 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 4, i32 0, i32 0, i32 4, i32 4, i32 0, i32 0, i32 4>
169 ret <8 x i16> %shuffle
172 define <8 x i16> @shuffle_v8i16_26405173(<8 x i16> %a, <8 x i16> %b) {
173 ; SSE2-LABEL: @shuffle_v8i16_26405173
175 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,1,3,4,5,6,7]
176 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,5,4,6]
177 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,2,1]
178 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,3,2,0,4,5,6,7]
179 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,5,6,4,7]
182 ; SSSE3-LABEL: @shuffle_v8i16_26405173
184 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[4,5,12,13,8,9,0,1,10,11,2,3,14,15,6,7]
186 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 2, i32 6, i32 4, i32 0, i32 5, i32 1, i32 7, i32 3>
187 ret <8 x i16> %shuffle
189 define <8 x i16> @shuffle_v8i16_20645173(<8 x i16> %a, <8 x i16> %b) {
190 ; SSE2-LABEL: @shuffle_v8i16_20645173
192 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,1,3,4,5,6,7]
193 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,5,4,6]
194 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,2,1]
195 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,0,3,2,4,5,6,7]
196 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,5,6,4,7]
199 ; SSSE3-LABEL: @shuffle_v8i16_20645173
201 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[4,5,0,1,12,13,8,9,10,11,2,3,14,15,6,7]
203 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 2, i32 0, i32 6, i32 4, i32 5, i32 1, i32 7, i32 3>
204 ret <8 x i16> %shuffle
206 define <8 x i16> @shuffle_v8i16_26401375(<8 x i16> %a, <8 x i16> %b) {
207 ; SSE2-LABEL: @shuffle_v8i16_26401375
209 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,1,3,4,5,6,7]
210 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,5,4,6]
211 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,1,2]
212 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,3,2,0,4,5,6,7]
215 ; SSSE3-LABEL: @shuffle_v8i16_26401375
217 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,4,5,8,9,12,13,2,3,6,7,14,15,10,11]
218 ; SSSE3-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,3,2,0,4,5,6,7]
220 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 2, i32 6, i32 4, i32 0, i32 1, i32 3, i32 7, i32 5>
221 ret <8 x i16> %shuffle
224 define <8 x i16> @shuffle_v8i16_00444444(<8 x i16> %a, <8 x i16> %b) {
225 ; SSE2-LABEL: @shuffle_v8i16_00444444
227 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
228 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,2,2,4,5,6,7]
229 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
232 ; SSSE3-LABEL: @shuffle_v8i16_00444444
234 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9]
236 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 0, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4>
237 ret <8 x i16> %shuffle
239 define <8 x i16> @shuffle_v8i16_44004444(<8 x i16> %a, <8 x i16> %b) {
240 ; SSE2-LABEL: @shuffle_v8i16_44004444
242 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
243 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,2,0,0,4,5,6,7]
244 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
247 ; SSSE3-LABEL: @shuffle_v8i16_44004444
249 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9]
251 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 4, i32 4, i32 0, i32 0, i32 4, i32 4, i32 4, i32 4>
252 ret <8 x i16> %shuffle
254 define <8 x i16> @shuffle_v8i16_04404444(<8 x i16> %a, <8 x i16> %b) {
255 ; SSE2-LABEL: @shuffle_v8i16_04404444
257 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
258 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
259 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
262 ; SSSE3-LABEL: @shuffle_v8i16_04404444
264 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
266 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 4, i32 4, i32 0, i32 4, i32 4, i32 4, i32 4>
267 ret <8 x i16> %shuffle
269 define <8 x i16> @shuffle_v8i16_04400000(<8 x i16> %a, <8 x i16> %b) {
270 ; SSE2-LABEL: @shuffle_v8i16_04400000
272 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,0,3]
273 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
274 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
277 ; SSSE3-LABEL: @shuffle_v8i16_04400000
279 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1]
281 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 4, i32 4, i32 0, i32 0, i32 0, i32 0, i32 0>
282 ret <8 x i16> %shuffle
284 define <8 x i16> @shuffle_v8i16_04404567(<8 x i16> %a, <8 x i16> %b) {
285 ; ALL-LABEL: @shuffle_v8i16_04404567
287 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
288 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
290 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 4, i32 4, i32 0, i32 4, i32 5, i32 6, i32 7>
291 ret <8 x i16> %shuffle
294 define <8 x i16> @shuffle_v8i16_0X444444(<8 x i16> %a, <8 x i16> %b) {
295 ; SSE2-LABEL: @shuffle_v8i16_0X444444
297 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
298 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,1,2,2,4,5,6,7]
299 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
302 ; SSSE3-LABEL: @shuffle_v8i16_0X444444
304 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,{{[0-9]+,[0-9]+}},8,9,8,9,8,9,8,9,8,9,8,9]
306 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 undef, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4>
307 ret <8 x i16> %shuffle
309 define <8 x i16> @shuffle_v8i16_44X04444(<8 x i16> %a, <8 x i16> %b) {
310 ; SSE2-LABEL: @shuffle_v8i16_44X04444
312 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
313 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,2,2,0,4,5,6,7]
314 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
317 ; SSSE3-LABEL: @shuffle_v8i16_44X04444
319 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,8,9,{{[0-9]+,[0-9]+}},0,1,8,9,8,9,8,9,8,9]
321 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 4, i32 4, i32 undef, i32 0, i32 4, i32 4, i32 4, i32 4>
322 ret <8 x i16> %shuffle
324 define <8 x i16> @shuffle_v8i16_X4404444(<8 x i16> %a, <8 x i16> %b) {
325 ; SSE2-LABEL: @shuffle_v8i16_X4404444
327 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
328 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
329 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
332 ; SSSE3-LABEL: @shuffle_v8i16_X4404444
334 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[{{[0-9]+,[0-9]+}},8,9,8,9,0,1,8,9,8,9,8,9,8,9]
336 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 4, i32 4, i32 0, i32 4, i32 4, i32 4, i32 4>
337 ret <8 x i16> %shuffle
340 define <8 x i16> @shuffle_v8i16_0127XXXX(<8 x i16> %a, <8 x i16> %b) {
341 ; SSE2-LABEL: @shuffle_v8i16_0127XXXX
343 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,1,3]
344 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,7,6,7]
345 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
348 ; SSSE3-LABEL: @shuffle_v8i16_0127XXXX
350 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,2,3,4,5,14,15,{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}}]
352 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 7, i32 undef, i32 undef, i32 undef, i32 undef>
353 ret <8 x i16> %shuffle
356 define <8 x i16> @shuffle_v8i16_XXXX4563(<8 x i16> %a, <8 x i16> %b) {
357 ; SSE2-LABEL: @shuffle_v8i16_XXXX4563
359 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
360 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,3,4,5,6,7]
361 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,0]
364 ; SSSE3-LABEL: @shuffle_v8i16_XXXX4563
366 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}},8,9,10,11,12,13,6,7]
368 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 undef, i32 undef, i32 undef, i32 4, i32 5, i32 6, i32 3>
369 ret <8 x i16> %shuffle
372 define <8 x i16> @shuffle_v8i16_4563XXXX(<8 x i16> %a, <8 x i16> %b) {
373 ; SSE2-LABEL: @shuffle_v8i16_4563XXXX
375 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
376 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,3,4,5,6,7]
377 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,0,2,3]
380 ; SSSE3-LABEL: @shuffle_v8i16_4563XXXX
382 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,10,11,12,13,6,7,{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}}]
384 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
385 ret <8 x i16> %shuffle
388 define <8 x i16> @shuffle_v8i16_01274563(<8 x i16> %a, <8 x i16> %b) {
389 ; SSE2-LABEL: @shuffle_v8i16_01274563
391 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,1,3]
392 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,6,5,4,7]
393 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,1,2]
396 ; SSSE3-LABEL: @shuffle_v8i16_01274563
398 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,2,3,4,5,14,15,8,9,10,11,12,13,6,7]
400 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 7, i32 4, i32 5, i32 6, i32 3>
401 ret <8 x i16> %shuffle
404 define <8 x i16> @shuffle_v8i16_45630127(<8 x i16> %a, <8 x i16> %b) {
405 ; SSE2-LABEL: @shuffle_v8i16_45630127
407 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
408 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,1,2,4,5,6,7]
409 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,0,1,3]
410 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,6,7,5,4]
413 ; SSSE3-LABEL: @shuffle_v8i16_45630127
415 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15]
417 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 3, i32 0, i32 1, i32 2, i32 7>
418 ret <8 x i16> %shuffle
421 define <8 x i16> @shuffle_v8i16_08192a3b(<8 x i16> %a, <8 x i16> %b) {
422 ; ALL-LABEL: @shuffle_v8i16_08192a3b
424 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
426 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11>
427 ret <8 x i16> %shuffle
430 define <8 x i16> @shuffle_v8i16_0c1d2e3f(<8 x i16> %a, <8 x i16> %b) {
431 ; ALL-LABEL: @shuffle_v8i16_0c1d2e3f
433 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,3,2,3]
434 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
436 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 12, i32 1, i32 13, i32 2, i32 14, i32 3, i32 15>
437 ret <8 x i16> %shuffle
440 define <8 x i16> @shuffle_v8i16_4c5d6e7f(<8 x i16> %a, <8 x i16> %b) {
441 ; ALL-LABEL: @shuffle_v8i16_4c5d6e7f
443 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,3,2,3]
444 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,2,3]
445 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
447 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
448 ret <8 x i16> %shuffle
451 define <8 x i16> @shuffle_v8i16_48596a7b(<8 x i16> %a, <8 x i16> %b) {
452 ; ALL-LABEL: @shuffle_v8i16_48596a7b
454 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,2,3]
455 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
457 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 4, i32 8, i32 5, i32 9, i32 6, i32 10, i32 7, i32 11>
458 ret <8 x i16> %shuffle
461 define <8 x i16> @shuffle_v8i16_08196e7f(<8 x i16> %a, <8 x i16> %b) {
462 ; ALL-LABEL: @shuffle_v8i16_08196e7f
464 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[0,3,2,3]
465 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,2,3]
466 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
468 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 6, i32 14, i32 7, i32 15>
469 ret <8 x i16> %shuffle
472 define <8 x i16> @shuffle_v8i16_0c1d6879(<8 x i16> %a, <8 x i16> %b) {
473 ; ALL-LABEL: @shuffle_v8i16_0c1d6879
475 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,0,2,3]
476 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,2,3]
477 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
479 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 12, i32 1, i32 13, i32 6, i32 8, i32 7, i32 9>
480 ret <8 x i16> %shuffle
483 define <8 x i16> @shuffle_v8i16_109832ba(<8 x i16> %a, <8 x i16> %b) {
484 ; ALL-LABEL: @shuffle_v8i16_109832ba
486 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
487 ; ALL-NEXT: pshuflw {{.*}} # xmm1 = xmm0[2,0,3,1,4,5,6,7]
488 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,2,3]
489 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,0,3,1,4,5,6,7]
490 ; ALL-NEXT: punpcklqdq %xmm0, %xmm1
491 ; ALL-NEXT: movdqa %xmm1, %xmm0
493 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 1, i32 0, i32 9, i32 8, i32 3, i32 2, i32 11, i32 10>
494 ret <8 x i16> %shuffle
497 define <8 x i16> @shuffle_v8i16_8091a2b3(<8 x i16> %a, <8 x i16> %b) {
498 ; ALL-LABEL: @shuffle_v8i16_8091a2b3
500 ; ALL-NEXT: punpcklwd %xmm0, %xmm1
501 ; ALL-NEXT: movdqa %xmm1, %xmm0
503 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 8, i32 0, i32 9, i32 1, i32 10, i32 2, i32 11, i32 3>
504 ret <8 x i16> %shuffle
506 define <8 x i16> @shuffle_v8i16_c4d5e6f7(<8 x i16> %a, <8 x i16> %b) {
507 ; ALL-LABEL: @shuffle_v8i16_c4d5e6f7
509 ; ALL-NEXT: pshufd {{.*}} # xmm2 = xmm0[2,3,2,3]
510 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm1[2,3,2,3]
511 ; ALL-NEXT: punpcklwd %xmm2, %xmm0
513 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 12, i32 4, i32 13, i32 5, i32 14, i32 6, i32 15, i32 7>
514 ret <8 x i16> %shuffle
517 define <8 x i16> @shuffle_v8i16_0213cedf(<8 x i16> %a, <8 x i16> %b) {
518 ; ALL-LABEL: @shuffle_v8i16_0213cedf
520 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,1,3,4,5,6,7]
521 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,3,2,3]
522 ; ALL-NEXT: pshuflw {{.*}} # xmm1 = xmm1[0,2,1,3,4,5,6,7]
523 ; ALL-NEXT: punpcklqdq %xmm1, %xmm0
525 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 2, i32 1, i32 3, i32 12, i32 14, i32 13, i32 15>
526 ret <8 x i16> %shuffle
529 define <8 x i16> @shuffle_v8i16_032dXXXX(<8 x i16> %a, <8 x i16> %b) {
530 ; SSE2-LABEL: @shuffle_v8i16_032dXXXX
532 ; SSE2-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
533 ; SSE2-NEXT: punpcklwd %xmm1, %xmm0
534 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,3,4,5,6,7]
535 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
536 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
537 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,1,4,5,6,7]
540 ; SSSE3-LABEL: @shuffle_v8i16_032dXXXX
542 ; SSSE3-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
543 ; SSSE3-NEXT: punpcklwd %xmm1, %xmm0 # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
544 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,12,13,8,9,6,7,{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}}]
546 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 3, i32 2, i32 13, i32 undef, i32 undef, i32 undef, i32 undef>
547 ret <8 x i16> %shuffle
549 define <8 x i16> @shuffle_v8i16_XXXcXXXX(<8 x i16> %a, <8 x i16> %b) {
550 ; ALL-LABEL: @shuffle_v8i16_XXXcXXXX
552 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm1[2,1,2,3]
553 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,1,2,1,4,5,6,7]
555 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 undef, i32 undef, i32 13, i32 undef, i32 undef, i32 undef, i32 undef>
556 ret <8 x i16> %shuffle
559 define <8 x i16> @shuffle_v8i16_012dXXXX(<8 x i16> %a, <8 x i16> %b) {
560 ; SSE2-LABEL: @shuffle_v8i16_012dXXXX
562 ; SSE2-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
563 ; SSE2-NEXT: punpcklwd %xmm1, %xmm0
564 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
565 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
566 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,1,2,3]
567 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,2,0,3,4,5,6,7]
570 ; SSSE3-LABEL: @shuffle_v8i16_012dXXXX
572 ; SSSE3-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
573 ; SSSE3-NEXT: punpcklwd %xmm1, %xmm0 # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
574 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,4,5,8,9,6,7,{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}}]
576 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 13, i32 undef, i32 undef, i32 undef, i32 undef>
577 ret <8 x i16> %shuffle
580 define <8 x i16> @shuffle_v8i16_XXXXcde3(<8 x i16> %a, <8 x i16> %b) {
581 ; SSE2-LABEL: @shuffle_v8i16_XXXXcde3
583 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
584 ; SSE2-NEXT: punpckhwd %xmm0, %xmm1
585 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm1[0,2,2,3,4,5,6,7]
586 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,7,6,7]
587 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,0,2]
590 ; SSSE3-LABEL: @shuffle_v8i16_XXXXcde3
592 ; SSSE3-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
593 ; SSSE3-NEXT: punpckhwd %xmm0, %xmm1 # xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
594 ; SSSE3-NEXT: pshufb {{.*}} # xmm1 = xmm1[{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}},0,1,4,5,8,9,14,15]
595 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
597 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 undef, i32 undef, i32 undef, i32 12, i32 13, i32 14, i32 3>
598 ret <8 x i16> %shuffle
601 define <8 x i16> @shuffle_v8i16_cde3XXXX(<8 x i16> %a, <8 x i16> %b) {
602 ; SSE2-LABEL: @shuffle_v8i16_cde3XXXX
604 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
605 ; SSE2-NEXT: punpckhwd %xmm0, %xmm1
606 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm1[0,2,2,3,4,5,6,7]
607 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,7,6,7]
608 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
611 ; SSSE3-LABEL: @shuffle_v8i16_cde3XXXX
613 ; SSSE3-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
614 ; SSSE3-NEXT: punpckhwd %xmm0, %xmm1 # xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
615 ; SSSE3-NEXT: pshufb {{.*}} # xmm1 = xmm1[0,1,4,5,8,9,14,15,{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}}]
616 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
618 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 12, i32 13, i32 14, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
619 ret <8 x i16> %shuffle
622 define <8 x i16> @shuffle_v8i16_012dcde3(<8 x i16> %a, <8 x i16> %b) {
623 ; SSE2-LABEL: @shuffle_v8i16_012dcde3
625 ; SSE2-NEXT: pshufd {{.*}} # xmm2 = xmm0[0,1,2,1]
626 ; SSE2-NEXT: pshufd {{.*}} # xmm3 = xmm1[2,1,2,3]
627 ; SSE2-NEXT: punpckhwd %xmm2, %xmm1
628 ; SSE2-NEXT: pshuflw {{.*}} # xmm1 = xmm1[0,2,2,3,4,5,6,7]
629 ; SSE2-NEXT: pshufhw {{.*}} # xmm1 = xmm1[0,1,2,3,4,7,6,7]
630 ; SSE2-NEXT: pshufd {{.*}} # xmm1 = xmm1[0,2,2,3]
631 ; SSE2-NEXT: punpcklwd %xmm3, %xmm0
632 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
633 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
634 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,1,2,3]
635 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,2,0,3,4,5,6,7]
636 ; SSE2-NEXT: punpcklqdq %xmm1, %xmm0
639 ; SSSE3-LABEL: @shuffle_v8i16_012dcde3
641 ; SSSE3-NEXT: pshufd {{.*}} # xmm2 = xmm0[0,1,2,1]
642 ; SSSE3-NEXT: pshufd {{.*}} # xmm3 = xmm1[2,1,2,3]
643 ; SSSE3-NEXT: punpckhwd %xmm2, %xmm1 # xmm1 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
644 ; SSSE3-NEXT: pshufb {{.*}} # xmm1 = xmm1[0,1,4,5,8,9,14,15,{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}}]
645 ; SSSE3-NEXT: punpcklwd %xmm3, %xmm0 # xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
646 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,4,5,8,9,6,7,{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}}]
647 ; SSSE3-NEXT: punpcklqdq %xmm1, %xmm0
649 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 13, i32 12, i32 13, i32 14, i32 3>
650 ret <8 x i16> %shuffle