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[0,1,2,3,7,5,6,4]
177 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,2,1]
178 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,2,3,0,4,5,6,7]
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[0,1,2,3,7,5,6,4]
194 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,2,1]
195 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,0,2,3,4,5,6,7]
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[0,1,2,3,7,5,6,4]
211 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,1,2]
212 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,2,3,0,4,5,6,7]
215 ; SSSE3-LABEL: @shuffle_v8i16_26401375
217 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[4,5,12,13,8,9,0,1,2,3,6,7,14,15,10,11]
219 %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>
220 ret <8 x i16> %shuffle
223 define <8 x i16> @shuffle_v8i16_66751643(<8 x i16> %a, <8 x i16> %b) {
224 ; SSE2-LABEL: @shuffle_v8i16_66751643
226 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[3,1,2,3,4,5,6,7]
227 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,5,7]
228 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,2,0]
229 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,1,3,2,4,5,6,7]
230 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,5,4,6]
233 ; SSSE3-LABEL: @shuffle_v8i16_66751643
235 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[12,13,12,13,14,15,10,11,2,3,12,13,8,9,6,7]
237 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 6, i32 6, i32 7, i32 5, i32 1, i32 6, i32 4, i32 3>
238 ret <8 x i16> %shuffle
241 define <8 x i16> @shuffle_v8i16_60514754(<8 x i16> %a, <8 x i16> %b) {
242 ; SSE2-LABEL: @shuffle_v8i16_60514754
244 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,6,5,4,7]
245 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
246 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,0,3,1,4,5,6,7]
247 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,6,7,5,6]
250 ; SSSE3-LABEL: @shuffle_v8i16_60514754
252 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[12,13,0,1,10,11,2,3,8,9,14,15,10,11,8,9]
254 %shuffle = shufflevector <8 x i16> %a, <8 x i16> undef, <8 x i32> <i32 6, i32 0, i32 5, i32 1, i32 4, i32 7, i32 5, i32 4>
255 ret <8 x i16> %shuffle
258 define <8 x i16> @shuffle_v8i16_00444444(<8 x i16> %a, <8 x i16> %b) {
259 ; SSE2-LABEL: @shuffle_v8i16_00444444
261 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
262 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,2,2,4,5,6,7]
263 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
266 ; SSSE3-LABEL: @shuffle_v8i16_00444444
268 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9]
270 %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>
271 ret <8 x i16> %shuffle
273 define <8 x i16> @shuffle_v8i16_44004444(<8 x i16> %a, <8 x i16> %b) {
274 ; SSE2-LABEL: @shuffle_v8i16_44004444
276 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
277 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,2,0,0,4,5,6,7]
278 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
281 ; SSSE3-LABEL: @shuffle_v8i16_44004444
283 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9]
285 %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>
286 ret <8 x i16> %shuffle
288 define <8 x i16> @shuffle_v8i16_04404444(<8 x i16> %a, <8 x i16> %b) {
289 ; SSE2-LABEL: @shuffle_v8i16_04404444
291 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
292 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
293 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
296 ; SSSE3-LABEL: @shuffle_v8i16_04404444
298 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
300 %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>
301 ret <8 x i16> %shuffle
303 define <8 x i16> @shuffle_v8i16_04400000(<8 x i16> %a, <8 x i16> %b) {
304 ; SSE2-LABEL: @shuffle_v8i16_04400000
306 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,0,3]
307 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
308 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
311 ; SSSE3-LABEL: @shuffle_v8i16_04400000
313 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1]
315 %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>
316 ret <8 x i16> %shuffle
318 define <8 x i16> @shuffle_v8i16_04404567(<8 x i16> %a, <8 x i16> %b) {
319 ; ALL-LABEL: @shuffle_v8i16_04404567
321 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
322 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
324 %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>
325 ret <8 x i16> %shuffle
328 define <8 x i16> @shuffle_v8i16_0X444444(<8 x i16> %a, <8 x i16> %b) {
329 ; SSE2-LABEL: @shuffle_v8i16_0X444444
331 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
332 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,1,2,2,4,5,6,7]
333 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
336 ; SSSE3-LABEL: @shuffle_v8i16_0X444444
338 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,{{[0-9]+,[0-9]+}},8,9,8,9,8,9,8,9,8,9,8,9]
340 %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>
341 ret <8 x i16> %shuffle
343 define <8 x i16> @shuffle_v8i16_44X04444(<8 x i16> %a, <8 x i16> %b) {
344 ; SSE2-LABEL: @shuffle_v8i16_44X04444
346 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
347 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,2,2,0,4,5,6,7]
348 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
351 ; SSSE3-LABEL: @shuffle_v8i16_44X04444
353 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,8,9,{{[0-9]+,[0-9]+}},0,1,8,9,8,9,8,9,8,9]
355 %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>
356 ret <8 x i16> %shuffle
358 define <8 x i16> @shuffle_v8i16_X4404444(<8 x i16> %a, <8 x i16> %b) {
359 ; SSE2-LABEL: @shuffle_v8i16_X4404444
361 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
362 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
363 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
366 ; SSSE3-LABEL: @shuffle_v8i16_X4404444
368 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[{{[0-9]+,[0-9]+}},8,9,8,9,0,1,8,9,8,9,8,9,8,9]
370 %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>
371 ret <8 x i16> %shuffle
374 define <8 x i16> @shuffle_v8i16_0127XXXX(<8 x i16> %a, <8 x i16> %b) {
375 ; SSE2-LABEL: @shuffle_v8i16_0127XXXX
377 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,1,3]
378 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,7,6,7]
379 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
382 ; SSSE3-LABEL: @shuffle_v8i16_0127XXXX
384 ; 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]+}}]
386 %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>
387 ret <8 x i16> %shuffle
390 define <8 x i16> @shuffle_v8i16_XXXX4563(<8 x i16> %a, <8 x i16> %b) {
391 ; SSE2-LABEL: @shuffle_v8i16_XXXX4563
393 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
394 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,3,4,5,6,7]
395 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,0]
398 ; SSSE3-LABEL: @shuffle_v8i16_XXXX4563
400 ; 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]
402 %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>
403 ret <8 x i16> %shuffle
406 define <8 x i16> @shuffle_v8i16_4563XXXX(<8 x i16> %a, <8 x i16> %b) {
407 ; SSE2-LABEL: @shuffle_v8i16_4563XXXX
409 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
410 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,3,4,5,6,7]
411 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,0,2,3]
414 ; SSSE3-LABEL: @shuffle_v8i16_4563XXXX
416 ; 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]+}}]
418 %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>
419 ret <8 x i16> %shuffle
422 define <8 x i16> @shuffle_v8i16_01274563(<8 x i16> %a, <8 x i16> %b) {
423 ; SSE2-LABEL: @shuffle_v8i16_01274563
425 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,1,3]
426 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,6,5,4,7]
427 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,1,2]
430 ; SSSE3-LABEL: @shuffle_v8i16_01274563
432 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,2,3,4,5,14,15,8,9,10,11,12,13,6,7]
434 %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>
435 ret <8 x i16> %shuffle
438 define <8 x i16> @shuffle_v8i16_45630127(<8 x i16> %a, <8 x i16> %b) {
439 ; SSE2-LABEL: @shuffle_v8i16_45630127
441 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
442 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,1,4,5,6,7]
443 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,0,3,1]
446 ; SSSE3-LABEL: @shuffle_v8i16_45630127
448 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15]
450 %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>
451 ret <8 x i16> %shuffle
454 define <8 x i16> @shuffle_v8i16_37102735(<8 x i16> %a, <8 x i16> %b) {
455 ; SSE2-LABEL: @shuffle_v8i16_37102735
457 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,5,7]
458 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,1,3]
459 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,5,6,4]
460 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
461 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[3,2,1,0,4,5,6,7]
462 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,4,5,6]
465 ; SSSE3-LABEL: @shuffle_v8i16_37102735
467 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[6,7,14,15,2,3,0,1,4,5,14,15,6,7,10,11]
469 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 3, i32 7, i32 1, i32 0, i32 2, i32 7, i32 3, i32 5>
470 ret <8 x i16> %shuffle
473 define <8 x i16> @shuffle_v8i16_08192a3b(<8 x i16> %a, <8 x i16> %b) {
474 ; ALL-LABEL: @shuffle_v8i16_08192a3b
476 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
478 %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>
479 ret <8 x i16> %shuffle
482 define <8 x i16> @shuffle_v8i16_0c1d2e3f(<8 x i16> %a, <8 x i16> %b) {
483 ; ALL-LABEL: @shuffle_v8i16_0c1d2e3f
485 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,3,2,3]
486 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
488 %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>
489 ret <8 x i16> %shuffle
492 define <8 x i16> @shuffle_v8i16_4c5d6e7f(<8 x i16> %a, <8 x i16> %b) {
493 ; ALL-LABEL: @shuffle_v8i16_4c5d6e7f
495 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,3,2,3]
496 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,2,3]
497 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
499 %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>
500 ret <8 x i16> %shuffle
503 define <8 x i16> @shuffle_v8i16_48596a7b(<8 x i16> %a, <8 x i16> %b) {
504 ; ALL-LABEL: @shuffle_v8i16_48596a7b
506 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,2,3]
507 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
509 %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>
510 ret <8 x i16> %shuffle
513 define <8 x i16> @shuffle_v8i16_08196e7f(<8 x i16> %a, <8 x i16> %b) {
514 ; ALL-LABEL: @shuffle_v8i16_08196e7f
516 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[0,3,2,3]
517 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,2,3]
518 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
520 %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>
521 ret <8 x i16> %shuffle
524 define <8 x i16> @shuffle_v8i16_0c1d6879(<8 x i16> %a, <8 x i16> %b) {
525 ; ALL-LABEL: @shuffle_v8i16_0c1d6879
527 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,0,2,3]
528 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,2,3]
529 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
531 %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>
532 ret <8 x i16> %shuffle
535 define <8 x i16> @shuffle_v8i16_109832ba(<8 x i16> %a, <8 x i16> %b) {
536 ; ALL-LABEL: @shuffle_v8i16_109832ba
538 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
539 ; ALL-NEXT: pshuflw {{.*}} # xmm1 = xmm0[2,0,3,1,4,5,6,7]
540 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,2,3]
541 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,0,3,1,4,5,6,7]
542 ; ALL-NEXT: punpcklqdq %xmm0, %xmm1
543 ; ALL-NEXT: movdqa %xmm1, %xmm0
545 %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>
546 ret <8 x i16> %shuffle
549 define <8 x i16> @shuffle_v8i16_8091a2b3(<8 x i16> %a, <8 x i16> %b) {
550 ; ALL-LABEL: @shuffle_v8i16_8091a2b3
552 ; ALL-NEXT: punpcklwd %xmm0, %xmm1
553 ; ALL-NEXT: movdqa %xmm1, %xmm0
555 %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>
556 ret <8 x i16> %shuffle
558 define <8 x i16> @shuffle_v8i16_c4d5e6f7(<8 x i16> %a, <8 x i16> %b) {
559 ; ALL-LABEL: @shuffle_v8i16_c4d5e6f7
561 ; ALL-NEXT: pshufd {{.*}} # xmm2 = xmm0[2,3,2,3]
562 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm1[2,3,2,3]
563 ; ALL-NEXT: punpcklwd %xmm2, %xmm0
565 %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>
566 ret <8 x i16> %shuffle
569 define <8 x i16> @shuffle_v8i16_0213cedf(<8 x i16> %a, <8 x i16> %b) {
570 ; ALL-LABEL: @shuffle_v8i16_0213cedf
572 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,1,3,4,5,6,7]
573 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,3,2,3]
574 ; ALL-NEXT: pshuflw {{.*}} # xmm1 = xmm1[0,2,1,3,4,5,6,7]
575 ; ALL-NEXT: punpcklqdq %xmm1, %xmm0
577 %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>
578 ret <8 x i16> %shuffle
581 define <8 x i16> @shuffle_v8i16_443aXXXX(<8 x i16> %a, <8 x i16> %b) {
582 ; SSE2-LABEL: @shuffle_v8i16_443aXXXX
584 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,1,2,3]
585 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,2,3,4,5,6,7]
586 ; SSE2-NEXT: punpcklwd %xmm1, %xmm0 # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
587 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,3,4,5,6,7]
588 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,6,5,6,7]
589 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
592 ; SSSE3-LABEL: @shuffle_v8i16_443aXXXX
594 ; SSSE3-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,1,2,3]
595 ; SSSE3-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,2,3,4,5,6,7]
596 ; SSSE3-NEXT: punpcklwd %xmm1, %xmm0 # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
597 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,4,5,12,13,10,11,{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}}]
599 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 4, i32 4, i32 3, i32 10, i32 undef, i32 undef, i32 undef, i32 undef>
600 ret <8 x i16> %shuffle
603 define <8 x i16> @shuffle_v8i16_032dXXXX(<8 x i16> %a, <8 x i16> %b) {
604 ; SSE2-LABEL: @shuffle_v8i16_032dXXXX
606 ; SSE2-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
607 ; SSE2-NEXT: punpcklwd %xmm1, %xmm0
608 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,3,4,5,6,7]
609 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
610 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
611 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,1,4,5,6,7]
614 ; SSSE3-LABEL: @shuffle_v8i16_032dXXXX
616 ; SSSE3-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
617 ; SSSE3-NEXT: punpcklwd %xmm1, %xmm0 # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
618 ; 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]+}}]
620 %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>
621 ret <8 x i16> %shuffle
623 define <8 x i16> @shuffle_v8i16_XXXcXXXX(<8 x i16> %a, <8 x i16> %b) {
624 ; ALL-LABEL: @shuffle_v8i16_XXXcXXXX
626 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm1[2,1,2,3]
627 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,1,2,1,4,5,6,7]
629 %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>
630 ret <8 x i16> %shuffle
633 define <8 x i16> @shuffle_v8i16_012dXXXX(<8 x i16> %a, <8 x i16> %b) {
634 ; SSE2-LABEL: @shuffle_v8i16_012dXXXX
636 ; SSE2-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
637 ; SSE2-NEXT: punpcklwd %xmm1, %xmm0
638 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
639 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
640 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,1,2,3]
641 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,2,0,3,4,5,6,7]
644 ; SSSE3-LABEL: @shuffle_v8i16_012dXXXX
646 ; SSSE3-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
647 ; SSSE3-NEXT: punpcklwd %xmm1, %xmm0 # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
648 ; 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]+}}]
650 %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>
651 ret <8 x i16> %shuffle
654 define <8 x i16> @shuffle_v8i16_XXXXcde3(<8 x i16> %a, <8 x i16> %b) {
655 ; SSE2-LABEL: @shuffle_v8i16_XXXXcde3
657 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
658 ; SSE2-NEXT: punpckhwd %xmm0, %xmm1
659 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm1[0,2,2,3,4,5,6,7]
660 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,7,6,7]
661 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,0,2]
664 ; SSSE3-LABEL: @shuffle_v8i16_XXXXcde3
666 ; SSSE3-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
667 ; SSSE3-NEXT: punpckhwd %xmm0, %xmm1 # xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
668 ; 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]
669 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
671 %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>
672 ret <8 x i16> %shuffle
675 define <8 x i16> @shuffle_v8i16_cde3XXXX(<8 x i16> %a, <8 x i16> %b) {
676 ; SSE2-LABEL: @shuffle_v8i16_cde3XXXX
678 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
679 ; SSE2-NEXT: punpckhwd %xmm0, %xmm1
680 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm1[0,2,2,3,4,5,6,7]
681 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,7,6,7]
682 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
685 ; SSSE3-LABEL: @shuffle_v8i16_cde3XXXX
687 ; SSSE3-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
688 ; SSSE3-NEXT: punpckhwd %xmm0, %xmm1 # xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
689 ; 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]+}}]
690 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
692 %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>
693 ret <8 x i16> %shuffle
696 define <8 x i16> @shuffle_v8i16_012dcde3(<8 x i16> %a, <8 x i16> %b) {
697 ; SSE2-LABEL: @shuffle_v8i16_012dcde3
699 ; SSE2-NEXT: pshufd {{.*}} # xmm2 = xmm0[0,1,2,1]
700 ; SSE2-NEXT: pshufd {{.*}} # xmm3 = xmm1[2,1,2,3]
701 ; SSE2-NEXT: punpckhwd %xmm2, %xmm1
702 ; SSE2-NEXT: pshuflw {{.*}} # xmm1 = xmm1[0,2,2,3,4,5,6,7]
703 ; SSE2-NEXT: pshufhw {{.*}} # xmm1 = xmm1[0,1,2,3,4,7,6,7]
704 ; SSE2-NEXT: pshufd {{.*}} # xmm1 = xmm1[0,2,2,3]
705 ; SSE2-NEXT: punpcklwd %xmm3, %xmm0
706 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
707 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
708 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,1,2,3]
709 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,2,0,3,4,5,6,7]
710 ; SSE2-NEXT: punpcklqdq %xmm1, %xmm0
713 ; SSSE3-LABEL: @shuffle_v8i16_012dcde3
715 ; SSSE3-NEXT: pshufd {{.*}} # xmm2 = xmm0[0,1,2,1]
716 ; SSSE3-NEXT: pshufd {{.*}} # xmm3 = xmm1[2,1,2,3]
717 ; SSSE3-NEXT: punpckhwd %xmm2, %xmm1 # xmm1 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
718 ; 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]+}}]
719 ; SSSE3-NEXT: punpcklwd %xmm3, %xmm0 # xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
720 ; 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]+}}]
721 ; SSSE3-NEXT: punpcklqdq %xmm1, %xmm0
723 %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>
724 ret <8 x i16> %shuffle
727 define <8 x i16> @shuffle_v8i16_XXX1X579(<8 x i16> %a, <8 x i16> %b) {
728 ; SSE2-LABEL: @shuffle_v8i16_XXX1X579
730 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,5,6,7]
731 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
732 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,1,3,2,4,5,6,7]
733 ; SSE2-NEXT: punpcklwd {{.*}} # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
734 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
735 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
736 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,1,2,2,4,5,6,7]
737 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,5,7]
740 ; SSSE3-LABEL: @shuffle_v8i16_XXX1X579
742 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[{{[0-9]+,[0-9]+}},2,3,10,11,14,15,{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}}]
743 ; SSSE3-NEXT: punpcklwd {{.*}} # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
744 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[{{[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+,[0-9]+}},4,5,{{[0-9]+,[0-9]+}},8,9,12,13,6,7]
746 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 undef, i32 undef, i32 1, i32 undef, i32 5, i32 7, i32 9>
747 ret <8 x i16> %shuffle
750 define <8 x i16> @shuffle_v8i16_XX4X8acX(<8 x i16> %a, <8 x i16> %b) {
751 ; SSE2-LABEL: @shuffle_v8i16_XX4X8acX
753 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,1,2,3]
754 ; SSE2-NEXT: pshuflw {{.*}} # xmm1 = xmm1[0,2,2,3,4,5,6,7]
755 ; SSE2-NEXT: pshufd {{.*}} # xmm1 = xmm1[0,2,2,3]
756 ; SSE2-NEXT: punpcklwd {{.*}} # xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
757 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm1[0,1,2,0,4,5,6,7]
758 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
759 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,1,1,3,4,5,6,7]
760 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,6,4,7]
763 ; SSSE3-LABEL: @shuffle_v8i16_XX4X8acX
765 ; SSSE3-NEXT: pshufd {{.*}} # [[X:xmm[0-9]+]] = xmm0[2,1,2,3]
766 ; SSSE3-NEXT: pshuflw {{.*}} # xmm0 = xmm1[0,2,2,3,4,5,6,7]
767 ; SSSE3-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
768 ; SSSE3-NEXT: punpcklwd {{.*}} # xmm0 = xmm0[0],[[X]][0],xmm0[1],[[X]][1],xmm0[2],[[X]][2],xmm0[3],[[X]][3]
769 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[{{[0-9]+,[0-9]+,[0-9]+,[0-9]+}},2,3,{{[0-9]+,[0-9]+}},0,1,4,5,8,9,{{[0-9]+,[0-9]+}}]
771 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 undef, i32 4, i32 undef, i32 8, i32 10, i32 12, i32 undef>
772 ret <8 x i16> %shuffle
775 define <8 x i16> @shuffle_v8i16_8zzzzzzz(i16 %i) {
776 ; ALL-LABEL: @shuffle_v8i16_8zzzzzzz
778 ; ALL-NEXT: movzwl {{.*}}, %[[R:.*]]
779 ; ALL-NEXT: movd %[[R]], %xmm0
781 %a = insertelement <8 x i16> undef, i16 %i, i32 0
782 %shuffle = shufflevector <8 x i16> zeroinitializer, <8 x i16> %a, <8 x i32> <i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
783 ret <8 x i16> %shuffle
786 define <8 x i16> @shuffle_v8i16_z8zzzzzz(i16 %i) {
787 ; ALL-LABEL: @shuffle_v8i16_z8zzzzzz
789 ; ALL-NEXT: movzwl {{.*}}, %[[R:.*]]
790 ; ALL-NEXT: movd %[[R]], %xmm0
791 ; ALL-NEXT: pslldq $2, %xmm0
793 %a = insertelement <8 x i16> undef, i16 %i, i32 0
794 %shuffle = shufflevector <8 x i16> zeroinitializer, <8 x i16> %a, <8 x i32> <i32 2, i32 8, i32 3, i32 7, i32 6, i32 5, i32 4, i32 3>
795 ret <8 x i16> %shuffle
798 define <8 x i16> @shuffle_v8i16_zzzzz8zz(i16 %i) {
799 ; ALL-LABEL: @shuffle_v8i16_zzzzz8zz
801 ; ALL-NEXT: movzwl {{.*}}, %[[R:.*]]
802 ; ALL-NEXT: movd %[[R]], %xmm0
803 ; ALL-NEXT: pslldq $10, %xmm0
805 %a = insertelement <8 x i16> undef, i16 %i, i32 0
806 %shuffle = shufflevector <8 x i16> zeroinitializer, <8 x i16> %a, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 8, i32 0, i32 0>
807 ret <8 x i16> %shuffle
810 define <8 x i16> @shuffle_v8i16_zuuzuuz8(i16 %i) {
811 ; ALL-LABEL: @shuffle_v8i16_zuuzuuz8
813 ; ALL-NEXT: movzwl {{.*}}, %[[R:.*]]
814 ; ALL-NEXT: movd %[[R]], %xmm0
815 ; ALL-NEXT: pslldq $14, %xmm0
817 %a = insertelement <8 x i16> undef, i16 %i, i32 0
818 %shuffle = shufflevector <8 x i16> zeroinitializer, <8 x i16> %a, <8 x i32> <i32 0, i32 undef, i32 undef, i32 3, i32 undef, i32 undef, i32 6, i32 8>
819 ret <8 x i16> %shuffle
822 define <8 x i16> @shuffle_v8i16_zzBzzzzz(i16 %i) {
823 ; ALL-LABEL: @shuffle_v8i16_zzBzzzzz
825 ; ALL-NEXT: movzwl {{.*}}, %[[R:.*]]
826 ; ALL-NEXT: movd %[[R]], %xmm0
827 ; ALL-NEXT: pslldq $4, %xmm0
829 %a = insertelement <8 x i16> undef, i16 %i, i32 3
830 %shuffle = shufflevector <8 x i16> zeroinitializer, <8 x i16> %a, <8 x i32> <i32 0, i32 1, i32 11, i32 3, i32 4, i32 5, i32 6, i32 7>
831 ret <8 x i16> %shuffle