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[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_00444444(<8 x i16> %a, <8 x i16> %b) {
224 ; SSE2-LABEL: @shuffle_v8i16_00444444
226 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
227 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,2,2,4,5,6,7]
228 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
231 ; SSSE3-LABEL: @shuffle_v8i16_00444444
233 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9]
235 %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>
236 ret <8 x i16> %shuffle
238 define <8 x i16> @shuffle_v8i16_44004444(<8 x i16> %a, <8 x i16> %b) {
239 ; SSE2-LABEL: @shuffle_v8i16_44004444
241 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
242 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,2,0,0,4,5,6,7]
243 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
246 ; SSSE3-LABEL: @shuffle_v8i16_44004444
248 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9]
250 %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>
251 ret <8 x i16> %shuffle
253 define <8 x i16> @shuffle_v8i16_04404444(<8 x i16> %a, <8 x i16> %b) {
254 ; SSE2-LABEL: @shuffle_v8i16_04404444
256 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
257 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
258 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
261 ; SSSE3-LABEL: @shuffle_v8i16_04404444
263 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
265 %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>
266 ret <8 x i16> %shuffle
268 define <8 x i16> @shuffle_v8i16_04400000(<8 x i16> %a, <8 x i16> %b) {
269 ; SSE2-LABEL: @shuffle_v8i16_04400000
271 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,0,3]
272 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
273 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
276 ; SSSE3-LABEL: @shuffle_v8i16_04400000
278 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1]
280 %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>
281 ret <8 x i16> %shuffle
283 define <8 x i16> @shuffle_v8i16_04404567(<8 x i16> %a, <8 x i16> %b) {
284 ; ALL-LABEL: @shuffle_v8i16_04404567
286 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
287 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
289 %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>
290 ret <8 x i16> %shuffle
293 define <8 x i16> @shuffle_v8i16_0X444444(<8 x i16> %a, <8 x i16> %b) {
294 ; SSE2-LABEL: @shuffle_v8i16_0X444444
296 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
297 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,1,2,2,4,5,6,7]
298 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
301 ; SSSE3-LABEL: @shuffle_v8i16_0X444444
303 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,{{[0-9]+,[0-9]+}},8,9,8,9,8,9,8,9,8,9,8,9]
305 %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>
306 ret <8 x i16> %shuffle
308 define <8 x i16> @shuffle_v8i16_44X04444(<8 x i16> %a, <8 x i16> %b) {
309 ; SSE2-LABEL: @shuffle_v8i16_44X04444
311 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
312 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,2,2,0,4,5,6,7]
313 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
316 ; SSSE3-LABEL: @shuffle_v8i16_44X04444
318 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,8,9,{{[0-9]+,[0-9]+}},0,1,8,9,8,9,8,9,8,9]
320 %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>
321 ret <8 x i16> %shuffle
323 define <8 x i16> @shuffle_v8i16_X4404444(<8 x i16> %a, <8 x i16> %b) {
324 ; SSE2-LABEL: @shuffle_v8i16_X4404444
326 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
327 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,0,4,5,6,7]
328 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
331 ; SSSE3-LABEL: @shuffle_v8i16_X4404444
333 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[{{[0-9]+,[0-9]+}},8,9,8,9,0,1,8,9,8,9,8,9,8,9]
335 %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>
336 ret <8 x i16> %shuffle
339 define <8 x i16> @shuffle_v8i16_0127XXXX(<8 x i16> %a, <8 x i16> %b) {
340 ; SSE2-LABEL: @shuffle_v8i16_0127XXXX
342 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,1,3]
343 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,7,6,7]
344 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
347 ; SSSE3-LABEL: @shuffle_v8i16_0127XXXX
349 ; 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]+}}]
351 %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>
352 ret <8 x i16> %shuffle
355 define <8 x i16> @shuffle_v8i16_XXXX4563(<8 x i16> %a, <8 x i16> %b) {
356 ; SSE2-LABEL: @shuffle_v8i16_XXXX4563
358 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
359 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,3,4,5,6,7]
360 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,0]
363 ; SSSE3-LABEL: @shuffle_v8i16_XXXX4563
365 ; 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]
367 %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>
368 ret <8 x i16> %shuffle
371 define <8 x i16> @shuffle_v8i16_4563XXXX(<8 x i16> %a, <8 x i16> %b) {
372 ; SSE2-LABEL: @shuffle_v8i16_4563XXXX
374 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
375 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,3,4,5,6,7]
376 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,0,2,3]
379 ; SSSE3-LABEL: @shuffle_v8i16_4563XXXX
381 ; 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]+}}]
383 %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>
384 ret <8 x i16> %shuffle
387 define <8 x i16> @shuffle_v8i16_01274563(<8 x i16> %a, <8 x i16> %b) {
388 ; SSE2-LABEL: @shuffle_v8i16_01274563
390 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,1,3]
391 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,6,5,4,7]
392 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,1,2]
395 ; SSSE3-LABEL: @shuffle_v8i16_01274563
397 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,2,3,4,5,14,15,8,9,10,11,12,13,6,7]
399 %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>
400 ret <8 x i16> %shuffle
403 define <8 x i16> @shuffle_v8i16_45630127(<8 x i16> %a, <8 x i16> %b) {
404 ; SSE2-LABEL: @shuffle_v8i16_45630127
406 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
407 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,1,2,4,5,6,7]
408 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,0,1,3]
409 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,6,7,5,4]
412 ; SSSE3-LABEL: @shuffle_v8i16_45630127
414 ; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15]
416 %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>
417 ret <8 x i16> %shuffle
420 define <8 x i16> @shuffle_v8i16_08192a3b(<8 x i16> %a, <8 x i16> %b) {
421 ; ALL-LABEL: @shuffle_v8i16_08192a3b
423 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
425 %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>
426 ret <8 x i16> %shuffle
429 define <8 x i16> @shuffle_v8i16_0c1d2e3f(<8 x i16> %a, <8 x i16> %b) {
430 ; ALL-LABEL: @shuffle_v8i16_0c1d2e3f
432 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,3,2,3]
433 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
435 %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>
436 ret <8 x i16> %shuffle
439 define <8 x i16> @shuffle_v8i16_4c5d6e7f(<8 x i16> %a, <8 x i16> %b) {
440 ; ALL-LABEL: @shuffle_v8i16_4c5d6e7f
442 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,3,2,3]
443 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,2,3]
444 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
446 %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>
447 ret <8 x i16> %shuffle
450 define <8 x i16> @shuffle_v8i16_48596a7b(<8 x i16> %a, <8 x i16> %b) {
451 ; ALL-LABEL: @shuffle_v8i16_48596a7b
453 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,2,3]
454 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
456 %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>
457 ret <8 x i16> %shuffle
460 define <8 x i16> @shuffle_v8i16_08196e7f(<8 x i16> %a, <8 x i16> %b) {
461 ; ALL-LABEL: @shuffle_v8i16_08196e7f
463 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[0,3,2,3]
464 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,2,3]
465 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
467 %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>
468 ret <8 x i16> %shuffle
471 define <8 x i16> @shuffle_v8i16_0c1d6879(<8 x i16> %a, <8 x i16> %b) {
472 ; ALL-LABEL: @shuffle_v8i16_0c1d6879
474 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,0,2,3]
475 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,3,2,3]
476 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
478 %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>
479 ret <8 x i16> %shuffle
482 define <8 x i16> @shuffle_v8i16_109832ba(<8 x i16> %a, <8 x i16> %b) {
483 ; ALL-LABEL: @shuffle_v8i16_109832ba
485 ; ALL-NEXT: punpcklwd %xmm1, %xmm0
486 ; ALL-NEXT: pshuflw {{.*}} # xmm1 = xmm0[2,0,3,1,4,5,6,7]
487 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,3,2,3]
488 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[2,0,3,1,4,5,6,7]
489 ; ALL-NEXT: punpcklqdq %xmm0, %xmm1
490 ; ALL-NEXT: movdqa %xmm1, %xmm0
492 %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>
493 ret <8 x i16> %shuffle
496 define <8 x i16> @shuffle_v8i16_8091a2b3(<8 x i16> %a, <8 x i16> %b) {
497 ; ALL-LABEL: @shuffle_v8i16_8091a2b3
499 ; ALL-NEXT: punpcklwd %xmm0, %xmm1
500 ; ALL-NEXT: movdqa %xmm1, %xmm0
502 %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>
503 ret <8 x i16> %shuffle
505 define <8 x i16> @shuffle_v8i16_c4d5e6f7(<8 x i16> %a, <8 x i16> %b) {
506 ; ALL-LABEL: @shuffle_v8i16_c4d5e6f7
508 ; ALL-NEXT: pshufd {{.*}} # xmm2 = xmm0[2,3,2,3]
509 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm1[2,3,2,3]
510 ; ALL-NEXT: punpcklwd %xmm2, %xmm0
512 %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>
513 ret <8 x i16> %shuffle
516 define <8 x i16> @shuffle_v8i16_0213cedf(<8 x i16> %a, <8 x i16> %b) {
517 ; ALL-LABEL: @shuffle_v8i16_0213cedf
519 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,1,3,4,5,6,7]
520 ; ALL-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,3,2,3]
521 ; ALL-NEXT: pshuflw {{.*}} # xmm1 = xmm1[0,2,1,3,4,5,6,7]
522 ; ALL-NEXT: punpcklqdq %xmm1, %xmm0
524 %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>
525 ret <8 x i16> %shuffle
528 define <8 x i16> @shuffle_v8i16_032dXXXX(<8 x i16> %a, <8 x i16> %b) {
529 ; SSE2-LABEL: @shuffle_v8i16_032dXXXX
531 ; SSE2-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
532 ; SSE2-NEXT: punpcklwd %xmm1, %xmm0
533 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,3,4,5,6,7]
534 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
535 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
536 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,3,2,1,4,5,6,7]
539 ; SSSE3-LABEL: @shuffle_v8i16_032dXXXX
541 ; SSSE3-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
542 ; SSSE3-NEXT: punpcklwd %xmm1, %xmm0 # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
543 ; 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]+}}]
545 %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>
546 ret <8 x i16> %shuffle
548 define <8 x i16> @shuffle_v8i16_XXXcXXXX(<8 x i16> %a, <8 x i16> %b) {
549 ; ALL-LABEL: @shuffle_v8i16_XXXcXXXX
551 ; ALL-NEXT: pshufd {{.*}} # xmm0 = xmm1[2,1,2,3]
552 ; ALL-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,1,2,1,4,5,6,7]
554 %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>
555 ret <8 x i16> %shuffle
558 define <8 x i16> @shuffle_v8i16_012dXXXX(<8 x i16> %a, <8 x i16> %b) {
559 ; SSE2-LABEL: @shuffle_v8i16_012dXXXX
561 ; SSE2-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
562 ; SSE2-NEXT: punpcklwd %xmm1, %xmm0
563 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
564 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
565 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,1,2,3]
566 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,2,0,3,4,5,6,7]
569 ; SSSE3-LABEL: @shuffle_v8i16_012dXXXX
571 ; SSSE3-NEXT: pshufd {{.*}} # xmm1 = xmm1[2,1,2,3]
572 ; SSSE3-NEXT: punpcklwd %xmm1, %xmm0 # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
573 ; 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]+}}]
575 %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>
576 ret <8 x i16> %shuffle
579 define <8 x i16> @shuffle_v8i16_XXXXcde3(<8 x i16> %a, <8 x i16> %b) {
580 ; SSE2-LABEL: @shuffle_v8i16_XXXXcde3
582 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
583 ; SSE2-NEXT: punpckhwd %xmm0, %xmm1
584 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm1[0,2,2,3,4,5,6,7]
585 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,7,6,7]
586 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,0,2]
589 ; SSSE3-LABEL: @shuffle_v8i16_XXXXcde3
591 ; SSSE3-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
592 ; SSSE3-NEXT: punpckhwd %xmm0, %xmm1 # xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
593 ; 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]
594 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
596 %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>
597 ret <8 x i16> %shuffle
600 define <8 x i16> @shuffle_v8i16_cde3XXXX(<8 x i16> %a, <8 x i16> %b) {
601 ; SSE2-LABEL: @shuffle_v8i16_cde3XXXX
603 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
604 ; SSE2-NEXT: punpckhwd %xmm0, %xmm1
605 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm1[0,2,2,3,4,5,6,7]
606 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,7,6,7]
607 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
610 ; SSSE3-LABEL: @shuffle_v8i16_cde3XXXX
612 ; SSSE3-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
613 ; SSSE3-NEXT: punpckhwd %xmm0, %xmm1 # xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
614 ; 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]+}}]
615 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
617 %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>
618 ret <8 x i16> %shuffle
621 define <8 x i16> @shuffle_v8i16_012dcde3(<8 x i16> %a, <8 x i16> %b) {
622 ; SSE2-LABEL: @shuffle_v8i16_012dcde3
624 ; SSE2-NEXT: pshufd {{.*}} # xmm2 = xmm0[0,1,2,1]
625 ; SSE2-NEXT: pshufd {{.*}} # xmm3 = xmm1[2,1,2,3]
626 ; SSE2-NEXT: punpckhwd %xmm2, %xmm1
627 ; SSE2-NEXT: pshuflw {{.*}} # xmm1 = xmm1[0,2,2,3,4,5,6,7]
628 ; SSE2-NEXT: pshufhw {{.*}} # xmm1 = xmm1[0,1,2,3,4,7,6,7]
629 ; SSE2-NEXT: pshufd {{.*}} # xmm1 = xmm1[0,2,2,3]
630 ; SSE2-NEXT: punpcklwd %xmm3, %xmm0
631 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[3,1,2,0]
632 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
633 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[2,1,2,3]
634 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,2,0,3,4,5,6,7]
635 ; SSE2-NEXT: punpcklqdq %xmm1, %xmm0
638 ; SSSE3-LABEL: @shuffle_v8i16_012dcde3
640 ; SSSE3-NEXT: pshufd {{.*}} # xmm2 = xmm0[0,1,2,1]
641 ; SSSE3-NEXT: pshufd {{.*}} # xmm3 = xmm1[2,1,2,3]
642 ; SSSE3-NEXT: punpckhwd %xmm2, %xmm1 # xmm1 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
643 ; 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]+}}]
644 ; SSSE3-NEXT: punpcklwd %xmm3, %xmm0 # xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
645 ; 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]+}}]
646 ; SSSE3-NEXT: punpcklqdq %xmm1, %xmm0
648 %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>
649 ret <8 x i16> %shuffle
652 define <8 x i16> @shuffle_v8i16_XXX1X579(<8 x i16> %a, <8 x i16> %b) {
653 ; SSE2-LABEL: @shuffle_v8i16_XXX1X579
655 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,5,7,6,7]
656 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
657 ; SSE2-NEXT: punpcklwd {{.*}} # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
658 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
659 ; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1]
660 ; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,1,2,2,4,5,6,7]
661 ; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,5,7]
664 ; SSSE3-LABEL: @shuffle_v8i16_XXX1X579
666 ; SSSE3-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,5,7,6,7]
667 ; SSSE3-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
668 ; SSSE3-NEXT: punpcklwd {{.*}} # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
669 ; 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]
671 %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>
672 ret <8 x i16> %shuffle