| blob | c756fe7d197fdbbf02e832089e4979412f2f35b3 |
1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSE2
3 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+ssse3 | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSSE3
4 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSE41
5 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1OR2 --check-prefix=AVX1
6 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1OR2 --check-prefix=AVX2OR512VL --check-prefix=AVX2 --check-prefix=AVX2-SLOW
7 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2,+fast-variable-shuffle | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1OR2 --check-prefix=AVX2OR512VL --check-prefix=AVX2 --check-prefix=AVX2-FAST
8 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2OR512VL --check-prefix=AVX512VL --check-prefix=AVX512VL-SLOW
9 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+avx512bw,+fast-variable-shuffle | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2OR512VL --check-prefix=AVX512VL --check-prefix=AVX512VL-FAST
11 define <8 x i16> @shuffle_v8i16_01012323(<8 x i16> %a, <8 x i16> %b) {
12 ; SSE-LABEL: shuffle_v8i16_01012323:
13 ; SSE: # %bb.0:
14 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
15 ; SSE-NEXT: retq
16 ;
17 ; AVX-LABEL: shuffle_v8i16_01012323:
18 ; AVX: # %bb.0:
19 ; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,0,1,1]
20 ; AVX-NEXT: retq
21 %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>
22 ret <8 x i16> %shuffle
23 }
24 define <8 x i16> @shuffle_v8i16_67452301(<8 x i16> %a, <8 x i16> %b) {
25 ; SSE-LABEL: shuffle_v8i16_67452301:
26 ; SSE: # %bb.0:
27 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,2,1,0]
28 ; SSE-NEXT: retq
29 ;
30 ; AVX-LABEL: shuffle_v8i16_67452301:
31 ; AVX: # %bb.0:
32 ; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[3,2,1,0]
33 ; AVX-NEXT: retq
34 %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>
35 ret <8 x i16> %shuffle
36 }
37 define <8 x i16> @shuffle_v8i16_456789AB(<8 x i16> %a, <8 x i16> %b) {
38 ; SSE2-LABEL: shuffle_v8i16_456789AB:
39 ; SSE2: # %bb.0:
40 ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[2,3],xmm1[0,1]
41 ; SSE2-NEXT: retq
42 ;
43 ; SSSE3-LABEL: shuffle_v8i16_456789AB:
44 ; SSSE3: # %bb.0:
45 ; SSSE3-NEXT: palignr {{.*#+}} xmm1 = xmm0[8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7]
46 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
47 ; SSSE3-NEXT: retq
48 ;
49 ; SSE41-LABEL: shuffle_v8i16_456789AB:
50 ; SSE41: # %bb.0:
51 ; SSE41-NEXT: palignr {{.*#+}} xmm1 = xmm0[8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7]
52 ; SSE41-NEXT: movdqa %xmm1, %xmm0
53 ; SSE41-NEXT: retq
54 ;
55 ; AVX-LABEL: shuffle_v8i16_456789AB:
56 ; AVX: # %bb.0:
57 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7]
58 ; AVX-NEXT: retq
59 %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>
60 ret <8 x i16> %shuffle
61 }
63 define <8 x i16> @shuffle_v8i16_00000000(<8 x i16> %a, <8 x i16> %b) {
64 ; SSE-LABEL: shuffle_v8i16_00000000:
65 ; SSE: # %bb.0:
66 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,2,3,4,5,6,7]
67 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
68 ; SSE-NEXT: retq
69 ;
70 ; AVX1-LABEL: shuffle_v8i16_00000000:
71 ; AVX1: # %bb.0:
72 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,3,4,5,6,7]
73 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
74 ; AVX1-NEXT: retq
75 ;
76 ; AVX2OR512VL-LABEL: shuffle_v8i16_00000000:
77 ; AVX2OR512VL: # %bb.0:
78 ; AVX2OR512VL-NEXT: vpbroadcastw %xmm0, %xmm0
79 ; AVX2OR512VL-NEXT: retq
80 %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>
81 ret <8 x i16> %shuffle
82 }
83 define <8 x i16> @shuffle_v8i16_00004444(<8 x i16> %a, <8 x i16> %b) {
84 ; SSE-LABEL: shuffle_v8i16_00004444:
85 ; SSE: # %bb.0:
86 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
87 ; SSE-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
88 ; SSE-NEXT: retq
89 ;
90 ; AVX1-LABEL: shuffle_v8i16_00004444:
91 ; AVX1: # %bb.0:
92 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
93 ; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
94 ; AVX1-NEXT: retq
95 ;
96 ; AVX2-SLOW-LABEL: shuffle_v8i16_00004444:
97 ; AVX2-SLOW: # %bb.0:
98 ; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
99 ; AVX2-SLOW-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
100 ; AVX2-SLOW-NEXT: retq
101 ;
102 ; AVX2-FAST-LABEL: shuffle_v8i16_00004444:
103 ; AVX2-FAST: # %bb.0:
104 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,8,9,8,9,8,9,8,9]
105 ; AVX2-FAST-NEXT: retq
106 ;
107 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_00004444:
108 ; AVX512VL-SLOW: # %bb.0:
109 ; AVX512VL-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
110 ; AVX512VL-SLOW-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
111 ; AVX512VL-SLOW-NEXT: retq
112 ;
113 ; AVX512VL-FAST-LABEL: shuffle_v8i16_00004444:
114 ; AVX512VL-FAST: # %bb.0:
115 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,8,9,8,9,8,9,8,9]
116 ; AVX512VL-FAST-NEXT: retq
117 %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>
118 ret <8 x i16> %shuffle
119 }
120 define <8 x i16> @shuffle_v8i16_u0u1u2u3(<8 x i16> %a, <8 x i16> %b) {
121 ; SSE-LABEL: shuffle_v8i16_u0u1u2u3:
122 ; SSE: # %bb.0:
123 ; SSE-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
124 ; SSE-NEXT: retq
125 ;
126 ; AVX-LABEL: shuffle_v8i16_u0u1u2u3:
127 ; AVX: # %bb.0:
128 ; AVX-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
129 ; AVX-NEXT: retq
130 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 0, i32 undef, i32 1, i32 undef, i32 2, i32 undef, i32 3>
131 ret <8 x i16> %shuffle
132 }
133 define <8 x i16> @shuffle_v8i16_u4u5u6u7(<8 x i16> %a, <8 x i16> %b) {
134 ; SSE-LABEL: shuffle_v8i16_u4u5u6u7:
135 ; SSE: # %bb.0:
136 ; SSE-NEXT: punpckhwd {{.*#+}} xmm0 = xmm0[4,4,5,5,6,6,7,7]
137 ; SSE-NEXT: retq
138 ;
139 ; AVX-LABEL: shuffle_v8i16_u4u5u6u7:
140 ; AVX: # %bb.0:
141 ; AVX-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4,4,5,5,6,6,7,7]
142 ; AVX-NEXT: retq
143 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 4, i32 undef, i32 5, i32 undef, i32 6, i32 undef, i32 7>
144 ret <8 x i16> %shuffle
145 }
146 define <8 x i16> @shuffle_v8i16_31206745(<8 x i16> %a, <8 x i16> %b) {
147 ; SSE-LABEL: shuffle_v8i16_31206745:
148 ; SSE: # %bb.0:
149 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,0,4,5,6,7]
150 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
151 ; SSE-NEXT: retq
152 ;
153 ; AVX1-LABEL: shuffle_v8i16_31206745:
154 ; AVX1: # %bb.0:
155 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[3,1,2,0,4,5,6,7]
156 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
157 ; AVX1-NEXT: retq
158 ;
159 ; AVX2-SLOW-LABEL: shuffle_v8i16_31206745:
160 ; AVX2-SLOW: # %bb.0:
161 ; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[3,1,2,0,4,5,6,7]
162 ; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
163 ; AVX2-SLOW-NEXT: retq
164 ;
165 ; AVX2-FAST-LABEL: shuffle_v8i16_31206745:
166 ; AVX2-FAST: # %bb.0:
167 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[6,7,2,3,4,5,0,1,12,13,14,15,8,9,10,11]
168 ; AVX2-FAST-NEXT: retq
169 ;
170 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_31206745:
171 ; AVX512VL-SLOW: # %bb.0:
172 ; AVX512VL-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[3,1,2,0,4,5,6,7]
173 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
174 ; AVX512VL-SLOW-NEXT: retq
175 ;
176 ; AVX512VL-FAST-LABEL: shuffle_v8i16_31206745:
177 ; AVX512VL-FAST: # %bb.0:
178 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[6,7,2,3,4,5,0,1,12,13,14,15,8,9,10,11]
179 ; AVX512VL-FAST-NEXT: retq
180 %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>
181 ret <8 x i16> %shuffle
182 }
183 define <8 x i16> @shuffle_v8i16_44440000(<8 x i16> %a, <8 x i16> %b) {
184 ; SSE2-LABEL: shuffle_v8i16_44440000:
185 ; SSE2: # %bb.0:
186 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
187 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
188 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
189 ; SSE2-NEXT: retq
190 ;
191 ; SSSE3-LABEL: shuffle_v8i16_44440000:
192 ; SSSE3: # %bb.0:
193 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,8,9,0,1,0,1,0,1,0,1]
194 ; SSSE3-NEXT: retq
195 ;
196 ; SSE41-LABEL: shuffle_v8i16_44440000:
197 ; SSE41: # %bb.0:
198 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,8,9,0,1,0,1,0,1,0,1]
199 ; SSE41-NEXT: retq
200 ;
201 ; AVX-LABEL: shuffle_v8i16_44440000:
202 ; AVX: # %bb.0:
203 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,8,9,0,1,0,1,0,1,0,1]
204 ; AVX-NEXT: retq
205 %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>
206 ret <8 x i16> %shuffle
207 }
208 define <8 x i16> @shuffle_v8i16_23016745(<8 x i16> %a, <8 x i16> %b) {
209 ; SSE-LABEL: shuffle_v8i16_23016745:
210 ; SSE: # %bb.0:
211 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,0,3,2]
212 ; SSE-NEXT: retq
213 ;
214 ; AVX-LABEL: shuffle_v8i16_23016745:
215 ; AVX: # %bb.0:
216 ; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[1,0,3,2]
217 ; AVX-NEXT: retq
218 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5>
219 ret <8 x i16> %shuffle
220 }
221 define <8 x i16> @shuffle_v8i16_23026745(<8 x i16> %a, <8 x i16> %b) {
222 ; SSE-LABEL: shuffle_v8i16_23026745:
223 ; SSE: # %bb.0:
224 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[2,3,0,2,4,5,6,7]
225 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
226 ; SSE-NEXT: retq
227 ;
228 ; AVX1-LABEL: shuffle_v8i16_23026745:
229 ; AVX1: # %bb.0:
230 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[2,3,0,2,4,5,6,7]
231 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
232 ; AVX1-NEXT: retq
233 ;
234 ; AVX2-SLOW-LABEL: shuffle_v8i16_23026745:
235 ; AVX2-SLOW: # %bb.0:
236 ; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[2,3,0,2,4,5,6,7]
237 ; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
238 ; AVX2-SLOW-NEXT: retq
239 ;
240 ; AVX2-FAST-LABEL: shuffle_v8i16_23026745:
241 ; AVX2-FAST: # %bb.0:
242 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5,6,7,0,1,4,5,12,13,14,15,8,9,10,11]
243 ; AVX2-FAST-NEXT: retq
244 ;
245 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_23026745:
246 ; AVX512VL-SLOW: # %bb.0:
247 ; AVX512VL-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[2,3,0,2,4,5,6,7]
248 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
249 ; AVX512VL-SLOW-NEXT: retq
250 ;
251 ; AVX512VL-FAST-LABEL: shuffle_v8i16_23026745:
252 ; AVX512VL-FAST: # %bb.0:
253 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5,6,7,0,1,4,5,12,13,14,15,8,9,10,11]
254 ; AVX512VL-FAST-NEXT: retq
255 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 2, i32 3, i32 0, i32 2, i32 6, i32 7, i32 4, i32 5>
256 ret <8 x i16> %shuffle
257 }
258 define <8 x i16> @shuffle_v8i16_23016747(<8 x i16> %a, <8 x i16> %b) {
259 ; SSE-LABEL: shuffle_v8i16_23016747:
260 ; SSE: # %bb.0:
261 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,0,2,3]
262 ; SSE-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,7,4,7]
263 ; SSE-NEXT: retq
264 ;
265 ; AVX1-LABEL: shuffle_v8i16_23016747:
266 ; AVX1: # %bb.0:
267 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,2,3]
268 ; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,7,4,7]
269 ; AVX1-NEXT: retq
270 ;
271 ; AVX2-SLOW-LABEL: shuffle_v8i16_23016747:
272 ; AVX2-SLOW: # %bb.0:
273 ; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,2,3]
274 ; AVX2-SLOW-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,7,4,7]
275 ; AVX2-SLOW-NEXT: retq
276 ;
277 ; AVX2-FAST-LABEL: shuffle_v8i16_23016747:
278 ; AVX2-FAST: # %bb.0:
279 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5,6,7,0,1,2,3,12,13,14,15,8,9,14,15]
280 ; AVX2-FAST-NEXT: retq
281 ;
282 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_23016747:
283 ; AVX512VL-SLOW: # %bb.0:
284 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,0,2,3]
285 ; AVX512VL-SLOW-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,7,4,7]
286 ; AVX512VL-SLOW-NEXT: retq
287 ;
288 ; AVX512VL-FAST-LABEL: shuffle_v8i16_23016747:
289 ; AVX512VL-FAST: # %bb.0:
290 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5,6,7,0,1,2,3,12,13,14,15,8,9,14,15]
291 ; AVX512VL-FAST-NEXT: retq
292 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 7>
293 ret <8 x i16> %shuffle
294 }
295 define <8 x i16> @shuffle_v8i16_75643120(<8 x i16> %a, <8 x i16> %b) {
296 ; SSE2-LABEL: shuffle_v8i16_75643120:
297 ; SSE2: # %bb.0:
298 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
299 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,0,4,5,6,7]
300 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,6,4]
301 ; SSE2-NEXT: retq
302 ;
303 ; SSSE3-LABEL: shuffle_v8i16_75643120:
304 ; SSSE3: # %bb.0:
305 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[14,15,10,11,12,13,8,9,6,7,2,3,4,5,0,1]
306 ; SSSE3-NEXT: retq
307 ;
308 ; SSE41-LABEL: shuffle_v8i16_75643120:
309 ; SSE41: # %bb.0:
310 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[14,15,10,11,12,13,8,9,6,7,2,3,4,5,0,1]
311 ; SSE41-NEXT: retq
312 ;
313 ; AVX-LABEL: shuffle_v8i16_75643120:
314 ; AVX: # %bb.0:
315 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[14,15,10,11,12,13,8,9,6,7,2,3,4,5,0,1]
316 ; AVX-NEXT: retq
317 %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>
318 ret <8 x i16> %shuffle
319 }
321 define <8 x i16> @shuffle_v8i16_10545410(<8 x i16> %a, <8 x i16> %b) {
322 ; SSE2-LABEL: shuffle_v8i16_10545410:
323 ; SSE2: # %bb.0:
324 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,0]
325 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,0,3,2,4,5,6,7]
326 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,4,7,6]
327 ; SSE2-NEXT: retq
328 ;
329 ; SSSE3-LABEL: shuffle_v8i16_10545410:
330 ; SSSE3: # %bb.0:
331 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,0,1,10,11,8,9,10,11,8,9,2,3,0,1]
332 ; SSSE3-NEXT: retq
333 ;
334 ; SSE41-LABEL: shuffle_v8i16_10545410:
335 ; SSE41: # %bb.0:
336 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,0,1,10,11,8,9,10,11,8,9,2,3,0,1]
337 ; SSE41-NEXT: retq
338 ;
339 ; AVX-LABEL: shuffle_v8i16_10545410:
340 ; AVX: # %bb.0:
341 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,0,1,10,11,8,9,10,11,8,9,2,3,0,1]
342 ; AVX-NEXT: retq
343 %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>
344 ret <8 x i16> %shuffle
345 }
346 define <8 x i16> @shuffle_v8i16_54105410(<8 x i16> %a, <8 x i16> %b) {
347 ; SSE2-LABEL: shuffle_v8i16_54105410:
348 ; SSE2: # %bb.0:
349 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,0]
350 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,2,1,0,4,5,6,7]
351 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,4,7,6]
352 ; SSE2-NEXT: retq
353 ;
354 ; SSSE3-LABEL: shuffle_v8i16_54105410:
355 ; SSSE3: # %bb.0:
356 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,10,11,8,9,2,3,0,1]
357 ; SSSE3-NEXT: retq
358 ;
359 ; SSE41-LABEL: shuffle_v8i16_54105410:
360 ; SSE41: # %bb.0:
361 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,10,11,8,9,2,3,0,1]
362 ; SSE41-NEXT: retq
363 ;
364 ; AVX-LABEL: shuffle_v8i16_54105410:
365 ; AVX: # %bb.0:
366 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,10,11,8,9,2,3,0,1]
367 ; AVX-NEXT: retq
368 %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>
369 ret <8 x i16> %shuffle
370 }
371 define <8 x i16> @shuffle_v8i16_54101054(<8 x i16> %a, <8 x i16> %b) {
372 ; SSE2-LABEL: shuffle_v8i16_54101054:
373 ; SSE2: # %bb.0:
374 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,0]
375 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,2,1,0,4,5,6,7]
376 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,6,5,4]
377 ; SSE2-NEXT: retq
378 ;
379 ; SSSE3-LABEL: shuffle_v8i16_54101054:
380 ; SSSE3: # %bb.0:
381 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,2,3,0,1,10,11,8,9]
382 ; SSSE3-NEXT: retq
383 ;
384 ; SSE41-LABEL: shuffle_v8i16_54101054:
385 ; SSE41: # %bb.0:
386 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,2,3,0,1,10,11,8,9]
387 ; SSE41-NEXT: retq
388 ;
389 ; AVX-LABEL: shuffle_v8i16_54101054:
390 ; AVX: # %bb.0:
391 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,2,3,0,1,10,11,8,9]
392 ; AVX-NEXT: retq
393 %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>
394 ret <8 x i16> %shuffle
395 }
396 define <8 x i16> @shuffle_v8i16_04400440(<8 x i16> %a, <8 x i16> %b) {
397 ; SSE2-LABEL: shuffle_v8i16_04400440:
398 ; SSE2: # %bb.0:
399 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,0]
400 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7]
401 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,4,4,6]
402 ; SSE2-NEXT: retq
403 ;
404 ; SSSE3-LABEL: shuffle_v8i16_04400440:
405 ; SSSE3: # %bb.0:
406 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,8,9,8,9,0,1]
407 ; SSSE3-NEXT: retq
408 ;
409 ; SSE41-LABEL: shuffle_v8i16_04400440:
410 ; SSE41: # %bb.0:
411 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,8,9,8,9,0,1]
412 ; SSE41-NEXT: retq
413 ;
414 ; AVX-LABEL: shuffle_v8i16_04400440:
415 ; AVX: # %bb.0:
416 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,8,9,8,9,0,1]
417 ; AVX-NEXT: retq
418 %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>
419 ret <8 x i16> %shuffle
420 }
421 define <8 x i16> @shuffle_v8i16_40044004(<8 x i16> %a, <8 x i16> %b) {
422 ; SSE2-LABEL: shuffle_v8i16_40044004:
423 ; SSE2: # %bb.0:
424 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,0]
425 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[2,0,0,2,4,5,6,7]
426 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,4]
427 ; SSE2-NEXT: retq
428 ;
429 ; SSSE3-LABEL: shuffle_v8i16_40044004:
430 ; SSSE3: # %bb.0:
431 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,0,1,0,1,8,9,8,9,0,1,0,1,8,9]
432 ; SSSE3-NEXT: retq
433 ;
434 ; SSE41-LABEL: shuffle_v8i16_40044004:
435 ; SSE41: # %bb.0:
436 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,0,1,0,1,8,9,8,9,0,1,0,1,8,9]
437 ; SSE41-NEXT: retq
438 ;
439 ; AVX-LABEL: shuffle_v8i16_40044004:
440 ; AVX: # %bb.0:
441 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,0,1,0,1,8,9,8,9,0,1,0,1,8,9]
442 ; AVX-NEXT: retq
443 %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>
444 ret <8 x i16> %shuffle
445 }
447 define <8 x i16> @shuffle_v8i16_26405173(<8 x i16> %a, <8 x i16> %b) {
448 ; SSE2-LABEL: shuffle_v8i16_26405173:
449 ; SSE2: # %bb.0:
450 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
451 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,6,4]
452 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,3,2,1]
453 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,2,3,0,4,5,6,7]
454 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,6,4,7]
455 ; SSE2-NEXT: retq
456 ;
457 ; SSSE3-LABEL: shuffle_v8i16_26405173:
458 ; SSSE3: # %bb.0:
459 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[4,5,12,13,8,9,0,1,10,11,2,3,14,15,6,7]
460 ; SSSE3-NEXT: retq
461 ;
462 ; SSE41-LABEL: shuffle_v8i16_26405173:
463 ; SSE41: # %bb.0:
464 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[4,5,12,13,8,9,0,1,10,11,2,3,14,15,6,7]
465 ; SSE41-NEXT: retq
466 ;
467 ; AVX-LABEL: shuffle_v8i16_26405173:
468 ; AVX: # %bb.0:
469 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5,12,13,8,9,0,1,10,11,2,3,14,15,6,7]
470 ; AVX-NEXT: retq
471 %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>
472 ret <8 x i16> %shuffle
473 }
474 define <8 x i16> @shuffle_v8i16_20645173(<8 x i16> %a, <8 x i16> %b) {
475 ; SSE2-LABEL: shuffle_v8i16_20645173:
476 ; SSE2: # %bb.0:
477 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
478 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,6,4]
479 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,3,2,1]
480 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,0,2,3,4,5,6,7]
481 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,6,4,7]
482 ; SSE2-NEXT: retq
483 ;
484 ; SSSE3-LABEL: shuffle_v8i16_20645173:
485 ; SSSE3: # %bb.0:
486 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[4,5,0,1,12,13,8,9,10,11,2,3,14,15,6,7]
487 ; SSSE3-NEXT: retq
488 ;
489 ; SSE41-LABEL: shuffle_v8i16_20645173:
490 ; SSE41: # %bb.0:
491 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[4,5,0,1,12,13,8,9,10,11,2,3,14,15,6,7]
492 ; SSE41-NEXT: retq
493 ;
494 ; AVX-LABEL: shuffle_v8i16_20645173:
495 ; AVX: # %bb.0:
496 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5,0,1,12,13,8,9,10,11,2,3,14,15,6,7]
497 ; AVX-NEXT: retq
498 %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>
499 ret <8 x i16> %shuffle
500 }
501 define <8 x i16> @shuffle_v8i16_26401375(<8 x i16> %a, <8 x i16> %b) {
502 ; SSE2-LABEL: shuffle_v8i16_26401375:
503 ; SSE2: # %bb.0:
504 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
505 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,6,4]
506 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,3,1,2]
507 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,2,3,0,4,5,6,7]
508 ; SSE2-NEXT: retq
509 ;
510 ; SSSE3-LABEL: shuffle_v8i16_26401375:
511 ; SSSE3: # %bb.0:
512 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[4,5,12,13,8,9,0,1,2,3,6,7,14,15,10,11]
513 ; SSSE3-NEXT: retq
514 ;
515 ; SSE41-LABEL: shuffle_v8i16_26401375:
516 ; SSE41: # %bb.0:
517 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[4,5,12,13,8,9,0,1,2,3,6,7,14,15,10,11]
518 ; SSE41-NEXT: retq
519 ;
520 ; AVX-LABEL: shuffle_v8i16_26401375:
521 ; AVX: # %bb.0:
522 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5,12,13,8,9,0,1,2,3,6,7,14,15,10,11]
523 ; AVX-NEXT: retq
524 %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>
525 ret <8 x i16> %shuffle
526 }
528 define <8 x i16> @shuffle_v8i16_66751643(<8 x i16> %a, <8 x i16> %b) {
529 ; SSE2-LABEL: shuffle_v8i16_66751643:
530 ; SSE2: # %bb.0:
531 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
532 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,5,7]
533 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,2,0]
534 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,1,3,2,4,5,6,7]
535 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,4,6]
536 ; SSE2-NEXT: retq
537 ;
538 ; SSSE3-LABEL: shuffle_v8i16_66751643:
539 ; SSSE3: # %bb.0:
540 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[12,13,12,13,14,15,10,11,2,3,12,13,8,9,6,7]
541 ; SSSE3-NEXT: retq
542 ;
543 ; SSE41-LABEL: shuffle_v8i16_66751643:
544 ; SSE41: # %bb.0:
545 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[12,13,12,13,14,15,10,11,2,3,12,13,8,9,6,7]
546 ; SSE41-NEXT: retq
547 ;
548 ; AVX-LABEL: shuffle_v8i16_66751643:
549 ; AVX: # %bb.0:
550 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[12,13,12,13,14,15,10,11,2,3,12,13,8,9,6,7]
551 ; AVX-NEXT: retq
552 %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>
553 ret <8 x i16> %shuffle
554 }
556 define <8 x i16> @shuffle_v8i16_60514754(<8 x i16> %a, <8 x i16> %b) {
557 ; SSE2-LABEL: shuffle_v8i16_60514754:
558 ; SSE2: # %bb.0:
559 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,4,7]
560 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
561 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7]
562 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,7,5,6]
563 ; SSE2-NEXT: retq
564 ;
565 ; SSSE3-LABEL: shuffle_v8i16_60514754:
566 ; SSSE3: # %bb.0:
567 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[12,13,0,1,10,11,2,3,8,9,14,15,10,11,8,9]
568 ; SSSE3-NEXT: retq
569 ;
570 ; SSE41-LABEL: shuffle_v8i16_60514754:
571 ; SSE41: # %bb.0:
572 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[12,13,0,1,10,11,2,3,8,9,14,15,10,11,8,9]
573 ; SSE41-NEXT: retq
574 ;
575 ; AVX-LABEL: shuffle_v8i16_60514754:
576 ; AVX: # %bb.0:
577 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[12,13,0,1,10,11,2,3,8,9,14,15,10,11,8,9]
578 ; AVX-NEXT: retq
579 %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>
580 ret <8 x i16> %shuffle
581 }
583 define <8 x i16> @shuffle_v8i16_00444444(<8 x i16> %a, <8 x i16> %b) {
584 ; SSE2-LABEL: shuffle_v8i16_00444444:
585 ; SSE2: # %bb.0:
586 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
587 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7]
588 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
589 ; SSE2-NEXT: retq
590 ;
591 ; SSSE3-LABEL: shuffle_v8i16_00444444:
592 ; SSSE3: # %bb.0:
593 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9]
594 ; SSSE3-NEXT: retq
595 ;
596 ; SSE41-LABEL: shuffle_v8i16_00444444:
597 ; SSE41: # %bb.0:
598 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9]
599 ; SSE41-NEXT: retq
600 ;
601 ; AVX-LABEL: shuffle_v8i16_00444444:
602 ; AVX: # %bb.0:
603 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9]
604 ; AVX-NEXT: retq
605 %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>
606 ret <8 x i16> %shuffle
607 }
608 define <8 x i16> @shuffle_v8i16_44004444(<8 x i16> %a, <8 x i16> %b) {
609 ; SSE2-LABEL: shuffle_v8i16_44004444:
610 ; SSE2: # %bb.0:
611 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
612 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[2,2,0,0,4,5,6,7]
613 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
614 ; SSE2-NEXT: retq
615 ;
616 ; SSSE3-LABEL: shuffle_v8i16_44004444:
617 ; SSSE3: # %bb.0:
618 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9]
619 ; SSSE3-NEXT: retq
620 ;
621 ; SSE41-LABEL: shuffle_v8i16_44004444:
622 ; SSE41: # %bb.0:
623 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9]
624 ; SSE41-NEXT: retq
625 ;
626 ; AVX-LABEL: shuffle_v8i16_44004444:
627 ; AVX: # %bb.0:
628 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9]
629 ; AVX-NEXT: retq
630 %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>
631 ret <8 x i16> %shuffle
632 }
633 define <8 x i16> @shuffle_v8i16_04404444(<8 x i16> %a, <8 x i16> %b) {
634 ; SSE2-LABEL: shuffle_v8i16_04404444:
635 ; SSE2: # %bb.0:
636 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
637 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7]
638 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
639 ; SSE2-NEXT: retq
640 ;
641 ; SSSE3-LABEL: shuffle_v8i16_04404444:
642 ; SSSE3: # %bb.0:
643 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
644 ; SSSE3-NEXT: retq
645 ;
646 ; SSE41-LABEL: shuffle_v8i16_04404444:
647 ; SSE41: # %bb.0:
648 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
649 ; SSE41-NEXT: retq
650 ;
651 ; AVX-LABEL: shuffle_v8i16_04404444:
652 ; AVX: # %bb.0:
653 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
654 ; AVX-NEXT: retq
655 %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>
656 ret <8 x i16> %shuffle
657 }
658 define <8 x i16> @shuffle_v8i16_04400000(<8 x i16> %a, <8 x i16> %b) {
659 ; SSE2-LABEL: shuffle_v8i16_04400000:
660 ; SSE2: # %bb.0:
661 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,0,3]
662 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7]
663 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
664 ; SSE2-NEXT: retq
665 ;
666 ; SSSE3-LABEL: shuffle_v8i16_04400000:
667 ; SSSE3: # %bb.0:
668 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1]
669 ; SSSE3-NEXT: retq
670 ;
671 ; SSE41-LABEL: shuffle_v8i16_04400000:
672 ; SSE41: # %bb.0:
673 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1]
674 ; SSE41-NEXT: retq
675 ;
676 ; AVX-LABEL: shuffle_v8i16_04400000:
677 ; AVX: # %bb.0:
678 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1]
679 ; AVX-NEXT: retq
680 %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>
681 ret <8 x i16> %shuffle
682 }
683 define <8 x i16> @shuffle_v8i16_04404567(<8 x i16> %a, <8 x i16> %b) {
684 ; SSE-LABEL: shuffle_v8i16_04404567:
685 ; SSE: # %bb.0:
686 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
687 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7]
688 ; SSE-NEXT: retq
689 ;
690 ; AVX1-LABEL: shuffle_v8i16_04404567:
691 ; AVX1: # %bb.0:
692 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
693 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7]
694 ; AVX1-NEXT: retq
695 ;
696 ; AVX2-SLOW-LABEL: shuffle_v8i16_04404567:
697 ; AVX2-SLOW: # %bb.0:
698 ; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
699 ; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7]
700 ; AVX2-SLOW-NEXT: retq
701 ;
702 ; AVX2-FAST-LABEL: shuffle_v8i16_04404567:
703 ; AVX2-FAST: # %bb.0:
704 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,10,11,12,13,14,15]
705 ; AVX2-FAST-NEXT: retq
706 ;
707 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_04404567:
708 ; AVX512VL-SLOW: # %bb.0:
709 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
710 ; AVX512VL-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7]
711 ; AVX512VL-SLOW-NEXT: retq
712 ;
713 ; AVX512VL-FAST-LABEL: shuffle_v8i16_04404567:
714 ; AVX512VL-FAST: # %bb.0:
715 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,10,11,12,13,14,15]
716 ; AVX512VL-FAST-NEXT: retq
717 %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>
718 ret <8 x i16> %shuffle
719 }
721 define <8 x i16> @shuffle_v8i16_0X444444(<8 x i16> %a, <8 x i16> %b) {
722 ; SSE2-LABEL: shuffle_v8i16_0X444444:
723 ; SSE2: # %bb.0:
724 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
725 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,1,2,2,4,5,6,7]
726 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
727 ; SSE2-NEXT: retq
728 ;
729 ; SSSE3-LABEL: shuffle_v8i16_0X444444:
730 ; SSSE3: # %bb.0:
731 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,8,9,8,9,8,9,8,9,8,9,8,9]
732 ; SSSE3-NEXT: retq
733 ;
734 ; SSE41-LABEL: shuffle_v8i16_0X444444:
735 ; SSE41: # %bb.0:
736 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,8,9,8,9,8,9,8,9,8,9,8,9]
737 ; SSE41-NEXT: retq
738 ;
739 ; AVX-LABEL: shuffle_v8i16_0X444444:
740 ; AVX: # %bb.0:
741 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,8,9,8,9,8,9,8,9,8,9,8,9]
742 ; AVX-NEXT: retq
743 %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>
744 ret <8 x i16> %shuffle
745 }
746 define <8 x i16> @shuffle_v8i16_44X04444(<8 x i16> %a, <8 x i16> %b) {
747 ; SSE2-LABEL: shuffle_v8i16_44X04444:
748 ; SSE2: # %bb.0:
749 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
750 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[2,2,2,0,4,5,6,7]
751 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
752 ; SSE2-NEXT: retq
753 ;
754 ; SSSE3-LABEL: shuffle_v8i16_44X04444:
755 ; SSSE3: # %bb.0:
756 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
757 ; SSSE3-NEXT: retq
758 ;
759 ; SSE41-LABEL: shuffle_v8i16_44X04444:
760 ; SSE41: # %bb.0:
761 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
762 ; SSE41-NEXT: retq
763 ;
764 ; AVX-LABEL: shuffle_v8i16_44X04444:
765 ; AVX: # %bb.0:
766 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
767 ; AVX-NEXT: retq
768 %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>
769 ret <8 x i16> %shuffle
770 }
771 define <8 x i16> @shuffle_v8i16_X4404444(<8 x i16> %a, <8 x i16> %b) {
772 ; SSE2-LABEL: shuffle_v8i16_X4404444:
773 ; SSE2: # %bb.0:
774 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
775 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,0,4,5,6,7]
776 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
777 ; SSE2-NEXT: retq
778 ;
779 ; SSSE3-LABEL: shuffle_v8i16_X4404444:
780 ; SSSE3: # %bb.0:
781 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
782 ; SSSE3-NEXT: retq
783 ;
784 ; SSE41-LABEL: shuffle_v8i16_X4404444:
785 ; SSE41: # %bb.0:
786 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
787 ; SSE41-NEXT: retq
788 ;
789 ; AVX-LABEL: shuffle_v8i16_X4404444:
790 ; AVX: # %bb.0:
791 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
792 ; AVX-NEXT: retq
793 %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>
794 ret <8 x i16> %shuffle
795 }
797 define <8 x i16> @shuffle_v8i16_0127XXXX(<8 x i16> %a, <8 x i16> %b) {
798 ; SSE2-LABEL: shuffle_v8i16_0127XXXX:
799 ; SSE2: # %bb.0:
800 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
801 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,7]
802 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
803 ; SSE2-NEXT: retq
804 ;
805 ; SSSE3-LABEL: shuffle_v8i16_0127XXXX:
806 ; SSSE3: # %bb.0:
807 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,14,15,4,5,14,15,12,13,14,15]
808 ; SSSE3-NEXT: retq
809 ;
810 ; SSE41-LABEL: shuffle_v8i16_0127XXXX:
811 ; SSE41: # %bb.0:
812 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,14,15,4,5,14,15,12,13,14,15]
813 ; SSE41-NEXT: retq
814 ;
815 ; AVX-LABEL: shuffle_v8i16_0127XXXX:
816 ; AVX: # %bb.0:
817 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,14,15,4,5,14,15,12,13,14,15]
818 ; AVX-NEXT: retq
819 %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>
820 ret <8 x i16> %shuffle
821 }
823 define <8 x i16> @shuffle_v8i16_XXXX4563(<8 x i16> %a, <8 x i16> %b) {
824 ; SSE2-LABEL: shuffle_v8i16_XXXX4563:
825 ; SSE2: # %bb.0:
826 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
827 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
828 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,2,0]
829 ; SSE2-NEXT: retq
830 ;
831 ; SSSE3-LABEL: shuffle_v8i16_XXXX4563:
832 ; SSSE3: # %bb.0:
833 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[12,13,6,7,4,5,6,7,8,9,10,11,12,13,6,7]
834 ; SSSE3-NEXT: retq
835 ;
836 ; SSE41-LABEL: shuffle_v8i16_XXXX4563:
837 ; SSE41: # %bb.0:
838 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[12,13,6,7,4,5,6,7,8,9,10,11,12,13,6,7]
839 ; SSE41-NEXT: retq
840 ;
841 ; AVX-LABEL: shuffle_v8i16_XXXX4563:
842 ; AVX: # %bb.0:
843 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[12,13,6,7,4,5,6,7,8,9,10,11,12,13,6,7]
844 ; AVX-NEXT: retq
845 %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>
846 ret <8 x i16> %shuffle
847 }
849 define <8 x i16> @shuffle_v8i16_4563XXXX(<8 x i16> %a, <8 x i16> %b) {
850 ; SSE2-LABEL: shuffle_v8i16_4563XXXX:
851 ; SSE2: # %bb.0:
852 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
853 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
854 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,0,2,3]
855 ; SSE2-NEXT: retq
856 ;
857 ; SSSE3-LABEL: shuffle_v8i16_4563XXXX:
858 ; SSSE3: # %bb.0:
859 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,8,9,10,11,0,1,2,3]
860 ; SSSE3-NEXT: retq
861 ;
862 ; SSE41-LABEL: shuffle_v8i16_4563XXXX:
863 ; SSE41: # %bb.0:
864 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,8,9,10,11,0,1,2,3]
865 ; SSE41-NEXT: retq
866 ;
867 ; AVX-LABEL: shuffle_v8i16_4563XXXX:
868 ; AVX: # %bb.0:
869 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,8,9,10,11,0,1,2,3]
870 ; AVX-NEXT: retq
871 %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>
872 ret <8 x i16> %shuffle
873 }
875 define <8 x i16> @shuffle_v8i16_01274563(<8 x i16> %a, <8 x i16> %b) {
876 ; SSE2-LABEL: shuffle_v8i16_01274563:
877 ; SSE2: # %bb.0:
878 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
879 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,4,7]
880 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,3,1,2]
881 ; SSE2-NEXT: retq
882 ;
883 ; SSSE3-LABEL: shuffle_v8i16_01274563:
884 ; SSSE3: # %bb.0:
885 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,14,15,8,9,10,11,12,13,6,7]
886 ; SSSE3-NEXT: retq
887 ;
888 ; SSE41-LABEL: shuffle_v8i16_01274563:
889 ; SSE41: # %bb.0:
890 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,14,15,8,9,10,11,12,13,6,7]
891 ; SSE41-NEXT: retq
892 ;
893 ; AVX-LABEL: shuffle_v8i16_01274563:
894 ; AVX: # %bb.0:
895 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,14,15,8,9,10,11,12,13,6,7]
896 ; AVX-NEXT: retq
897 %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>
898 ret <8 x i16> %shuffle
899 }
901 define <8 x i16> @shuffle_v8i16_45630127(<8 x i16> %a, <8 x i16> %b) {
902 ; SSE2-LABEL: shuffle_v8i16_45630127:
903 ; SSE2: # %bb.0:
904 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
905 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,1,4,5,6,7]
906 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,0,3,1]
907 ; SSE2-NEXT: retq
908 ;
909 ; SSSE3-LABEL: shuffle_v8i16_45630127:
910 ; SSSE3: # %bb.0:
911 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15]
912 ; SSSE3-NEXT: retq
913 ;
914 ; SSE41-LABEL: shuffle_v8i16_45630127:
915 ; SSE41: # %bb.0:
916 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15]
917 ; SSE41-NEXT: retq
918 ;
919 ; AVX-LABEL: shuffle_v8i16_45630127:
920 ; AVX: # %bb.0:
921 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15]
922 ; AVX-NEXT: retq
923 %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>
924 ret <8 x i16> %shuffle
925 }
927 define <8 x i16> @shuffle_v8i16_37102735(<8 x i16> %a, <8 x i16> %b) {
928 ; SSE2-LABEL: shuffle_v8i16_37102735:
929 ; SSE2: # %bb.0:
930 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,5,7]
931 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
932 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,6,4]
933 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
934 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,2,1,0,4,5,6,7]
935 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,4,5,6]
936 ; SSE2-NEXT: retq
937 ;
938 ; SSSE3-LABEL: shuffle_v8i16_37102735:
939 ; SSSE3: # %bb.0:
940 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[6,7,14,15,2,3,0,1,4,5,14,15,6,7,10,11]
941 ; SSSE3-NEXT: retq
942 ;
943 ; SSE41-LABEL: shuffle_v8i16_37102735:
944 ; SSE41: # %bb.0:
945 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[6,7,14,15,2,3,0,1,4,5,14,15,6,7,10,11]
946 ; SSE41-NEXT: retq
947 ;
948 ; AVX-LABEL: shuffle_v8i16_37102735:
949 ; AVX: # %bb.0:
950 ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[6,7,14,15,2,3,0,1,4,5,14,15,6,7,10,11]
951 ; AVX-NEXT: retq
952 %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>
953 ret <8 x i16> %shuffle
954 }
956 define <8 x i16> @shuffle_v8i16_08192a3b(<8 x i16> %a, <8 x i16> %b) {
957 ; SSE-LABEL: shuffle_v8i16_08192a3b:
958 ; SSE: # %bb.0:
959 ; SSE-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
960 ; SSE-NEXT: retq
961 ;
962 ; AVX-LABEL: shuffle_v8i16_08192a3b:
963 ; AVX: # %bb.0:
964 ; AVX-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
965 ; AVX-NEXT: retq
966 %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>
967 ret <8 x i16> %shuffle
968 }
970 define <8 x i16> @shuffle_v8i16_0c1d2e3f(<8 x i16> %a, <8 x i16> %b) {
971 ; SSE-LABEL: shuffle_v8i16_0c1d2e3f:
972 ; SSE: # %bb.0:
973 ; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
974 ; SSE-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
975 ; SSE-NEXT: retq
976 ;
977 ; AVX1OR2-LABEL: shuffle_v8i16_0c1d2e3f:
978 ; AVX1OR2: # %bb.0:
979 ; AVX1OR2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
980 ; AVX1OR2-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
981 ; AVX1OR2-NEXT: retq
982 ;
983 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_0c1d2e3f:
984 ; AVX512VL-SLOW: # %bb.0:
985 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
986 ; AVX512VL-SLOW-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
987 ; AVX512VL-SLOW-NEXT: retq
988 ;
989 ; AVX512VL-FAST-LABEL: shuffle_v8i16_0c1d2e3f:
990 ; AVX512VL-FAST: # %bb.0:
991 ; AVX512VL-FAST-NEXT: vmovdqa {{.*#+}} xmm2 = [0,12,1,13,2,14,3,15]
992 ; AVX512VL-FAST-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
993 ; AVX512VL-FAST-NEXT: retq
994 %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>
995 ret <8 x i16> %shuffle
996 }
998 define <8 x i16> @shuffle_v8i16_4c5d6e7f(<8 x i16> %a, <8 x i16> %b) {
999 ; SSE-LABEL: shuffle_v8i16_4c5d6e7f:
1000 ; SSE: # %bb.0:
1001 ; SSE-NEXT: punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
1002 ; SSE-NEXT: retq
1003 ;
1004 ; AVX-LABEL: shuffle_v8i16_4c5d6e7f:
1005 ; AVX: # %bb.0:
1006 ; AVX-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
1007 ; AVX-NEXT: retq
1008 %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>
1009 ret <8 x i16> %shuffle
1010 }
1012 define <8 x i16> @shuffle_v8i16_48596a7b(<8 x i16> %a, <8 x i16> %b) {
1013 ; SSE-LABEL: shuffle_v8i16_48596a7b:
1014 ; SSE: # %bb.0:
1015 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
1016 ; SSE-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1017 ; SSE-NEXT: retq
1018 ;
1019 ; AVX1OR2-LABEL: shuffle_v8i16_48596a7b:
1020 ; AVX1OR2: # %bb.0:
1021 ; AVX1OR2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
1022 ; AVX1OR2-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1023 ; AVX1OR2-NEXT: retq
1024 ;
1025 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_48596a7b:
1026 ; AVX512VL-SLOW: # %bb.0:
1027 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
1028 ; AVX512VL-SLOW-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1029 ; AVX512VL-SLOW-NEXT: retq
1030 ;
1031 ; AVX512VL-FAST-LABEL: shuffle_v8i16_48596a7b:
1032 ; AVX512VL-FAST: # %bb.0:
1033 ; AVX512VL-FAST-NEXT: vmovdqa {{.*#+}} xmm2 = [4,8,5,9,6,10,7,11]
1034 ; AVX512VL-FAST-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1035 ; AVX512VL-FAST-NEXT: retq
1036 %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>
1037 ret <8 x i16> %shuffle
1038 }
1040 define <8 x i16> @shuffle_v8i16_08196e7f(<8 x i16> %a, <8 x i16> %b) {
1041 ; SSE-LABEL: shuffle_v8i16_08196e7f:
1042 ; SSE: # %bb.0:
1043 ; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
1044 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
1045 ; SSE-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1046 ; SSE-NEXT: retq
1047 ;
1048 ; AVX1OR2-LABEL: shuffle_v8i16_08196e7f:
1049 ; AVX1OR2: # %bb.0:
1050 ; AVX1OR2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
1051 ; AVX1OR2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
1052 ; AVX1OR2-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1053 ; AVX1OR2-NEXT: retq
1054 ;
1055 ; AVX512VL-LABEL: shuffle_v8i16_08196e7f:
1056 ; AVX512VL: # %bb.0:
1057 ; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [0,8,1,9,6,14,7,15]
1058 ; AVX512VL-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1059 ; AVX512VL-NEXT: retq
1060 %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>
1061 ret <8 x i16> %shuffle
1062 }
1064 define <8 x i16> @shuffle_v8i16_0c1d6879(<8 x i16> %a, <8 x i16> %b) {
1065 ; SSE-LABEL: shuffle_v8i16_0c1d6879:
1066 ; SSE: # %bb.0:
1067 ; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,0,2,3]
1068 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
1069 ; SSE-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1070 ; SSE-NEXT: retq
1071 ;
1072 ; AVX1OR2-LABEL: shuffle_v8i16_0c1d6879:
1073 ; AVX1OR2: # %bb.0:
1074 ; AVX1OR2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,0,2,3]
1075 ; AVX1OR2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
1076 ; AVX1OR2-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1077 ; AVX1OR2-NEXT: retq
1078 ;
1079 ; AVX512VL-LABEL: shuffle_v8i16_0c1d6879:
1080 ; AVX512VL: # %bb.0:
1081 ; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [0,12,1,13,6,8,7,9]
1082 ; AVX512VL-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1083 ; AVX512VL-NEXT: retq
1084 %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>
1085 ret <8 x i16> %shuffle
1086 }
1088 define <8 x i16> @shuffle_v8i16_109832ba(<8 x i16> %a, <8 x i16> %b) {
1089 ; SSE-LABEL: shuffle_v8i16_109832ba:
1090 ; SSE: # %bb.0:
1091 ; SSE-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1092 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7]
1093 ; SSE-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,4,7,5]
1094 ; SSE-NEXT: retq
1095 ;
1096 ; AVX1-LABEL: shuffle_v8i16_109832ba:
1097 ; AVX1: # %bb.0:
1098 ; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1099 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7]
1100 ; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,4,7,5]
1101 ; AVX1-NEXT: retq
1102 ;
1103 ; AVX2-SLOW-LABEL: shuffle_v8i16_109832ba:
1104 ; AVX2-SLOW: # %bb.0:
1105 ; AVX2-SLOW-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1106 ; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7]
1107 ; AVX2-SLOW-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,4,7,5]
1108 ; AVX2-SLOW-NEXT: retq
1109 ;
1110 ; AVX2-FAST-LABEL: shuffle_v8i16_109832ba:
1111 ; AVX2-FAST: # %bb.0:
1112 ; AVX2-FAST-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
1113 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5,0,1,6,7,2,3,12,13,8,9,14,15,10,11]
1114 ; AVX2-FAST-NEXT: retq
1115 ;
1116 ; AVX512VL-LABEL: shuffle_v8i16_109832ba:
1117 ; AVX512VL: # %bb.0:
1118 ; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [1,0,9,8,3,2,11,10]
1119 ; AVX512VL-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1120 ; AVX512VL-NEXT: retq
1121 %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>
1122 ret <8 x i16> %shuffle
1123 }
1125 define <8 x i16> @shuffle_v8i16_8091a2b3(<8 x i16> %a, <8 x i16> %b) {
1126 ; SSE-LABEL: shuffle_v8i16_8091a2b3:
1127 ; SSE: # %bb.0:
1128 ; SSE-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
1129 ; SSE-NEXT: movdqa %xmm1, %xmm0
1130 ; SSE-NEXT: retq
1131 ;
1132 ; AVX-LABEL: shuffle_v8i16_8091a2b3:
1133 ; AVX: # %bb.0:
1134 ; AVX-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
1135 ; AVX-NEXT: retq
1136 %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>
1137 ret <8 x i16> %shuffle
1138 }
1139 define <8 x i16> @shuffle_v8i16_c4d5e6f7(<8 x i16> %a, <8 x i16> %b) {
1140 ; SSE-LABEL: shuffle_v8i16_c4d5e6f7:
1141 ; SSE: # %bb.0:
1142 ; SSE-NEXT: punpckhwd {{.*#+}} xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
1143 ; SSE-NEXT: movdqa %xmm1, %xmm0
1144 ; SSE-NEXT: retq
1145 ;
1146 ; AVX-LABEL: shuffle_v8i16_c4d5e6f7:
1147 ; AVX: # %bb.0:
1148 ; AVX-NEXT: vpunpckhwd {{.*#+}} xmm0 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
1149 ; AVX-NEXT: retq
1150 %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>
1151 ret <8 x i16> %shuffle
1152 }
1154 define <8 x i16> @shuffle_v8i16_0213cedf(<8 x i16> %a, <8 x i16> %b) {
1155 ; SSE2-LABEL: shuffle_v8i16_0213cedf:
1156 ; SSE2: # %bb.0:
1157 ; SSE2-NEXT: pshuflw {{.*#+}} xmm2 = xmm0[0,2,1,3,4,5,6,7]
1158 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm1[0,1,2,3,4,6,5,7]
1159 ; SSE2-NEXT: movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
1160 ; SSE2-NEXT: retq
1161 ;
1162 ; SSSE3-LABEL: shuffle_v8i16_0213cedf:
1163 ; SSSE3: # %bb.0:
1164 ; SSSE3-NEXT: pshuflw {{.*#+}} xmm2 = xmm0[0,2,1,3,4,5,6,7]
1165 ; SSSE3-NEXT: pshufhw {{.*#+}} xmm0 = xmm1[0,1,2,3,4,6,5,7]
1166 ; SSSE3-NEXT: movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
1167 ; SSSE3-NEXT: retq
1168 ;
1169 ; SSE41-LABEL: shuffle_v8i16_0213cedf:
1170 ; SSE41: # %bb.0:
1171 ; SSE41-NEXT: pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,5,7]
1172 ; SSE41-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
1173 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
1174 ; SSE41-NEXT: retq
1175 ;
1176 ; AVX1-LABEL: shuffle_v8i16_0213cedf:
1177 ; AVX1: # %bb.0:
1178 ; AVX1-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,5,7]
1179 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
1180 ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
1181 ; AVX1-NEXT: retq
1182 ;
1183 ; AVX2-SLOW-LABEL: shuffle_v8i16_0213cedf:
1184 ; AVX2-SLOW: # %bb.0:
1185 ; AVX2-SLOW-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,5,7]
1186 ; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
1187 ; AVX2-SLOW-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
1188 ; AVX2-SLOW-NEXT: retq
1189 ;
1190 ; AVX2-FAST-LABEL: shuffle_v8i16_0213cedf:
1191 ; AVX2-FAST: # %bb.0:
1192 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[8,9,12,13,10,11,14,15,8,9,12,13,10,11,14,15]
1193 ; AVX2-FAST-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
1194 ; AVX2-FAST-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
1195 ; AVX2-FAST-NEXT: retq
1196 ;
1197 ; AVX512VL-LABEL: shuffle_v8i16_0213cedf:
1198 ; AVX512VL: # %bb.0:
1199 ; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [0,2,1,3,12,14,13,15]
1200 ; AVX512VL-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1201 ; AVX512VL-NEXT: retq
1202 %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>
1203 ret <8 x i16> %shuffle
1204 }
1206 define <8 x i16> @shuffle_v8i16_443aXXXX(<8 x i16> %a, <8 x i16> %b) {
1207 ; SSE2-LABEL: shuffle_v8i16_443aXXXX:
1208 ; SSE2: # %bb.0:
1209 ; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [65535,65535,0,65535,65535,65535,65535,65535]
1210 ; SSE2-NEXT: pand %xmm2, %xmm0
1211 ; SSE2-NEXT: pandn %xmm1, %xmm2
1212 ; SSE2-NEXT: por %xmm0, %xmm2
1213 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm2[2,1,2,3]
1214 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,3,2,4,5,6,7]
1215 ; SSE2-NEXT: retq
1216 ;
1217 ; SSSE3-LABEL: shuffle_v8i16_443aXXXX:
1218 ; SSSE3: # %bb.0:
1219 ; SSSE3-NEXT: pshufb {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,xmm1[4,5,u,u,u,u,u,u,u,u]
1220 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,6,7],zero,zero,xmm0[u,u,u,u,u,u,u,u]
1221 ; SSSE3-NEXT: por %xmm1, %xmm0
1222 ; SSSE3-NEXT: retq
1223 ;
1224 ; SSE41-LABEL: shuffle_v8i16_443aXXXX:
1225 ; SSE41: # %bb.0:
1226 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7]
1227 ; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,1,2,3]
1228 ; SSE41-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,3,2,4,5,6,7]
1229 ; SSE41-NEXT: retq
1230 ;
1231 ; AVX1-LABEL: shuffle_v8i16_443aXXXX:
1232 ; AVX1: # %bb.0:
1233 ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7]
1234 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,1,2,3]
1235 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,3,2,4,5,6,7]
1236 ; AVX1-NEXT: retq
1237 ;
1238 ; AVX2-SLOW-LABEL: shuffle_v8i16_443aXXXX:
1239 ; AVX2-SLOW: # %bb.0:
1240 ; AVX2-SLOW-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7]
1241 ; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,1,2,3]
1242 ; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,3,2,4,5,6,7]
1243 ; AVX2-SLOW-NEXT: retq
1244 ;
1245 ; AVX2-FAST-LABEL: shuffle_v8i16_443aXXXX:
1246 ; AVX2-FAST: # %bb.0:
1247 ; AVX2-FAST-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7]
1248 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,6,7,4,5,8,9,10,11,12,13,14,15]
1249 ; AVX2-FAST-NEXT: retq
1250 ;
1251 ; AVX512VL-LABEL: shuffle_v8i16_443aXXXX:
1252 ; AVX512VL: # %bb.0:
1253 ; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [4,4,3,10,4,5,6,7]
1254 ; AVX512VL-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1255 ; AVX512VL-NEXT: retq
1256 %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>
1257 ret <8 x i16> %shuffle
1258 }
1260 define <8 x i16> @shuffle_v8i16_032dXXXX(<8 x i16> %a, <8 x i16> %b) {
1261 ; SSE2-LABEL: shuffle_v8i16_032dXXXX:
1262 ; SSE2: # %bb.0:
1263 ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
1264 ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[3,1,2,0]
1265 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
1266 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,1,2,3]
1267 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,1,4,5,6,7]
1268 ; SSE2-NEXT: retq
1269 ;
1270 ; SSSE3-LABEL: shuffle_v8i16_032dXXXX:
1271 ; SSSE3: # %bb.0:
1272 ; SSSE3-NEXT: pshufb {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,xmm1[10,11,u,u,u,u,u,u,u,u]
1273 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,6,7,4,5],zero,zero,xmm0[u,u,u,u,u,u,u,u]
1274 ; SSSE3-NEXT: por %xmm1, %xmm0
1275 ; SSSE3-NEXT: retq
1276 ;
1277 ; SSE41-LABEL: shuffle_v8i16_032dXXXX:
1278 ; SSE41: # %bb.0:
1279 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
1280 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,6,7,4,5,10,11,0,1,10,11,0,1,2,3]
1281 ; SSE41-NEXT: retq
1282 ;
1283 ; AVX1-LABEL: shuffle_v8i16_032dXXXX:
1284 ; AVX1: # %bb.0:
1285 ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
1286 ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,6,7,4,5,10,11,0,1,10,11,0,1,2,3]
1287 ; AVX1-NEXT: retq
1288 ;
1289 ; AVX2-LABEL: shuffle_v8i16_032dXXXX:
1290 ; AVX2: # %bb.0:
1291 ; AVX2-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
1292 ; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,6,7,4,5,10,11,0,1,10,11,0,1,2,3]
1293 ; AVX2-NEXT: retq
1294 ;
1295 ; AVX512VL-LABEL: shuffle_v8i16_032dXXXX:
1296 ; AVX512VL: # %bb.0:
1297 ; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [0,3,2,13,0,13,0,1]
1298 ; AVX512VL-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1299 ; AVX512VL-NEXT: retq
1300 %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>
1301 ret <8 x i16> %shuffle
1302 }
1303 define <8 x i16> @shuffle_v8i16_XXXdXXXX(<8 x i16> %a, <8 x i16> %b) {
1304 ; SSE-LABEL: shuffle_v8i16_XXXdXXXX:
1305 ; SSE: # %bb.0:
1306 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm1[2,2,3,3]
1307 ; SSE-NEXT: retq
1308 ;
1309 ; AVX-LABEL: shuffle_v8i16_XXXdXXXX:
1310 ; AVX: # %bb.0:
1311 ; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm1[2,2,3,3]
1312 ; AVX-NEXT: retq
1313 %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>
1314 ret <8 x i16> %shuffle
1315 }
1317 define <8 x i16> @shuffle_v8i16_012dXXXX(<8 x i16> %a, <8 x i16> %b) {
1318 ; SSE2-LABEL: shuffle_v8i16_012dXXXX:
1319 ; SSE2: # %bb.0:
1320 ; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [65535,65535,65535,0,65535,65535,65535,65535]
1321 ; SSE2-NEXT: pand %xmm2, %xmm0
1322 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,2,3,3]
1323 ; SSE2-NEXT: pandn %xmm1, %xmm2
1324 ; SSE2-NEXT: por %xmm2, %xmm0
1325 ; SSE2-NEXT: retq
1326 ;
1327 ; SSSE3-LABEL: shuffle_v8i16_012dXXXX:
1328 ; SSSE3: # %bb.0:
1329 ; SSSE3-NEXT: pshufb {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,xmm1[10,11,u,u,u,u,u,u,u,u]
1330 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5],zero,zero,xmm0[u,u,u,u,u,u,u,u]
1331 ; SSSE3-NEXT: por %xmm1, %xmm0
1332 ; SSSE3-NEXT: retq
1333 ;
1334 ; SSE41-LABEL: shuffle_v8i16_012dXXXX:
1335 ; SSE41: # %bb.0:
1336 ; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,2,3,3]
1337 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7]
1338 ; SSE41-NEXT: retq
1339 ;
1340 ; AVX1OR2-LABEL: shuffle_v8i16_012dXXXX:
1341 ; AVX1OR2: # %bb.0:
1342 ; AVX1OR2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,2,3,3]
1343 ; AVX1OR2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7]
1344 ; AVX1OR2-NEXT: retq
1345 ;
1346 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_012dXXXX:
1347 ; AVX512VL-SLOW: # %bb.0:
1348 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,2,3,3]
1349 ; AVX512VL-SLOW-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7]
1350 ; AVX512VL-SLOW-NEXT: retq
1351 ;
1352 ; AVX512VL-FAST-LABEL: shuffle_v8i16_012dXXXX:
1353 ; AVX512VL-FAST: # %bb.0:
1354 ; AVX512VL-FAST-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,2,13,4,5,6,7]
1355 ; AVX512VL-FAST-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1356 ; AVX512VL-FAST-NEXT: retq
1357 %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>
1358 ret <8 x i16> %shuffle
1359 }
1361 define <8 x i16> @shuffle_v8i16_XXXXcde3(<8 x i16> %a, <8 x i16> %b) {
1362 ; SSE2-LABEL: shuffle_v8i16_XXXXcde3:
1363 ; SSE2: # %bb.0:
1364 ; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [65535,65535,65535,65535,65535,65535,65535,0]
1365 ; SSE2-NEXT: pand %xmm2, %xmm1
1366 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
1367 ; SSE2-NEXT: pandn %xmm0, %xmm2
1368 ; SSE2-NEXT: por %xmm1, %xmm2
1369 ; SSE2-NEXT: movdqa %xmm2, %xmm0
1370 ; SSE2-NEXT: retq
1371 ;
1372 ; SSSE3-LABEL: shuffle_v8i16_XXXXcde3:
1373 ; SSSE3: # %bb.0:
1374 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[u,u,u,u,u,u,u,u],zero,zero,zero,zero,zero,zero,xmm0[6,7]
1375 ; SSSE3-NEXT: pshufb {{.*#+}} xmm1 = xmm1[u,u,u,u,u,u,u,u,8,9,10,11,12,13],zero,zero
1376 ; SSSE3-NEXT: por %xmm1, %xmm0
1377 ; SSSE3-NEXT: retq
1378 ;
1379 ; SSE41-LABEL: shuffle_v8i16_XXXXcde3:
1380 ; SSE41: # %bb.0:
1381 ; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
1382 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm1[0,1,2,3,4,5,6],xmm0[7]
1383 ; SSE41-NEXT: retq
1384 ;
1385 ; AVX1-LABEL: shuffle_v8i16_XXXXcde3:
1386 ; AVX1: # %bb.0:
1387 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
1388 ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3,4,5,6],xmm0[7]
1389 ; AVX1-NEXT: retq
1390 ;
1391 ; AVX2-LABEL: shuffle_v8i16_XXXXcde3:
1392 ; AVX2: # %bb.0:
1393 ; AVX2-NEXT: vpbroadcastq %xmm0, %xmm0
1394 ; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3,4,5,6],xmm0[7]
1395 ; AVX2-NEXT: retq
1396 ;
1397 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_XXXXcde3:
1398 ; AVX512VL-SLOW: # %bb.0:
1399 ; AVX512VL-SLOW-NEXT: vpbroadcastq %xmm0, %xmm0
1400 ; AVX512VL-SLOW-NEXT: vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3,4,5,6],xmm0[7]
1401 ; AVX512VL-SLOW-NEXT: retq
1402 ;
1403 ; AVX512VL-FAST-LABEL: shuffle_v8i16_XXXXcde3:
1404 ; AVX512VL-FAST: # %bb.0:
1405 ; AVX512VL-FAST-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,2,3,4,5,6,11]
1406 ; AVX512VL-FAST-NEXT: vpermi2w %xmm0, %xmm1, %xmm2
1407 ; AVX512VL-FAST-NEXT: vmovdqa %xmm2, %xmm0
1408 ; AVX512VL-FAST-NEXT: retq
1409 %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>
1410 ret <8 x i16> %shuffle
1411 }
1413 define <8 x i16> @shuffle_v8i16_cde3XXXX(<8 x i16> %a, <8 x i16> %b) {
1414 ; SSE2-LABEL: shuffle_v8i16_cde3XXXX:
1415 ; SSE2: # %bb.0:
1416 ; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [65535,65535,65535,0,65535,65535,65535,65535]
1417 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
1418 ; SSE2-NEXT: pand %xmm2, %xmm1
1419 ; SSE2-NEXT: pandn %xmm0, %xmm2
1420 ; SSE2-NEXT: por %xmm1, %xmm2
1421 ; SSE2-NEXT: movdqa %xmm2, %xmm0
1422 ; SSE2-NEXT: retq
1423 ;
1424 ; SSSE3-LABEL: shuffle_v8i16_cde3XXXX:
1425 ; SSSE3: # %bb.0:
1426 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,xmm0[6,7,u,u,u,u,u,u,u,u]
1427 ; SSSE3-NEXT: pshufb {{.*#+}} xmm1 = xmm1[8,9,10,11,12,13],zero,zero,xmm1[u,u,u,u,u,u,u,u]
1428 ; SSSE3-NEXT: por %xmm1, %xmm0
1429 ; SSSE3-NEXT: retq
1430 ;
1431 ; SSE41-LABEL: shuffle_v8i16_cde3XXXX:
1432 ; SSE41: # %bb.0:
1433 ; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
1434 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[3],xmm1[4,5,6,7]
1435 ; SSE41-NEXT: retq
1436 ;
1437 ; AVX1OR2-LABEL: shuffle_v8i16_cde3XXXX:
1438 ; AVX1OR2: # %bb.0:
1439 ; AVX1OR2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
1440 ; AVX1OR2-NEXT: vpblendw {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[3],xmm1[4,5,6,7]
1441 ; AVX1OR2-NEXT: retq
1442 ;
1443 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_cde3XXXX:
1444 ; AVX512VL-SLOW: # %bb.0:
1445 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
1446 ; AVX512VL-SLOW-NEXT: vpblendw {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[3],xmm1[4,5,6,7]
1447 ; AVX512VL-SLOW-NEXT: retq
1448 ;
1449 ; AVX512VL-FAST-LABEL: shuffle_v8i16_cde3XXXX:
1450 ; AVX512VL-FAST: # %bb.0:
1451 ; AVX512VL-FAST-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,6,11,0,1,2,3]
1452 ; AVX512VL-FAST-NEXT: vpermi2w %xmm0, %xmm1, %xmm2
1453 ; AVX512VL-FAST-NEXT: vmovdqa %xmm2, %xmm0
1454 ; AVX512VL-FAST-NEXT: retq
1455 %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>
1456 ret <8 x i16> %shuffle
1457 }
1459 define <8 x i16> @shuffle_v8i16_012dcde3(<8 x i16> %a, <8 x i16> %b) {
1460 ; SSE2-LABEL: shuffle_v8i16_012dcde3:
1461 ; SSE2: # %bb.0:
1462 ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
1463 ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,3,2,1]
1464 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,0,4,5,6,7]
1465 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
1466 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,1,4,5,6,7]
1467 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,5,7]
1468 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,3,2,1]
1469 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,3,0,2,4,5,6,7]
1470 ; SSE2-NEXT: retq
1471 ;
1472 ; SSSE3-LABEL: shuffle_v8i16_012dcde3:
1473 ; SSSE3: # %bb.0:
1474 ; SSSE3-NEXT: pshufb {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,xmm1[10,11,8,9,10,11,12,13],zero,zero
1475 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5],zero,zero,zero,zero,zero,zero,zero,zero,xmm0[6,7]
1476 ; SSSE3-NEXT: por %xmm1, %xmm0
1477 ; SSSE3-NEXT: retq
1478 ;
1479 ; SSE41-LABEL: shuffle_v8i16_012dcde3:
1480 ; SSE41: # %bb.0:
1481 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
1482 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7]
1483 ; SSE41-NEXT: retq
1484 ;
1485 ; AVX1-LABEL: shuffle_v8i16_012dcde3:
1486 ; AVX1: # %bb.0:
1487 ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
1488 ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7]
1489 ; AVX1-NEXT: retq
1490 ;
1491 ; AVX2-LABEL: shuffle_v8i16_012dcde3:
1492 ; AVX2: # %bb.0:
1493 ; AVX2-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
1494 ; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7]
1495 ; AVX2-NEXT: retq
1496 ;
1497 ; AVX512VL-LABEL: shuffle_v8i16_012dcde3:
1498 ; AVX512VL: # %bb.0:
1499 ; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,2,13,12,13,14,3]
1500 ; AVX512VL-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1501 ; AVX512VL-NEXT: retq
1502 %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>
1503 ret <8 x i16> %shuffle
1504 }
1506 define <8 x i16> @shuffle_v8i16_0923cde7(<8 x i16> %a, <8 x i16> %b) {
1507 ; SSE2-LABEL: shuffle_v8i16_0923cde7:
1508 ; SSE2: # %bb.0:
1509 ; SSE2-NEXT: movaps {{.*#+}} xmm2 = [65535,0,65535,65535,0,0,0,65535]
1510 ; SSE2-NEXT: andps %xmm2, %xmm0
1511 ; SSE2-NEXT: andnps %xmm1, %xmm2
1512 ; SSE2-NEXT: orps %xmm2, %xmm0
1513 ; SSE2-NEXT: retq
1514 ;
1515 ; SSSE3-LABEL: shuffle_v8i16_0923cde7:
1516 ; SSSE3: # %bb.0:
1517 ; SSSE3-NEXT: movaps {{.*#+}} xmm2 = [65535,0,65535,65535,0,0,0,65535]
1518 ; SSSE3-NEXT: andps %xmm2, %xmm0
1519 ; SSSE3-NEXT: andnps %xmm1, %xmm2
1520 ; SSSE3-NEXT: orps %xmm2, %xmm0
1521 ; SSSE3-NEXT: retq
1522 ;
1523 ; SSE41-LABEL: shuffle_v8i16_0923cde7:
1524 ; SSE41: # %bb.0:
1525 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3],xmm1[4,5,6],xmm0[7]
1526 ; SSE41-NEXT: retq
1527 ;
1528 ; AVX-LABEL: shuffle_v8i16_0923cde7:
1529 ; AVX: # %bb.0:
1530 ; AVX-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3],xmm1[4,5,6],xmm0[7]
1531 ; AVX-NEXT: retq
1532 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 9, i32 2, i32 3, i32 12, i32 13, i32 14, i32 7>
1533 ret <8 x i16> %shuffle
1534 }
1536 define <8 x i16> @shuffle_v8i16_XXX1X579(<8 x i16> %a, <8 x i16> %b) {
1537 ; SSE2-LABEL: shuffle_v8i16_XXX1X579:
1538 ; SSE2: # %bb.0:
1539 ; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm1[0,1,2,0]
1540 ; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [65535,65535,65535,65535,65535,65535,65535,0]
1541 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,1,2,1,4,5,6,7]
1542 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,7,7]
1543 ; SSE2-NEXT: pand %xmm1, %xmm0
1544 ; SSE2-NEXT: pandn %xmm2, %xmm1
1545 ; SSE2-NEXT: por %xmm0, %xmm1
1546 ; SSE2-NEXT: movdqa %xmm1, %xmm0
1547 ; SSE2-NEXT: retq
1548 ;
1549 ; SSSE3-LABEL: shuffle_v8i16_XXX1X579:
1550 ; SSSE3: # %bb.0:
1551 ; SSSE3-NEXT: pshufb {{.*#+}} xmm1 = xmm1[u,u,u,u,u,u],zero,zero,xmm1[u,u],zero,zero,zero,zero,xmm1[2,3]
1552 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[u,u,u,u,u,u,2,3,u,u,10,11,14,15],zero,zero
1553 ; SSSE3-NEXT: por %xmm1, %xmm0
1554 ; SSSE3-NEXT: retq
1555 ;
1556 ; SSE41-LABEL: shuffle_v8i16_XXX1X579:
1557 ; SSE41: # %bb.0:
1558 ; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,1,2,0]
1559 ; SSE41-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,1,2,1,4,5,6,7]
1560 ; SSE41-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,7,7]
1561 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
1562 ; SSE41-NEXT: retq
1563 ;
1564 ; AVX1-LABEL: shuffle_v8i16_XXX1X579:
1565 ; AVX1: # %bb.0:
1566 ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,1,2,0]
1567 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,1,2,1,4,5,6,7]
1568 ; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,7,7]
1569 ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
1570 ; AVX1-NEXT: retq
1571 ;
1572 ; AVX2-SLOW-LABEL: shuffle_v8i16_XXX1X579:
1573 ; AVX2-SLOW: # %bb.0:
1574 ; AVX2-SLOW-NEXT: vpbroadcastd %xmm1, %xmm1
1575 ; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,1,2,1,4,5,6,7]
1576 ; AVX2-SLOW-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,7,7]
1577 ; AVX2-SLOW-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
1578 ; AVX2-SLOW-NEXT: retq
1579 ;
1580 ; AVX2-FAST-LABEL: shuffle_v8i16_XXX1X579:
1581 ; AVX2-FAST: # %bb.0:
1582 ; AVX2-FAST-NEXT: vpbroadcastd %xmm1, %xmm1
1583 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,2,3,8,9,10,11,14,15,14,15]
1584 ; AVX2-FAST-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
1585 ; AVX2-FAST-NEXT: retq
1586 ;
1587 ; AVX512VL-LABEL: shuffle_v8i16_XXX1X579:
1588 ; AVX512VL: # %bb.0:
1589 ; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,2,1,4,5,7,9]
1590 ; AVX512VL-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1591 ; AVX512VL-NEXT: retq
1592 %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>
1593 ret <8 x i16> %shuffle
1594 }
1596 define <8 x i16> @shuffle_v8i16_XX4X8acX(<8 x i16> %a, <8 x i16> %b) {
1597 ; SSE2-LABEL: shuffle_v8i16_XX4X8acX:
1598 ; SSE2: # %bb.0:
1599 ; SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
1600 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,1,2,0]
1601 ; SSE2-NEXT: pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,7,4,7]
1602 ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[2,2],xmm1[2,3]
1603 ; SSE2-NEXT: retq
1604 ;
1605 ; SSSE3-LABEL: shuffle_v8i16_XX4X8acX:
1606 ; SSSE3: # %bb.0:
1607 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[u,u,u,u,8,9,u,u],zero,zero,zero,zero,zero,zero,xmm0[u,u]
1608 ; SSSE3-NEXT: pshufb {{.*#+}} xmm1 = xmm1[u,u,u,u],zero,zero,xmm1[u,u,0,1,4,5,8,9,u,u]
1609 ; SSSE3-NEXT: por %xmm1, %xmm0
1610 ; SSSE3-NEXT: retq
1611 ;
1612 ; SSE41-LABEL: shuffle_v8i16_XX4X8acX:
1613 ; SSE41: # %bb.0:
1614 ; SSE41-NEXT: pshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,4,5]
1615 ; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,2,3,3]
1616 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
1617 ; SSE41-NEXT: retq
1618 ;
1619 ; AVX1-LABEL: shuffle_v8i16_XX4X8acX:
1620 ; AVX1: # %bb.0:
1621 ; AVX1-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,4,5]
1622 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,2,3,3]
1623 ; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
1624 ; AVX1-NEXT: retq
1625 ;
1626 ; AVX2-LABEL: shuffle_v8i16_XX4X8acX:
1627 ; AVX2: # %bb.0:
1628 ; AVX2-NEXT: vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,4,5]
1629 ; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,2,3,3]
1630 ; AVX2-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
1631 ; AVX2-NEXT: retq
1632 ;
1633 ; AVX512VL-LABEL: shuffle_v8i16_XX4X8acX:
1634 ; AVX512VL: # %bb.0:
1635 ; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [4,5,4,5,8,10,12,10]
1636 ; AVX512VL-NEXT: vpermt2w %xmm1, %xmm2, %xmm0
1637 ; AVX512VL-NEXT: retq
1638 %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>
1639 ret <8 x i16> %shuffle
1640 }
1642 define <8 x i16> @shuffle_v8i16_8zzzzzzz(i16 %i) {
1643 ; SSE-LABEL: shuffle_v8i16_8zzzzzzz:
1644 ; SSE: # %bb.0:
1645 ; SSE-NEXT: movzwl %di, %eax
1646 ; SSE-NEXT: movd %eax, %xmm0
1647 ; SSE-NEXT: retq
1648 ;
1649 ; AVX-LABEL: shuffle_v8i16_8zzzzzzz:
1650 ; AVX: # %bb.0:
1651 ; AVX-NEXT: movzwl %di, %eax
1652 ; AVX-NEXT: vmovd %eax, %xmm0
1653 ; AVX-NEXT: retq
1654 %a = insertelement <8 x i16> undef, i16 %i, i32 0
1655 %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>
1656 ret <8 x i16> %shuffle
1657 }
1659 define <8 x i16> @shuffle_v8i16_z8zzzzzz(i16 %i) {
1660 ; SSE-LABEL: shuffle_v8i16_z8zzzzzz:
1661 ; SSE: # %bb.0:
1662 ; SSE-NEXT: pxor %xmm0, %xmm0
1663 ; SSE-NEXT: pinsrw $1, %edi, %xmm0
1664 ; SSE-NEXT: retq
1665 ;
1666 ; AVX-LABEL: shuffle_v8i16_z8zzzzzz:
1667 ; AVX: # %bb.0:
1668 ; AVX-NEXT: vpxor %xmm0, %xmm0, %xmm0
1669 ; AVX-NEXT: vpinsrw $1, %edi, %xmm0, %xmm0
1670 ; AVX-NEXT: retq
1671 %a = insertelement <8 x i16> undef, i16 %i, i32 0
1672 %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>
1673 ret <8 x i16> %shuffle
1674 }
1676 define <8 x i16> @shuffle_v8i16_zzzzz8zz(i16 %i) {
1677 ; SSE-LABEL: shuffle_v8i16_zzzzz8zz:
1678 ; SSE: # %bb.0:
1679 ; SSE-NEXT: pxor %xmm0, %xmm0
1680 ; SSE-NEXT: pinsrw $5, %edi, %xmm0
1681 ; SSE-NEXT: retq
1682 ;
1683 ; AVX-LABEL: shuffle_v8i16_zzzzz8zz:
1684 ; AVX: # %bb.0:
1685 ; AVX-NEXT: vpxor %xmm0, %xmm0, %xmm0
1686 ; AVX-NEXT: vpinsrw $5, %edi, %xmm0, %xmm0
1687 ; AVX-NEXT: retq
1688 %a = insertelement <8 x i16> undef, i16 %i, i32 0
1689 %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>
1690 ret <8 x i16> %shuffle
1691 }
1693 define <8 x i16> @shuffle_v8i16_zuuzuuz8(i16 %i) {
1694 ; SSE-LABEL: shuffle_v8i16_zuuzuuz8:
1695 ; SSE: # %bb.0:
1696 ; SSE-NEXT: pxor %xmm0, %xmm0
1697 ; SSE-NEXT: pinsrw $7, %edi, %xmm0
1698 ; SSE-NEXT: retq
1699 ;
1700 ; AVX-LABEL: shuffle_v8i16_zuuzuuz8:
1701 ; AVX: # %bb.0:
1702 ; AVX-NEXT: vpxor %xmm0, %xmm0, %xmm0
1703 ; AVX-NEXT: vpinsrw $7, %edi, %xmm0, %xmm0
1704 ; AVX-NEXT: retq
1705 %a = insertelement <8 x i16> undef, i16 %i, i32 0
1706 %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>
1707 ret <8 x i16> %shuffle
1708 }
1710 define <8 x i16> @shuffle_v8i16_zzBzzzzz(i16 %i) {
1711 ; SSE-LABEL: shuffle_v8i16_zzBzzzzz:
1712 ; SSE: # %bb.0:
1713 ; SSE-NEXT: pxor %xmm0, %xmm0
1714 ; SSE-NEXT: pinsrw $2, %edi, %xmm0
1715 ; SSE-NEXT: retq
1716 ;
1717 ; AVX-LABEL: shuffle_v8i16_zzBzzzzz:
1718 ; AVX: # %bb.0:
1719 ; AVX-NEXT: vpxor %xmm0, %xmm0, %xmm0
1720 ; AVX-NEXT: vpinsrw $2, %edi, %xmm0, %xmm0
1721 ; AVX-NEXT: retq
1722 %a = insertelement <8 x i16> undef, i16 %i, i32 3
1723 %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>
1724 ret <8 x i16> %shuffle
1725 }
1727 define <8 x i16> @shuffle_v8i16_def01234(<8 x i16> %a, <8 x i16> %b) {
1728 ; SSE2-LABEL: shuffle_v8i16_def01234:
1729 ; SSE2: # %bb.0:
1730 ; SSE2-NEXT: psrldq {{.*#+}} xmm1 = xmm1[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
1731 ; SSE2-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9]
1732 ; SSE2-NEXT: por %xmm1, %xmm0
1733 ; SSE2-NEXT: retq
1734 ;
1735 ; SSSE3-LABEL: shuffle_v8i16_def01234:
1736 ; SSSE3: # %bb.0:
1737 ; SSSE3-NEXT: palignr {{.*#+}} xmm0 = xmm1[10,11,12,13,14,15],xmm0[0,1,2,3,4,5,6,7,8,9]
1738 ; SSSE3-NEXT: retq
1739 ;
1740 ; SSE41-LABEL: shuffle_v8i16_def01234:
1741 ; SSE41: # %bb.0:
1742 ; SSE41-NEXT: palignr {{.*#+}} xmm0 = xmm1[10,11,12,13,14,15],xmm0[0,1,2,3,4,5,6,7,8,9]
1743 ; SSE41-NEXT: retq
1744 ;
1745 ; AVX-LABEL: shuffle_v8i16_def01234:
1746 ; AVX: # %bb.0:
1747 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm1[10,11,12,13,14,15],xmm0[0,1,2,3,4,5,6,7,8,9]
1748 ; AVX-NEXT: retq
1749 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 13, i32 14, i32 15, i32 0, i32 1, i32 2, i32 3, i32 4>
1750 ret <8 x i16> %shuffle
1751 }
1753 define <8 x i16> @shuffle_v8i16_ueuu123u(<8 x i16> %a, <8 x i16> %b) {
1754 ; SSE2-LABEL: shuffle_v8i16_ueuu123u:
1755 ; SSE2: # %bb.0:
1756 ; SSE2-NEXT: psrldq {{.*#+}} xmm1 = xmm1[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
1757 ; SSE2-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9]
1758 ; SSE2-NEXT: por %xmm1, %xmm0
1759 ; SSE2-NEXT: retq
1760 ;
1761 ; SSSE3-LABEL: shuffle_v8i16_ueuu123u:
1762 ; SSSE3: # %bb.0:
1763 ; SSSE3-NEXT: palignr {{.*#+}} xmm0 = xmm1[10,11,12,13,14,15],xmm0[0,1,2,3,4,5,6,7,8,9]
1764 ; SSSE3-NEXT: retq
1765 ;
1766 ; SSE41-LABEL: shuffle_v8i16_ueuu123u:
1767 ; SSE41: # %bb.0:
1768 ; SSE41-NEXT: palignr {{.*#+}} xmm0 = xmm1[10,11,12,13,14,15],xmm0[0,1,2,3,4,5,6,7,8,9]
1769 ; SSE41-NEXT: retq
1770 ;
1771 ; AVX-LABEL: shuffle_v8i16_ueuu123u:
1772 ; AVX: # %bb.0:
1773 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm1[10,11,12,13,14,15],xmm0[0,1,2,3,4,5,6,7,8,9]
1774 ; AVX-NEXT: retq
1775 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 14, i32 undef, i32 undef, i32 1, i32 2, i32 3, i32 undef>
1776 ret <8 x i16> %shuffle
1777 }
1779 define <8 x i16> @shuffle_v8i16_56701234(<8 x i16> %a, <8 x i16> %b) {
1780 ; SSE2-LABEL: shuffle_v8i16_56701234:
1781 ; SSE2: # %bb.0:
1782 ; SSE2-NEXT: movdqa %xmm0, %xmm1
1783 ; SSE2-NEXT: psrldq {{.*#+}} xmm1 = xmm1[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
1784 ; SSE2-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9]
1785 ; SSE2-NEXT: por %xmm1, %xmm0
1786 ; SSE2-NEXT: retq
1787 ;
1788 ; SSSE3-LABEL: shuffle_v8i16_56701234:
1789 ; SSSE3: # %bb.0:
1790 ; SSSE3-NEXT: palignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
1791 ; SSSE3-NEXT: retq
1792 ;
1793 ; SSE41-LABEL: shuffle_v8i16_56701234:
1794 ; SSE41: # %bb.0:
1795 ; SSE41-NEXT: palignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
1796 ; SSE41-NEXT: retq
1797 ;
1798 ; AVX-LABEL: shuffle_v8i16_56701234:
1799 ; AVX: # %bb.0:
1800 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
1801 ; AVX-NEXT: retq
1802 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4>
1803 ret <8 x i16> %shuffle
1804 }
1806 define <8 x i16> @shuffle_v8i16_u6uu123u(<8 x i16> %a, <8 x i16> %b) {
1807 ; SSE2-LABEL: shuffle_v8i16_u6uu123u:
1808 ; SSE2: # %bb.0:
1809 ; SSE2-NEXT: movdqa %xmm0, %xmm1
1810 ; SSE2-NEXT: psrldq {{.*#+}} xmm1 = xmm1[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
1811 ; SSE2-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9]
1812 ; SSE2-NEXT: por %xmm1, %xmm0
1813 ; SSE2-NEXT: retq
1814 ;
1815 ; SSSE3-LABEL: shuffle_v8i16_u6uu123u:
1816 ; SSSE3: # %bb.0:
1817 ; SSSE3-NEXT: palignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
1818 ; SSSE3-NEXT: retq
1819 ;
1820 ; SSE41-LABEL: shuffle_v8i16_u6uu123u:
1821 ; SSE41: # %bb.0:
1822 ; SSE41-NEXT: palignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
1823 ; SSE41-NEXT: retq
1824 ;
1825 ; AVX-LABEL: shuffle_v8i16_u6uu123u:
1826 ; AVX: # %bb.0:
1827 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
1828 ; AVX-NEXT: retq
1829 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 6, i32 undef, i32 undef, i32 1, i32 2, i32 3, i32 undef>
1830 ret <8 x i16> %shuffle
1831 }
1833 define <8 x i16> @shuffle_v8i16_uuuu123u(<8 x i16> %a, <8 x i16> %b) {
1834 ; SSE-LABEL: shuffle_v8i16_uuuu123u:
1835 ; SSE: # %bb.0:
1836 ; SSE-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9]
1837 ; SSE-NEXT: retq
1838 ;
1839 ; AVX-LABEL: shuffle_v8i16_uuuu123u:
1840 ; AVX: # %bb.0:
1841 ; AVX-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9]
1842 ; AVX-NEXT: retq
1843 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 undef, i32 undef, i32 undef, i32 1, i32 2, i32 3, i32 undef>
1844 ret <8 x i16> %shuffle
1845 }
1847 define <8 x i16> @shuffle_v8i16_bcdef012(<8 x i16> %a, <8 x i16> %b) {
1848 ; SSE2-LABEL: shuffle_v8i16_bcdef012:
1849 ; SSE2: # %bb.0:
1850 ; SSE2-NEXT: psrldq {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
1851 ; SSE2-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
1852 ; SSE2-NEXT: por %xmm1, %xmm0
1853 ; SSE2-NEXT: retq
1854 ;
1855 ; SSSE3-LABEL: shuffle_v8i16_bcdef012:
1856 ; SSSE3: # %bb.0:
1857 ; SSSE3-NEXT: palignr {{.*#+}} xmm0 = xmm1[6,7,8,9,10,11,12,13,14,15],xmm0[0,1,2,3,4,5]
1858 ; SSSE3-NEXT: retq
1859 ;
1860 ; SSE41-LABEL: shuffle_v8i16_bcdef012:
1861 ; SSE41: # %bb.0:
1862 ; SSE41-NEXT: palignr {{.*#+}} xmm0 = xmm1[6,7,8,9,10,11,12,13,14,15],xmm0[0,1,2,3,4,5]
1863 ; SSE41-NEXT: retq
1864 ;
1865 ; AVX-LABEL: shuffle_v8i16_bcdef012:
1866 ; AVX: # %bb.0:
1867 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm1[6,7,8,9,10,11,12,13,14,15],xmm0[0,1,2,3,4,5]
1868 ; AVX-NEXT: retq
1869 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 11, i32 12, i32 13, i32 14, i32 15, i32 0, i32 1, i32 2>
1870 ret <8 x i16> %shuffle
1871 }
1873 define <8 x i16> @shuffle_v8i16_ucdeuu1u(<8 x i16> %a, <8 x i16> %b) {
1874 ; SSE2-LABEL: shuffle_v8i16_ucdeuu1u:
1875 ; SSE2: # %bb.0:
1876 ; SSE2-NEXT: psrldq {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
1877 ; SSE2-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
1878 ; SSE2-NEXT: por %xmm1, %xmm0
1879 ; SSE2-NEXT: retq
1880 ;
1881 ; SSSE3-LABEL: shuffle_v8i16_ucdeuu1u:
1882 ; SSSE3: # %bb.0:
1883 ; SSSE3-NEXT: palignr {{.*#+}} xmm0 = xmm1[6,7,8,9,10,11,12,13,14,15],xmm0[0,1,2,3,4,5]
1884 ; SSSE3-NEXT: retq
1885 ;
1886 ; SSE41-LABEL: shuffle_v8i16_ucdeuu1u:
1887 ; SSE41: # %bb.0:
1888 ; SSE41-NEXT: palignr {{.*#+}} xmm0 = xmm1[6,7,8,9,10,11,12,13,14,15],xmm0[0,1,2,3,4,5]
1889 ; SSE41-NEXT: retq
1890 ;
1891 ; AVX-LABEL: shuffle_v8i16_ucdeuu1u:
1892 ; AVX: # %bb.0:
1893 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm1[6,7,8,9,10,11,12,13,14,15],xmm0[0,1,2,3,4,5]
1894 ; AVX-NEXT: retq
1895 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 12, i32 13, i32 14, i32 undef, i32 undef, i32 1, i32 undef>
1896 ret <8 x i16> %shuffle
1897 }
1899 define <8 x i16> @shuffle_v8i16_34567012(<8 x i16> %a, <8 x i16> %b) {
1900 ; SSE2-LABEL: shuffle_v8i16_34567012:
1901 ; SSE2: # %bb.0:
1902 ; SSE2-NEXT: movdqa %xmm0, %xmm1
1903 ; SSE2-NEXT: psrldq {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
1904 ; SSE2-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
1905 ; SSE2-NEXT: por %xmm1, %xmm0
1906 ; SSE2-NEXT: retq
1907 ;
1908 ; SSSE3-LABEL: shuffle_v8i16_34567012:
1909 ; SSSE3: # %bb.0:
1910 ; SSSE3-NEXT: palignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
1911 ; SSSE3-NEXT: retq
1912 ;
1913 ; SSE41-LABEL: shuffle_v8i16_34567012:
1914 ; SSE41: # %bb.0:
1915 ; SSE41-NEXT: palignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
1916 ; SSE41-NEXT: retq
1917 ;
1918 ; AVX-LABEL: shuffle_v8i16_34567012:
1919 ; AVX: # %bb.0:
1920 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
1921 ; AVX-NEXT: retq
1922 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2>
1923 ret <8 x i16> %shuffle
1924 }
1926 define <8 x i16> @shuffle_v8i16_u456uu1u(<8 x i16> %a, <8 x i16> %b) {
1927 ; SSE2-LABEL: shuffle_v8i16_u456uu1u:
1928 ; SSE2: # %bb.0:
1929 ; SSE2-NEXT: movdqa %xmm0, %xmm1
1930 ; SSE2-NEXT: psrldq {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
1931 ; SSE2-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
1932 ; SSE2-NEXT: por %xmm1, %xmm0
1933 ; SSE2-NEXT: retq
1934 ;
1935 ; SSSE3-LABEL: shuffle_v8i16_u456uu1u:
1936 ; SSSE3: # %bb.0:
1937 ; SSSE3-NEXT: palignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
1938 ; SSSE3-NEXT: retq
1939 ;
1940 ; SSE41-LABEL: shuffle_v8i16_u456uu1u:
1941 ; SSE41: # %bb.0:
1942 ; SSE41-NEXT: palignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
1943 ; SSE41-NEXT: retq
1944 ;
1945 ; AVX-LABEL: shuffle_v8i16_u456uu1u:
1946 ; AVX: # %bb.0:
1947 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
1948 ; AVX-NEXT: retq
1949 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 4, i32 5, i32 6, i32 undef, i32 undef, i32 1, i32 undef>
1950 ret <8 x i16> %shuffle
1951 }
1953 define <8 x i16> @shuffle_v8i16_u456uuuu(<8 x i16> %a, <8 x i16> %b) {
1954 ; SSE-LABEL: shuffle_v8i16_u456uuuu:
1955 ; SSE: # %bb.0:
1956 ; SSE-NEXT: psrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
1957 ; SSE-NEXT: retq
1958 ;
1959 ; AVX-LABEL: shuffle_v8i16_u456uuuu:
1960 ; AVX: # %bb.0:
1961 ; AVX-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
1962 ; AVX-NEXT: retq
1963 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 4, i32 5, i32 6, i32 undef, i32 undef, i32 undef, i32 undef>
1964 ret <8 x i16> %shuffle
1965 }
1967 define <8 x i16> @shuffle_v8i16_3456789a(<8 x i16> %a, <8 x i16> %b) {
1968 ; SSE2-LABEL: shuffle_v8i16_3456789a:
1969 ; SSE2: # %bb.0:
1970 ; SSE2-NEXT: psrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
1971 ; SSE2-NEXT: pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1,2,3,4,5]
1972 ; SSE2-NEXT: por %xmm1, %xmm0
1973 ; SSE2-NEXT: retq
1974 ;
1975 ; SSSE3-LABEL: shuffle_v8i16_3456789a:
1976 ; SSSE3: # %bb.0:
1977 ; SSSE3-NEXT: palignr {{.*#+}} xmm1 = xmm0[6,7,8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5]
1978 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
1979 ; SSSE3-NEXT: retq
1980 ;
1981 ; SSE41-LABEL: shuffle_v8i16_3456789a:
1982 ; SSE41: # %bb.0:
1983 ; SSE41-NEXT: palignr {{.*#+}} xmm1 = xmm0[6,7,8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5]
1984 ; SSE41-NEXT: movdqa %xmm1, %xmm0
1985 ; SSE41-NEXT: retq
1986 ;
1987 ; AVX-LABEL: shuffle_v8i16_3456789a:
1988 ; AVX: # %bb.0:
1989 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5]
1990 ; AVX-NEXT: retq
1991 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10>
1992 ret <8 x i16> %shuffle
1993 }
1995 define <8 x i16> @shuffle_v8i16_u456uu9u(<8 x i16> %a, <8 x i16> %b) {
1996 ; SSE2-LABEL: shuffle_v8i16_u456uu9u:
1997 ; SSE2: # %bb.0:
1998 ; SSE2-NEXT: psrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
1999 ; SSE2-NEXT: pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1,2,3,4,5]
2000 ; SSE2-NEXT: por %xmm1, %xmm0
2001 ; SSE2-NEXT: retq
2002 ;
2003 ; SSSE3-LABEL: shuffle_v8i16_u456uu9u:
2004 ; SSSE3: # %bb.0:
2005 ; SSSE3-NEXT: palignr {{.*#+}} xmm1 = xmm0[6,7,8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5]
2006 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
2007 ; SSSE3-NEXT: retq
2008 ;
2009 ; SSE41-LABEL: shuffle_v8i16_u456uu9u:
2010 ; SSE41: # %bb.0:
2011 ; SSE41-NEXT: palignr {{.*#+}} xmm1 = xmm0[6,7,8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5]
2012 ; SSE41-NEXT: movdqa %xmm1, %xmm0
2013 ; SSE41-NEXT: retq
2014 ;
2015 ; AVX-LABEL: shuffle_v8i16_u456uu9u:
2016 ; AVX: # %bb.0:
2017 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5]
2018 ; AVX-NEXT: retq
2019 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 4, i32 5, i32 6, i32 undef, i32 undef, i32 9, i32 undef>
2020 ret <8 x i16> %shuffle
2021 }
2023 define <8 x i16> @shuffle_v8i16_56789abc(<8 x i16> %a, <8 x i16> %b) {
2024 ; SSE2-LABEL: shuffle_v8i16_56789abc:
2025 ; SSE2: # %bb.0:
2026 ; SSE2-NEXT: psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2027 ; SSE2-NEXT: pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,xmm1[0,1,2,3,4,5,6,7,8,9]
2028 ; SSE2-NEXT: por %xmm1, %xmm0
2029 ; SSE2-NEXT: retq
2030 ;
2031 ; SSSE3-LABEL: shuffle_v8i16_56789abc:
2032 ; SSSE3: # %bb.0:
2033 ; SSSE3-NEXT: palignr {{.*#+}} xmm1 = xmm0[10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7,8,9]
2034 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
2035 ; SSSE3-NEXT: retq
2036 ;
2037 ; SSE41-LABEL: shuffle_v8i16_56789abc:
2038 ; SSE41: # %bb.0:
2039 ; SSE41-NEXT: palignr {{.*#+}} xmm1 = xmm0[10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7,8,9]
2040 ; SSE41-NEXT: movdqa %xmm1, %xmm0
2041 ; SSE41-NEXT: retq
2042 ;
2043 ; AVX-LABEL: shuffle_v8i16_56789abc:
2044 ; AVX: # %bb.0:
2045 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7,8,9]
2046 ; AVX-NEXT: retq
2047 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12>
2048 ret <8 x i16> %shuffle
2049 }
2051 define <8 x i16> @shuffle_v8i16_u6uu9abu(<8 x i16> %a, <8 x i16> %b) {
2052 ; SSE2-LABEL: shuffle_v8i16_u6uu9abu:
2053 ; SSE2: # %bb.0:
2054 ; SSE2-NEXT: psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2055 ; SSE2-NEXT: pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,xmm1[0,1,2,3,4,5,6,7,8,9]
2056 ; SSE2-NEXT: por %xmm1, %xmm0
2057 ; SSE2-NEXT: retq
2058 ;
2059 ; SSSE3-LABEL: shuffle_v8i16_u6uu9abu:
2060 ; SSSE3: # %bb.0:
2061 ; SSSE3-NEXT: palignr {{.*#+}} xmm1 = xmm0[10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7,8,9]
2062 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
2063 ; SSSE3-NEXT: retq
2064 ;
2065 ; SSE41-LABEL: shuffle_v8i16_u6uu9abu:
2066 ; SSE41: # %bb.0:
2067 ; SSE41-NEXT: palignr {{.*#+}} xmm1 = xmm0[10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7,8,9]
2068 ; SSE41-NEXT: movdqa %xmm1, %xmm0
2069 ; SSE41-NEXT: retq
2070 ;
2071 ; AVX-LABEL: shuffle_v8i16_u6uu9abu:
2072 ; AVX: # %bb.0:
2073 ; AVX-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7,8,9]
2074 ; AVX-NEXT: retq
2075 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 undef, i32 6, i32 undef, i32 undef, i32 9, i32 10, i32 11, i32 undef>
2076 ret <8 x i16> %shuffle
2077 }
2079 define <8 x i16> @shuffle_v8i16_0uuu1uuu(<8 x i16> %a) {
2080 ; SSE2-LABEL: shuffle_v8i16_0uuu1uuu:
2081 ; SSE2: # %bb.0:
2082 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
2083 ; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,6,7]
2084 ; SSE2-NEXT: retq
2085 ;
2086 ; SSSE3-LABEL: shuffle_v8i16_0uuu1uuu:
2087 ; SSSE3: # %bb.0:
2088 ; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
2089 ; SSSE3-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,6,7]
2090 ; SSSE3-NEXT: retq
2091 ;
2092 ; SSE41-LABEL: shuffle_v8i16_0uuu1uuu:
2093 ; SSE41: # %bb.0:
2094 ; SSE41-NEXT: pmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
2095 ; SSE41-NEXT: retq
2096 ;
2097 ; AVX-LABEL: shuffle_v8i16_0uuu1uuu:
2098 ; AVX: # %bb.0:
2099 ; AVX-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
2100 ; AVX-NEXT: retq
2101 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 undef, i32 undef, i32 undef, i32 1, i32 undef, i32 undef, i32 undef>
2102 ret <8 x i16> %shuffle
2103 }
2105 define <8 x i16> @shuffle_v8i16_0zzz1zzz(<8 x i16> %a) {
2106 ; SSE2-LABEL: shuffle_v8i16_0zzz1zzz:
2107 ; SSE2: # %bb.0:
2108 ; SSE2-NEXT: pxor %xmm1, %xmm1
2109 ; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
2110 ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
2111 ; SSE2-NEXT: retq
2112 ;
2113 ; SSSE3-LABEL: shuffle_v8i16_0zzz1zzz:
2114 ; SSSE3: # %bb.0:
2115 ; SSSE3-NEXT: pxor %xmm1, %xmm1
2116 ; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
2117 ; SSSE3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
2118 ; SSSE3-NEXT: retq
2119 ;
2120 ; SSE41-LABEL: shuffle_v8i16_0zzz1zzz:
2121 ; SSE41: # %bb.0:
2122 ; SSE41-NEXT: pmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
2123 ; SSE41-NEXT: retq
2124 ;
2125 ; AVX-LABEL: shuffle_v8i16_0zzz1zzz:
2126 ; AVX: # %bb.0:
2127 ; AVX-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
2128 ; AVX-NEXT: retq
2129 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 9, i32 10, i32 11, i32 1, i32 13, i32 14, i32 15>
2130 ret <8 x i16> %shuffle
2131 }
2133 define <8 x i16> @shuffle_v8i16_0u1u2u3u(<8 x i16> %a) {
2134 ; SSE2-LABEL: shuffle_v8i16_0u1u2u3u:
2135 ; SSE2: # %bb.0:
2136 ; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
2137 ; SSE2-NEXT: retq
2138 ;
2139 ; SSSE3-LABEL: shuffle_v8i16_0u1u2u3u:
2140 ; SSSE3: # %bb.0:
2141 ; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
2142 ; SSSE3-NEXT: retq
2143 ;
2144 ; SSE41-LABEL: shuffle_v8i16_0u1u2u3u:
2145 ; SSE41: # %bb.0:
2146 ; SSE41-NEXT: pmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
2147 ; SSE41-NEXT: retq
2148 ;
2149 ; AVX-LABEL: shuffle_v8i16_0u1u2u3u:
2150 ; AVX: # %bb.0:
2151 ; AVX-NEXT: vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
2152 ; AVX-NEXT: retq
2153 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 undef, i32 1, i32 undef, i32 2, i32 undef, i32 3, i32 undef>
2154 ret <8 x i16> %shuffle
2155 }
2157 define <8 x i16> @shuffle_v8i16_0z1z2z3z(<8 x i16> %a) {
2158 ; SSE2-LABEL: shuffle_v8i16_0z1z2z3z:
2159 ; SSE2: # %bb.0:
2160 ; SSE2-NEXT: pxor %xmm1, %xmm1
2161 ; SSE2-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
2162 ; SSE2-NEXT: retq
2163 ;
2164 ; SSSE3-LABEL: shuffle_v8i16_0z1z2z3z:
2165 ; SSSE3: # %bb.0:
2166 ; SSSE3-NEXT: pxor %xmm1, %xmm1
2167 ; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
2168 ; SSSE3-NEXT: retq
2169 ;
2170 ; SSE41-LABEL: shuffle_v8i16_0z1z2z3z:
2171 ; SSE41: # %bb.0:
2172 ; SSE41-NEXT: pmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
2173 ; SSE41-NEXT: retq
2174 ;
2175 ; AVX-LABEL: shuffle_v8i16_0z1z2z3z:
2176 ; AVX: # %bb.0:
2177 ; AVX-NEXT: vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
2178 ; AVX-NEXT: retq
2179 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 9, i32 1, i32 11, i32 2, i32 13, i32 3, i32 15>
2180 ret <8 x i16> %shuffle
2181 }
2183 define <8 x i16> @shuffle_v8i16_01100110(<8 x i16> %a) {
2184 ; SSE-LABEL: shuffle_v8i16_01100110:
2185 ; SSE: # %bb.0:
2186 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,1,1,0,4,5,6,7]
2187 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
2188 ; SSE-NEXT: retq
2189 ;
2190 ; AVX1-LABEL: shuffle_v8i16_01100110:
2191 ; AVX1: # %bb.0:
2192 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,1,1,0,4,5,6,7]
2193 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
2194 ; AVX1-NEXT: retq
2195 ;
2196 ; AVX2-SLOW-LABEL: shuffle_v8i16_01100110:
2197 ; AVX2-SLOW: # %bb.0:
2198 ; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,1,1,0,4,5,6,7]
2199 ; AVX2-SLOW-NEXT: vpbroadcastq %xmm0, %xmm0
2200 ; AVX2-SLOW-NEXT: retq
2201 ;
2202 ; AVX2-FAST-LABEL: shuffle_v8i16_01100110:
2203 ; AVX2-FAST: # %bb.0:
2204 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,2,3,0,1,0,1,2,3,2,3,0,1]
2205 ; AVX2-FAST-NEXT: retq
2206 ;
2207 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_01100110:
2208 ; AVX512VL-SLOW: # %bb.0:
2209 ; AVX512VL-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,1,1,0,4,5,6,7]
2210 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
2211 ; AVX512VL-SLOW-NEXT: retq
2212 ;
2213 ; AVX512VL-FAST-LABEL: shuffle_v8i16_01100110:
2214 ; AVX512VL-FAST: # %bb.0:
2215 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,2,3,0,1,0,1,2,3,2,3,0,1]
2216 ; AVX512VL-FAST-NEXT: retq
2217 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 1, i32 1, i32 0, i32 0, i32 1, i32 1, i32 0>
2218 ret <8 x i16> %shuffle
2219 }
2221 define <8 x i16> @shuffle_v8i16_01u0u110(<8 x i16> %a) {
2222 ; SSE-LABEL: shuffle_v8i16_01u0u110:
2223 ; SSE: # %bb.0:
2224 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,1,1,0,4,5,6,7]
2225 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
2226 ; SSE-NEXT: retq
2227 ;
2228 ; AVX1-LABEL: shuffle_v8i16_01u0u110:
2229 ; AVX1: # %bb.0:
2230 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,1,1,0,4,5,6,7]
2231 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
2232 ; AVX1-NEXT: retq
2233 ;
2234 ; AVX2-SLOW-LABEL: shuffle_v8i16_01u0u110:
2235 ; AVX2-SLOW: # %bb.0:
2236 ; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,1,1,0,4,5,6,7]
2237 ; AVX2-SLOW-NEXT: vpbroadcastq %xmm0, %xmm0
2238 ; AVX2-SLOW-NEXT: retq
2239 ;
2240 ; AVX2-FAST-LABEL: shuffle_v8i16_01u0u110:
2241 ; AVX2-FAST: # %bb.0:
2242 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,2,3,0,1,0,1,2,3,2,3,0,1]
2243 ; AVX2-FAST-NEXT: retq
2244 ;
2245 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_01u0u110:
2246 ; AVX512VL-SLOW: # %bb.0:
2247 ; AVX512VL-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,1,1,0,4,5,6,7]
2248 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
2249 ; AVX512VL-SLOW-NEXT: retq
2250 ;
2251 ; AVX512VL-FAST-LABEL: shuffle_v8i16_01u0u110:
2252 ; AVX512VL-FAST: # %bb.0:
2253 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,2,3,0,1,0,1,2,3,2,3,0,1]
2254 ; AVX512VL-FAST-NEXT: retq
2255 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 1, i32 undef, i32 0, i32 undef, i32 1, i32 1, i32 0>
2256 ret <8 x i16> %shuffle
2257 }
2259 define <8 x i16> @shuffle_v8i16_467uu675(<8 x i16> %a) {
2260 ; SSE-LABEL: shuffle_v8i16_467uu675:
2261 ; SSE: # %bb.0:
2262 ; SSE-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,7,5]
2263 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
2264 ; SSE-NEXT: retq
2265 ;
2266 ; AVX1-LABEL: shuffle_v8i16_467uu675:
2267 ; AVX1: # %bb.0:
2268 ; AVX1-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,7,5]
2269 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
2270 ; AVX1-NEXT: retq
2271 ;
2272 ; AVX2-SLOW-LABEL: shuffle_v8i16_467uu675:
2273 ; AVX2-SLOW: # %bb.0:
2274 ; AVX2-SLOW-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,7,5]
2275 ; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
2276 ; AVX2-SLOW-NEXT: retq
2277 ;
2278 ; AVX2-FAST-LABEL: shuffle_v8i16_467uu675:
2279 ; AVX2-FAST: # %bb.0:
2280 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,12,13,14,15,10,11,8,9,12,13,14,15,10,11]
2281 ; AVX2-FAST-NEXT: retq
2282 ;
2283 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_467uu675:
2284 ; AVX512VL-SLOW: # %bb.0:
2285 ; AVX512VL-SLOW-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,7,5]
2286 ; AVX512VL-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
2287 ; AVX512VL-SLOW-NEXT: retq
2288 ;
2289 ; AVX512VL-FAST-LABEL: shuffle_v8i16_467uu675:
2290 ; AVX512VL-FAST: # %bb.0:
2291 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[8,9,12,13,14,15,10,11,8,9,12,13,14,15,10,11]
2292 ; AVX512VL-FAST-NEXT: retq
2293 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 4, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7, i32 5>
2294 ret <8 x i16> %shuffle
2295 }
2297 ;
2298 ; Shuffle to logical bit shifts
2299 ;
2300 define <8 x i16> @shuffle_v8i16_z0z2z4z6(<8 x i16> %a) {
2301 ; SSE-LABEL: shuffle_v8i16_z0z2z4z6:
2302 ; SSE: # %bb.0:
2303 ; SSE-NEXT: pslld $16, %xmm0
2304 ; SSE-NEXT: retq
2305 ;
2306 ; AVX-LABEL: shuffle_v8i16_z0z2z4z6:
2307 ; AVX: # %bb.0:
2308 ; AVX-NEXT: vpslld $16, %xmm0, %xmm0
2309 ; AVX-NEXT: retq
2310 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 8, i32 0, i32 8, i32 2, i32 8, i32 4, i32 8, i32 6>
2311 ret <8 x i16> %shuffle
2312 }
2314 define <8 x i16> @shuffle_v8i16_zzz0zzz4(<8 x i16> %a) {
2315 ; SSE-LABEL: shuffle_v8i16_zzz0zzz4:
2316 ; SSE: # %bb.0:
2317 ; SSE-NEXT: psllq $48, %xmm0
2318 ; SSE-NEXT: retq
2319 ;
2320 ; AVX-LABEL: shuffle_v8i16_zzz0zzz4:
2321 ; AVX: # %bb.0:
2322 ; AVX-NEXT: vpsllq $48, %xmm0, %xmm0
2323 ; AVX-NEXT: retq
2324 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 8, i32 8, i32 8, i32 0, i32 8, i32 8, i32 8, i32 4>
2325 ret <8 x i16> %shuffle
2326 }
2328 define <8 x i16> @shuffle_v8i16_zz01zX4X(<8 x i16> %a) {
2329 ; SSE-LABEL: shuffle_v8i16_zz01zX4X:
2330 ; SSE: # %bb.0:
2331 ; SSE-NEXT: psllq $32, %xmm0
2332 ; SSE-NEXT: retq
2333 ;
2334 ; AVX-LABEL: shuffle_v8i16_zz01zX4X:
2335 ; AVX: # %bb.0:
2336 ; AVX-NEXT: vpsllq $32, %xmm0, %xmm0
2337 ; AVX-NEXT: retq
2338 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 8, i32 8, i32 0, i32 1, i32 8, i32 undef, i32 4, i32 undef>
2339 ret <8 x i16> %shuffle
2340 }
2342 define <8 x i16> @shuffle_v8i16_z0X2z456(<8 x i16> %a) {
2343 ; SSE-LABEL: shuffle_v8i16_z0X2z456:
2344 ; SSE: # %bb.0:
2345 ; SSE-NEXT: psllq $16, %xmm0
2346 ; SSE-NEXT: retq
2347 ;
2348 ; AVX-LABEL: shuffle_v8i16_z0X2z456:
2349 ; AVX: # %bb.0:
2350 ; AVX-NEXT: vpsllq $16, %xmm0, %xmm0
2351 ; AVX-NEXT: retq
2352 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 8, i32 0, i32 undef, i32 2, i32 8, i32 4, i32 5, i32 6>
2353 ret <8 x i16> %shuffle
2354 }
2356 define <8 x i16> @shuffle_v8i16_1z3zXz7z(<8 x i16> %a) {
2357 ; SSE-LABEL: shuffle_v8i16_1z3zXz7z:
2358 ; SSE: # %bb.0:
2359 ; SSE-NEXT: psrld $16, %xmm0
2360 ; SSE-NEXT: retq
2361 ;
2362 ; AVX-LABEL: shuffle_v8i16_1z3zXz7z:
2363 ; AVX: # %bb.0:
2364 ; AVX-NEXT: vpsrld $16, %xmm0, %xmm0
2365 ; AVX-NEXT: retq
2366 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 1, i32 8, i32 3, i32 8, i32 undef, i32 8, i32 7, i32 8>
2367 ret <8 x i16> %shuffle
2368 }
2370 define <8 x i16> @shuffle_v8i16_1X3z567z(<8 x i16> %a) {
2371 ; SSE-LABEL: shuffle_v8i16_1X3z567z:
2372 ; SSE: # %bb.0:
2373 ; SSE-NEXT: psrlq $16, %xmm0
2374 ; SSE-NEXT: retq
2375 ;
2376 ; AVX-LABEL: shuffle_v8i16_1X3z567z:
2377 ; AVX: # %bb.0:
2378 ; AVX-NEXT: vpsrlq $16, %xmm0, %xmm0
2379 ; AVX-NEXT: retq
2380 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 1, i32 undef, i32 3, i32 8, i32 5, i32 6, i32 7, i32 8>
2381 ret <8 x i16> %shuffle
2382 }
2384 define <8 x i16> @shuffle_v8i16_23zz67zz(<8 x i16> %a) {
2385 ; SSE-LABEL: shuffle_v8i16_23zz67zz:
2386 ; SSE: # %bb.0:
2387 ; SSE-NEXT: psrlq $32, %xmm0
2388 ; SSE-NEXT: retq
2389 ;
2390 ; AVX-LABEL: shuffle_v8i16_23zz67zz:
2391 ; AVX: # %bb.0:
2392 ; AVX-NEXT: vpsrlq $32, %xmm0, %xmm0
2393 ; AVX-NEXT: retq
2394 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 2, i32 3, i32 8, i32 8, i32 6, i32 7, i32 8, i32 8>
2395 ret <8 x i16> %shuffle
2396 }
2398 define <8 x i16> @shuffle_v8i16_3zXXXzzz(<8 x i16> %a) {
2399 ; SSE-LABEL: shuffle_v8i16_3zXXXzzz:
2400 ; SSE: # %bb.0:
2401 ; SSE-NEXT: psrlq $48, %xmm0
2402 ; SSE-NEXT: retq
2403 ;
2404 ; AVX-LABEL: shuffle_v8i16_3zXXXzzz:
2405 ; AVX: # %bb.0:
2406 ; AVX-NEXT: vpsrlq $48, %xmm0, %xmm0
2407 ; AVX-NEXT: retq
2408 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 3, i32 8, i32 undef, i32 undef, i32 undef, i32 8, i32 8, i32 8>
2409 ret <8 x i16> %shuffle
2410 }
2412 define <8 x i16> @shuffle_v8i16_01u3zzuz(<8 x i16> %a) {
2413 ; SSE-LABEL: shuffle_v8i16_01u3zzuz:
2414 ; SSE: # %bb.0:
2415 ; SSE-NEXT: movq {{.*#+}} xmm0 = xmm0[0],zero
2416 ; SSE-NEXT: retq
2417 ;
2418 ; AVX-LABEL: shuffle_v8i16_01u3zzuz:
2419 ; AVX: # %bb.0:
2420 ; AVX-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero
2421 ; AVX-NEXT: retq
2422 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 1, i32 undef, i32 3, i32 8, i32 8, i32 undef, i32 8>
2423 ret <8 x i16> %shuffle
2424 }
2426 define <8 x i16> @shuffle_v8i16_0z234567(<8 x i16> %a) {
2427 ; SSE2-LABEL: shuffle_v8i16_0z234567:
2428 ; SSE2: # %bb.0:
2429 ; SSE2-NEXT: andps {{.*}}(%rip), %xmm0
2430 ; SSE2-NEXT: retq
2431 ;
2432 ; SSSE3-LABEL: shuffle_v8i16_0z234567:
2433 ; SSSE3: # %bb.0:
2434 ; SSSE3-NEXT: andps {{.*}}(%rip), %xmm0
2435 ; SSSE3-NEXT: retq
2436 ;
2437 ; SSE41-LABEL: shuffle_v8i16_0z234567:
2438 ; SSE41: # %bb.0:
2439 ; SSE41-NEXT: pxor %xmm1, %xmm1
2440 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6,7]
2441 ; SSE41-NEXT: retq
2442 ;
2443 ; AVX-LABEL: shuffle_v8i16_0z234567:
2444 ; AVX: # %bb.0:
2445 ; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
2446 ; AVX-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6,7]
2447 ; AVX-NEXT: retq
2448 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
2449 ret <8 x i16> %shuffle
2450 }
2452 define <8 x i16> @shuffle_v8i16_0zzzz5z7(<8 x i16> %a) {
2453 ; SSE2-LABEL: shuffle_v8i16_0zzzz5z7:
2454 ; SSE2: # %bb.0:
2455 ; SSE2-NEXT: andps {{.*}}(%rip), %xmm0
2456 ; SSE2-NEXT: retq
2457 ;
2458 ; SSSE3-LABEL: shuffle_v8i16_0zzzz5z7:
2459 ; SSSE3: # %bb.0:
2460 ; SSSE3-NEXT: andps {{.*}}(%rip), %xmm0
2461 ; SSSE3-NEXT: retq
2462 ;
2463 ; SSE41-LABEL: shuffle_v8i16_0zzzz5z7:
2464 ; SSE41: # %bb.0:
2465 ; SSE41-NEXT: pxor %xmm1, %xmm1
2466 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3,4],xmm0[5],xmm1[6],xmm0[7]
2467 ; SSE41-NEXT: retq
2468 ;
2469 ; AVX-LABEL: shuffle_v8i16_0zzzz5z7:
2470 ; AVX: # %bb.0:
2471 ; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
2472 ; AVX-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3,4],xmm0[5],xmm1[6],xmm0[7]
2473 ; AVX-NEXT: retq
2474 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 8, i32 8, i32 8, i32 8, i32 5, i32 8, i32 7>
2475 ret <8 x i16> %shuffle
2476 }
2478 define <8 x i16> @shuffle_v8i16_0123456z(<8 x i16> %a) {
2479 ; SSE2-LABEL: shuffle_v8i16_0123456z:
2480 ; SSE2: # %bb.0:
2481 ; SSE2-NEXT: andps {{.*}}(%rip), %xmm0
2482 ; SSE2-NEXT: retq
2483 ;
2484 ; SSSE3-LABEL: shuffle_v8i16_0123456z:
2485 ; SSSE3: # %bb.0:
2486 ; SSSE3-NEXT: andps {{.*}}(%rip), %xmm0
2487 ; SSSE3-NEXT: retq
2488 ;
2489 ; SSE41-LABEL: shuffle_v8i16_0123456z:
2490 ; SSE41: # %bb.0:
2491 ; SSE41-NEXT: pxor %xmm1, %xmm1
2492 ; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6],xmm1[7]
2493 ; SSE41-NEXT: retq
2494 ;
2495 ; AVX-LABEL: shuffle_v8i16_0123456z:
2496 ; AVX: # %bb.0:
2497 ; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
2498 ; AVX-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6],xmm1[7]
2499 ; AVX-NEXT: retq
2500 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 15>
2501 ret <8 x i16> %shuffle
2502 }
2504 define <8 x i16> @shuffle_v8i16_fu3ucc5u(<8 x i16> %a, <8 x i16> %b) {
2505 ; SSE-LABEL: shuffle_v8i16_fu3ucc5u:
2506 ; SSE: # %bb.0:
2507 ; SSE-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
2508 ; SSE-NEXT: pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,7,5,4,4]
2509 ; SSE-NEXT: punpckhdq {{.*#+}} xmm1 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
2510 ; SSE-NEXT: movdqa %xmm1, %xmm0
2511 ; SSE-NEXT: retq
2512 ;
2513 ; AVX1OR2-LABEL: shuffle_v8i16_fu3ucc5u:
2514 ; AVX1OR2: # %bb.0:
2515 ; AVX1OR2-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
2516 ; AVX1OR2-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,7,5,4,4]
2517 ; AVX1OR2-NEXT: vpunpckhdq {{.*#+}} xmm0 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
2518 ; AVX1OR2-NEXT: retq
2519 ;
2520 ; AVX512VL-LABEL: shuffle_v8i16_fu3ucc5u:
2521 ; AVX512VL: # %bb.0:
2522 ; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [7,5,11,12,4,4,13,14]
2523 ; AVX512VL-NEXT: vpermi2w %xmm0, %xmm1, %xmm2
2524 ; AVX512VL-NEXT: vmovdqa %xmm2, %xmm0
2525 ; AVX512VL-NEXT: retq
2526 %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 15, i32 undef, i32 3, i32 undef, i32 12, i32 12, i32 5, i32 undef>
2527 ret <8 x i16> %shuffle
2528 }
2530 define <8 x i16> @shuffle_v8i16_8012345u(<8 x i16> %a) {
2531 ; SSE-LABEL: shuffle_v8i16_8012345u:
2532 ; SSE: # %bb.0:
2533 ; SSE-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
2534 ; SSE-NEXT: retq
2535 ;
2536 ; AVX-LABEL: shuffle_v8i16_8012345u:
2537 ; AVX: # %bb.0:
2538 ; AVX-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
2539 ; AVX-NEXT: retq
2540 %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 8, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 undef>
2542 ret <8 x i16> %shuffle
2543 }
2545 ; PR40306
2546 define <8 x i16> @shuffle_v8i16_9zzzuuuu(<8 x i16> %x) {
2547 ; SSE-LABEL: shuffle_v8i16_9zzzuuuu:
2548 ; SSE: # %bb.0:
2549 ; SSE-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3]
2550 ; SSE-NEXT: psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2551 ; SSE-NEXT: retq
2552 ;
2553 ; AVX1-LABEL: shuffle_v8i16_9zzzuuuu:
2554 ; AVX1: # %bb.0:
2555 ; AVX1-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3]
2556 ; AVX1-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2557 ; AVX1-NEXT: retq
2558 ;
2559 ; AVX2-SLOW-LABEL: shuffle_v8i16_9zzzuuuu:
2560 ; AVX2-SLOW: # %bb.0:
2561 ; AVX2-SLOW-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3]
2562 ; AVX2-SLOW-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2563 ; AVX2-SLOW-NEXT: retq
2564 ;
2565 ; AVX2-FAST-LABEL: shuffle_v8i16_9zzzuuuu:
2566 ; AVX2-FAST: # %bb.0:
2567 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2568 ; AVX2-FAST-NEXT: retq
2569 ;
2570 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_9zzzuuuu:
2571 ; AVX512VL-SLOW: # %bb.0:
2572 ; AVX512VL-SLOW-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3]
2573 ; AVX512VL-SLOW-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2574 ; AVX512VL-SLOW-NEXT: retq
2575 ;
2576 ; AVX512VL-FAST-LABEL: shuffle_v8i16_9zzzuuuu:
2577 ; AVX512VL-FAST: # %bb.0:
2578 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2579 ; AVX512VL-FAST-NEXT: retq
2580 %r = shufflevector <8 x i16> zeroinitializer, <8 x i16> %x, <8 x i32> <i32 9, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
2581 ret <8 x i16> %r
2582 }
2584 ; PR40318
2585 define <8 x i16> @shuffle_v8i16_2zzzuuuu(<8 x i16> %x) {
2586 ; SSE-LABEL: shuffle_v8i16_2zzzuuuu:
2587 ; SSE: # %bb.0:
2588 ; SSE-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
2589 ; SSE-NEXT: psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2590 ; SSE-NEXT: retq
2591 ;
2592 ; AVX1-LABEL: shuffle_v8i16_2zzzuuuu:
2593 ; AVX1: # %bb.0:
2594 ; AVX1-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
2595 ; AVX1-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2596 ; AVX1-NEXT: retq
2597 ;
2598 ; AVX2-SLOW-LABEL: shuffle_v8i16_2zzzuuuu:
2599 ; AVX2-SLOW: # %bb.0:
2600 ; AVX2-SLOW-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
2601 ; AVX2-SLOW-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2602 ; AVX2-SLOW-NEXT: retq
2603 ;
2604 ; AVX2-FAST-LABEL: shuffle_v8i16_2zzzuuuu:
2605 ; AVX2-FAST: # %bb.0:
2606 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2607 ; AVX2-FAST-NEXT: retq
2608 ;
2609 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_2zzzuuuu:
2610 ; AVX512VL-SLOW: # %bb.0:
2611 ; AVX512VL-SLOW-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
2612 ; AVX512VL-SLOW-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2613 ; AVX512VL-SLOW-NEXT: retq
2614 ;
2615 ; AVX512VL-FAST-LABEL: shuffle_v8i16_2zzzuuuu:
2616 ; AVX512VL-FAST: # %bb.0:
2617 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[4,5],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
2618 ; AVX512VL-FAST-NEXT: retq
2619 %r = shufflevector <8 x i16> %x, <8 x i16> zeroinitializer, <8 x i32> <i32 2, i32 9, i32 10, i32 11, i32 undef, i32 undef, i32 undef, i32 undef>
2620 ret <8 x i16> %r
2621 }
2623 define <8 x i16> @shuffle_v8i16_3uu6zzzz(<8 x i16> %x) {
2624 ; SSE-LABEL: shuffle_v8i16_3uu6zzzz:
2625 ; SSE: # %bb.0:
2626 ; SSE-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
2627 ; SSE-NEXT: psrldq {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero
2628 ; SSE-NEXT: retq
2629 ;
2630 ; AVX1-LABEL: shuffle_v8i16_3uu6zzzz:
2631 ; AVX1: # %bb.0:
2632 ; AVX1-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
2633 ; AVX1-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero
2634 ; AVX1-NEXT: retq
2635 ;
2636 ; AVX2-SLOW-LABEL: shuffle_v8i16_3uu6zzzz:
2637 ; AVX2-SLOW: # %bb.0:
2638 ; AVX2-SLOW-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
2639 ; AVX2-SLOW-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero
2640 ; AVX2-SLOW-NEXT: retq
2641 ;
2642 ; AVX2-FAST-LABEL: shuffle_v8i16_3uu6zzzz:
2643 ; AVX2-FAST: # %bb.0:
2644 ; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13],zero,zero,zero,zero,zero,zero,zero,zero
2645 ; AVX2-FAST-NEXT: retq
2646 ;
2647 ; AVX512VL-SLOW-LABEL: shuffle_v8i16_3uu6zzzz:
2648 ; AVX512VL-SLOW: # %bb.0:
2649 ; AVX512VL-SLOW-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
2650 ; AVX512VL-SLOW-NEXT: vpsrldq {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero
2651 ; AVX512VL-SLOW-NEXT: retq
2652 ;
2653 ; AVX512VL-FAST-LABEL: shuffle_v8i16_3uu6zzzz:
2654 ; AVX512VL-FAST: # %bb.0:
2655 ; AVX512VL-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13],zero,zero,zero,zero,zero,zero,zero,zero
2656 ; AVX512VL-FAST-NEXT: retq
2657 %r = shufflevector <8 x i16> %x, <8 x i16> zeroinitializer, <8 x i32> <i32 3, i32 undef, i32 undef, i32 6, i32 8, i32 8, i32 8, i32 8>
2658 ret <8 x i16> %r
2659 }
2661 define <8 x i16> @mask_v8i16_012345ef(<8 x i16> %a, <8 x i16> %b) {
2662 ; SSE2-LABEL: mask_v8i16_012345ef:
2663 ; SSE2: # %bb.0:
2664 ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[3,0],xmm1[2,0]
2665 ; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[2,0]
2666 ; SSE2-NEXT: movaps %xmm1, %xmm0
2667 ; SSE2-NEXT: retq
2668 ;
2669 ; SSSE3-LABEL: mask_v8i16_012345ef:
2670 ; SSSE3: # %bb.0:
2671 ; SSSE3-NEXT: shufps {{.*#+}} xmm0 = xmm0[3,0],xmm1[2,0]
2672 ; SSSE3-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[2,0]
2673 ; SSSE3-NEXT: movaps %xmm1, %xmm0
2674 ; SSSE3-NEXT: retq
2675 ;
2676 ; SSE41-LABEL: mask_v8i16_012345ef:
2677 ; SSE41: # %bb.0:
2678 ; SSE41-NEXT: blendps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[3]
2679 ; SSE41-NEXT: retq
2680 ;
2681 ; AVX-LABEL: mask_v8i16_012345ef:
2682 ; AVX: # %bb.0:
2683 ; AVX-NEXT: vblendps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[3]
2684 ; AVX-NEXT: retq
2685 %1 = bitcast <8 x i16> %a to <2 x i64>
2686 %2 = bitcast <8 x i16> %b to <2 x i64>
2687 %3 = and <2 x i64> %1, <i64 0, i64 -4294967296>
2688 %4 = and <2 x i64> %2, <i64 -1, i64 4294967295>
2689 %5 = or <2 x i64> %4, %3
2690 %6 = bitcast <2 x i64> %5 to <8 x i16>
2691 ret <8 x i16> %6
2692 }
2694 define <8 x i16> @insert_dup_mem_v8i16_i32(i32* %ptr) {
2695 ; SSE-LABEL: insert_dup_mem_v8i16_i32:
2696 ; SSE: # %bb.0:
2697 ; SSE-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
2698 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,2,3,4,5,6,7]
2699 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2700 ; SSE-NEXT: retq
2701 ;
2702 ; AVX1-LABEL: insert_dup_mem_v8i16_i32:
2703 ; AVX1: # %bb.0:
2704 ; AVX1-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
2705 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,3,4,5,6,7]
2706 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2707 ; AVX1-NEXT: retq
2708 ;
2709 ; AVX2OR512VL-LABEL: insert_dup_mem_v8i16_i32:
2710 ; AVX2OR512VL: # %bb.0:
2711 ; AVX2OR512VL-NEXT: vpbroadcastw (%rdi), %xmm0
2712 ; AVX2OR512VL-NEXT: retq
2713 %tmp = load i32, i32* %ptr, align 4
2714 %tmp1 = insertelement <4 x i32> zeroinitializer, i32 %tmp, i32 0
2715 %tmp2 = bitcast <4 x i32> %tmp1 to <8 x i16>
2716 %tmp3 = shufflevector <8 x i16> %tmp2, <8 x i16> undef, <8 x i32> zeroinitializer
2717 ret <8 x i16> %tmp3
2718 }
2720 define <8 x i16> @insert_dup_mem_v8i16_sext_i16(i16* %ptr) {
2721 ; SSE-LABEL: insert_dup_mem_v8i16_sext_i16:
2722 ; SSE: # %bb.0:
2723 ; SSE-NEXT: movzwl (%rdi), %eax
2724 ; SSE-NEXT: movd %eax, %xmm0
2725 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,2,3,4,5,6,7]
2726 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2727 ; SSE-NEXT: retq
2728 ;
2729 ; AVX1-LABEL: insert_dup_mem_v8i16_sext_i16:
2730 ; AVX1: # %bb.0:
2731 ; AVX1-NEXT: movzwl (%rdi), %eax
2732 ; AVX1-NEXT: vmovd %eax, %xmm0
2733 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,3,4,5,6,7]
2734 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2735 ; AVX1-NEXT: retq
2736 ;
2737 ; AVX2OR512VL-LABEL: insert_dup_mem_v8i16_sext_i16:
2738 ; AVX2OR512VL: # %bb.0:
2739 ; AVX2OR512VL-NEXT: vpbroadcastw (%rdi), %xmm0
2740 ; AVX2OR512VL-NEXT: retq
2741 %tmp = load i16, i16* %ptr, align 2
2742 %tmp1 = sext i16 %tmp to i32
2743 %tmp2 = insertelement <4 x i32> zeroinitializer, i32 %tmp1, i32 0
2744 %tmp3 = bitcast <4 x i32> %tmp2 to <8 x i16>
2745 %tmp4 = shufflevector <8 x i16> %tmp3, <8 x i16> undef, <8 x i32> zeroinitializer
2746 ret <8 x i16> %tmp4
2747 }
2749 define <8 x i16> @insert_dup_elt1_mem_v8i16_i32(i32* %ptr) {
2750 ; SSE-LABEL: insert_dup_elt1_mem_v8i16_i32:
2751 ; SSE: # %bb.0:
2752 ; SSE-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
2753 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,1,2,3,4,5,6,7]
2754 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2755 ; SSE-NEXT: retq
2756 ;
2757 ; AVX1-LABEL: insert_dup_elt1_mem_v8i16_i32:
2758 ; AVX1: # %bb.0:
2759 ; AVX1-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
2760 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[1,1,2,3,4,5,6,7]
2761 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2762 ; AVX1-NEXT: retq
2763 ;
2764 ; AVX2OR512VL-LABEL: insert_dup_elt1_mem_v8i16_i32:
2765 ; AVX2OR512VL: # %bb.0:
2766 ; AVX2OR512VL-NEXT: vpbroadcastw 2(%rdi), %xmm0
2767 ; AVX2OR512VL-NEXT: retq
2768 %tmp = load i32, i32* %ptr, align 4
2769 %tmp1 = insertelement <4 x i32> zeroinitializer, i32 %tmp, i32 0
2770 %tmp2 = bitcast <4 x i32> %tmp1 to <8 x i16>
2771 %tmp3 = shufflevector <8 x i16> %tmp2, <8 x i16> undef, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
2772 ret <8 x i16> %tmp3
2773 }
2775 define <8 x i16> @insert_dup_elt3_mem_v8i16_i32(i32* %ptr) {
2776 ; SSE2-LABEL: insert_dup_elt3_mem_v8i16_i32:
2777 ; SSE2: # %bb.0:
2778 ; SSE2-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
2779 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,1,0,1,4,5,6,7]
2780 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2781 ; SSE2-NEXT: retq
2782 ;
2783 ; SSSE3-LABEL: insert_dup_elt3_mem_v8i16_i32:
2784 ; SSSE3: # %bb.0:
2785 ; SSSE3-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
2786 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
2787 ; SSSE3-NEXT: retq
2788 ;
2789 ; SSE41-LABEL: insert_dup_elt3_mem_v8i16_i32:
2790 ; SSE41: # %bb.0:
2791 ; SSE41-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
2792 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
2793 ; SSE41-NEXT: retq
2794 ;
2795 ; AVX1-LABEL: insert_dup_elt3_mem_v8i16_i32:
2796 ; AVX1: # %bb.0:
2797 ; AVX1-NEXT: vbroadcastss (%rdi), %xmm0
2798 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[3,3,2,3,4,5,6,7]
2799 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2800 ; AVX1-NEXT: retq
2801 ;
2802 ; AVX2OR512VL-LABEL: insert_dup_elt3_mem_v8i16_i32:
2803 ; AVX2OR512VL: # %bb.0:
2804 ; AVX2OR512VL-NEXT: vpbroadcastw 2(%rdi), %xmm0
2805 ; AVX2OR512VL-NEXT: retq
2806 %tmp = load i32, i32* %ptr, align 4
2807 %tmp1 = insertelement <4 x i32> zeroinitializer, i32 %tmp, i32 1
2808 %tmp2 = bitcast <4 x i32> %tmp1 to <8 x i16>
2809 %tmp3 = shufflevector <8 x i16> %tmp2, <8 x i16> undef, <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
2810 ret <8 x i16> %tmp3
2811 }
2813 define <8 x i16> @insert_dup_elt1_mem_v8i16_sext_i16(i16* %ptr) {
2814 ; SSE-LABEL: insert_dup_elt1_mem_v8i16_sext_i16:
2815 ; SSE: # %bb.0:
2816 ; SSE-NEXT: movswl (%rdi), %eax
2817 ; SSE-NEXT: movd %eax, %xmm0
2818 ; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,1,2,3,4,5,6,7]
2819 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2820 ; SSE-NEXT: retq
2821 ;
2822 ; AVX1-LABEL: insert_dup_elt1_mem_v8i16_sext_i16:
2823 ; AVX1: # %bb.0:
2824 ; AVX1-NEXT: movswl (%rdi), %eax
2825 ; AVX1-NEXT: vmovd %eax, %xmm0
2826 ; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[1,1,2,3,4,5,6,7]
2827 ; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2828 ; AVX1-NEXT: retq
2829 ;
2830 ; AVX2-LABEL: insert_dup_elt1_mem_v8i16_sext_i16:
2831 ; AVX2: # %bb.0:
2832 ; AVX2-NEXT: movswl (%rdi), %eax
2833 ; AVX2-NEXT: shrl $16, %eax
2834 ; AVX2-NEXT: vmovd %eax, %xmm0
2835 ; AVX2-NEXT: vpbroadcastw %xmm0, %xmm0
2836 ; AVX2-NEXT: retq
2837 ;
2838 ; AVX512VL-LABEL: insert_dup_elt1_mem_v8i16_sext_i16:
2839 ; AVX512VL: # %bb.0:
2840 ; AVX512VL-NEXT: movswl (%rdi), %eax
2841 ; AVX512VL-NEXT: shrl $16, %eax
2842 ; AVX512VL-NEXT: vpbroadcastw %eax, %xmm0
2843 ; AVX512VL-NEXT: retq
2844 %tmp = load i16, i16* %ptr, align 2
2845 %tmp1 = sext i16 %tmp to i32
2846 %tmp2 = insertelement <4 x i32> zeroinitializer, i32 %tmp1, i32 0
2847 %tmp3 = bitcast <4 x i32> %tmp2 to <8 x i16>
2848 %tmp4 = shufflevector <8 x i16> %tmp3, <8 x i16> undef, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
2849 ret <8 x i16> %tmp4
2850 }
2852 define <8 x i16> @insert_dup_elt3_mem_v8i16_sext_i16(i16* %ptr) {
2853 ; SSE2-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
2854 ; SSE2: # %bb.0:
2855 ; SSE2-NEXT: movswl (%rdi), %eax
2856 ; SSE2-NEXT: movd %eax, %xmm0
2857 ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,1,0,1,4,5,6,7]
2858 ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
2859 ; SSE2-NEXT: retq
2860 ;
2861 ; SSSE3-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
2862 ; SSSE3: # %bb.0:
2863 ; SSSE3-NEXT: movswl (%rdi), %eax
2864 ; SSSE3-NEXT: movd %eax, %xmm0
2865 ; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
2866 ; SSSE3-NEXT: retq
2867 ;
2868 ; SSE41-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
2869 ; SSE41: # %bb.0:
2870 ; SSE41-NEXT: movswl (%rdi), %eax
2871 ; SSE41-NEXT: movd %eax, %xmm0
2872 ; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
2873 ; SSE41-NEXT: retq
2874 ;
2875 ; AVX1-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
2876 ; AVX1: # %bb.0:
2877 ; AVX1-NEXT: movswl (%rdi), %eax
2878 ; AVX1-NEXT: vmovd %eax, %xmm0
2879 ; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
2880 ; AVX1-NEXT: retq
2881 ;
2882 ; AVX2-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
2883 ; AVX2: # %bb.0:
2884 ; AVX2-NEXT: movswl (%rdi), %eax
2885 ; AVX2-NEXT: shrl $16, %eax
2886 ; AVX2-NEXT: vmovd %eax, %xmm0
2887 ; AVX2-NEXT: vpbroadcastw %xmm0, %xmm0
2888 ; AVX2-NEXT: retq
2889 ;
2890 ; AVX512VL-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
2891 ; AVX512VL: # %bb.0:
2892 ; AVX512VL-NEXT: movswl (%rdi), %eax
2893 ; AVX512VL-NEXT: shrl $16, %eax
2894 ; AVX512VL-NEXT: vpbroadcastw %eax, %xmm0
2895 ; AVX512VL-NEXT: retq
2896 %tmp = load i16, i16* %ptr, align 2
2897 %tmp1 = sext i16 %tmp to i32
2898 %tmp2 = insertelement <4 x i32> zeroinitializer, i32 %tmp1, i32 1
2899 %tmp3 = bitcast <4 x i32> %tmp2 to <8 x i16>
2900 %tmp4 = shufflevector <8 x i16> %tmp3, <8 x i16> undef, <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
2901 ret <8 x i16> %tmp4
2902 }