1 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSE2
2 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse3 | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSE3
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=AVX1
6 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
8 define <2 x i64> @testv2i64(<2 x i64> %in) nounwind {
9 ; SSE2-LABEL: testv2i64:
11 ; SSE2-NEXT: movdqa %xmm0, %xmm1
12 ; SSE2-NEXT: psrlq $1, %xmm1
13 ; SSE2-NEXT: pand {{.*}}(%rip), %xmm1
14 ; SSE2-NEXT: psubq %xmm1, %xmm0
15 ; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [3689348814741910323,3689348814741910323]
16 ; SSE2-NEXT: movdqa %xmm0, %xmm2
17 ; SSE2-NEXT: pand %xmm1, %xmm2
18 ; SSE2-NEXT: psrlq $2, %xmm0
19 ; SSE2-NEXT: pand %xmm1, %xmm0
20 ; SSE2-NEXT: paddq %xmm2, %xmm0
21 ; SSE2-NEXT: movdqa %xmm0, %xmm1
22 ; SSE2-NEXT: psrlq $4, %xmm1
23 ; SSE2-NEXT: paddq %xmm0, %xmm1
24 ; SSE2-NEXT: pand {{.*}}(%rip), %xmm1
25 ; SSE2-NEXT: pxor %xmm0, %xmm0
26 ; SSE2-NEXT: psadbw %xmm0, %xmm1
27 ; SSE2-NEXT: movdqa %xmm1, %xmm0
30 ; SSE3-LABEL: testv2i64:
32 ; SSE3-NEXT: movdqa %xmm0, %xmm1
33 ; SSE3-NEXT: psrlq $1, %xmm1
34 ; SSE3-NEXT: pand {{.*}}(%rip), %xmm1
35 ; SSE3-NEXT: psubq %xmm1, %xmm0
36 ; SSE3-NEXT: movdqa {{.*#+}} xmm1 = [3689348814741910323,3689348814741910323]
37 ; SSE3-NEXT: movdqa %xmm0, %xmm2
38 ; SSE3-NEXT: pand %xmm1, %xmm2
39 ; SSE3-NEXT: psrlq $2, %xmm0
40 ; SSE3-NEXT: pand %xmm1, %xmm0
41 ; SSE3-NEXT: paddq %xmm2, %xmm0
42 ; SSE3-NEXT: movdqa %xmm0, %xmm1
43 ; SSE3-NEXT: psrlq $4, %xmm1
44 ; SSE3-NEXT: paddq %xmm0, %xmm1
45 ; SSE3-NEXT: pand {{.*}}(%rip), %xmm1
46 ; SSE3-NEXT: pxor %xmm0, %xmm0
47 ; SSE3-NEXT: psadbw %xmm0, %xmm1
48 ; SSE3-NEXT: movdqa %xmm1, %xmm0
51 ; SSSE3-LABEL: testv2i64:
53 ; SSSE3-NEXT: movdqa {{.*#+}} xmm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
54 ; SSSE3-NEXT: movdqa %xmm0, %xmm2
55 ; SSSE3-NEXT: pand %xmm1, %xmm2
56 ; SSSE3-NEXT: movdqa {{.*#+}} xmm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
57 ; SSSE3-NEXT: movdqa %xmm3, %xmm4
58 ; SSSE3-NEXT: pshufb %xmm2, %xmm4
59 ; SSSE3-NEXT: psrlw $4, %xmm0
60 ; SSSE3-NEXT: pand %xmm1, %xmm0
61 ; SSSE3-NEXT: pshufb %xmm0, %xmm3
62 ; SSSE3-NEXT: paddb %xmm4, %xmm3
63 ; SSSE3-NEXT: pxor %xmm0, %xmm0
64 ; SSSE3-NEXT: psadbw %xmm3, %xmm0
67 ; SSE41-LABEL: testv2i64:
69 ; SSE41-NEXT: movdqa {{.*#+}} xmm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
70 ; SSE41-NEXT: movdqa %xmm0, %xmm2
71 ; SSE41-NEXT: pand %xmm1, %xmm2
72 ; SSE41-NEXT: movdqa {{.*#+}} xmm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
73 ; SSE41-NEXT: movdqa %xmm3, %xmm4
74 ; SSE41-NEXT: pshufb %xmm2, %xmm4
75 ; SSE41-NEXT: psrlw $4, %xmm0
76 ; SSE41-NEXT: pand %xmm1, %xmm0
77 ; SSE41-NEXT: pshufb %xmm0, %xmm3
78 ; SSE41-NEXT: paddb %xmm4, %xmm3
79 ; SSE41-NEXT: pxor %xmm0, %xmm0
80 ; SSE41-NEXT: psadbw %xmm3, %xmm0
83 ; AVX-LABEL: testv2i64:
85 ; AVX-NEXT: vmovdqa {{.*#+}} xmm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
86 ; AVX-NEXT: vpand %xmm1, %xmm0, %xmm2
87 ; AVX-NEXT: vmovdqa {{.*#+}} xmm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
88 ; AVX-NEXT: vpshufb %xmm2, %xmm3, %xmm2
89 ; AVX-NEXT: vpsrlw $4, %xmm0, %xmm0
90 ; AVX-NEXT: vpand %xmm1, %xmm0, %xmm0
91 ; AVX-NEXT: vpshufb %xmm0, %xmm3, %xmm0
92 ; AVX-NEXT: vpaddb %xmm2, %xmm0, %xmm0
93 ; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
94 ; AVX-NEXT: vpsadbw %xmm0, %xmm1, %xmm0
96 %out = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> %in)
100 define <4 x i32> @testv4i32(<4 x i32> %in) nounwind {
101 ; SSE2-LABEL: testv4i32:
103 ; SSE2-NEXT: movdqa %xmm0, %xmm1
104 ; SSE2-NEXT: psrld $1, %xmm1
105 ; SSE2-NEXT: pand {{.*}}(%rip), %xmm1
106 ; SSE2-NEXT: psubd %xmm1, %xmm0
107 ; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [858993459,858993459,858993459,858993459]
108 ; SSE2-NEXT: movdqa %xmm0, %xmm2
109 ; SSE2-NEXT: pand %xmm1, %xmm2
110 ; SSE2-NEXT: psrld $2, %xmm0
111 ; SSE2-NEXT: pand %xmm1, %xmm0
112 ; SSE2-NEXT: paddd %xmm2, %xmm0
113 ; SSE2-NEXT: movdqa %xmm0, %xmm1
114 ; SSE2-NEXT: psrld $4, %xmm1
115 ; SSE2-NEXT: paddd %xmm0, %xmm1
116 ; SSE2-NEXT: pand {{.*}}(%rip), %xmm1
117 ; SSE2-NEXT: pxor %xmm0, %xmm0
118 ; SSE2-NEXT: movdqa %xmm1, %xmm2
119 ; SSE2-NEXT: punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm0[2],xmm2[3],xmm0[3]
120 ; SSE2-NEXT: psadbw %xmm0, %xmm2
121 ; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
122 ; SSE2-NEXT: psadbw %xmm0, %xmm1
123 ; SSE2-NEXT: packuswb %xmm2, %xmm1
124 ; SSE2-NEXT: movdqa %xmm1, %xmm0
127 ; SSE3-LABEL: testv4i32:
129 ; SSE3-NEXT: movdqa %xmm0, %xmm1
130 ; SSE3-NEXT: psrld $1, %xmm1
131 ; SSE3-NEXT: pand {{.*}}(%rip), %xmm1
132 ; SSE3-NEXT: psubd %xmm1, %xmm0
133 ; SSE3-NEXT: movdqa {{.*#+}} xmm1 = [858993459,858993459,858993459,858993459]
134 ; SSE3-NEXT: movdqa %xmm0, %xmm2
135 ; SSE3-NEXT: pand %xmm1, %xmm2
136 ; SSE3-NEXT: psrld $2, %xmm0
137 ; SSE3-NEXT: pand %xmm1, %xmm0
138 ; SSE3-NEXT: paddd %xmm2, %xmm0
139 ; SSE3-NEXT: movdqa %xmm0, %xmm1
140 ; SSE3-NEXT: psrld $4, %xmm1
141 ; SSE3-NEXT: paddd %xmm0, %xmm1
142 ; SSE3-NEXT: pand {{.*}}(%rip), %xmm1
143 ; SSE3-NEXT: pxor %xmm0, %xmm0
144 ; SSE3-NEXT: movdqa %xmm1, %xmm2
145 ; SSE3-NEXT: punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm0[2],xmm2[3],xmm0[3]
146 ; SSE3-NEXT: psadbw %xmm0, %xmm2
147 ; SSE3-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
148 ; SSE3-NEXT: psadbw %xmm0, %xmm1
149 ; SSE3-NEXT: packuswb %xmm2, %xmm1
150 ; SSE3-NEXT: movdqa %xmm1, %xmm0
153 ; SSSE3-LABEL: testv4i32:
155 ; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
156 ; SSSE3-NEXT: movdqa %xmm0, %xmm3
157 ; SSSE3-NEXT: pand %xmm2, %xmm3
158 ; SSSE3-NEXT: movdqa {{.*#+}} xmm1 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
159 ; SSSE3-NEXT: movdqa %xmm1, %xmm4
160 ; SSSE3-NEXT: pshufb %xmm3, %xmm4
161 ; SSSE3-NEXT: psrlw $4, %xmm0
162 ; SSSE3-NEXT: pand %xmm2, %xmm0
163 ; SSSE3-NEXT: pshufb %xmm0, %xmm1
164 ; SSSE3-NEXT: paddb %xmm4, %xmm1
165 ; SSSE3-NEXT: pxor %xmm0, %xmm0
166 ; SSSE3-NEXT: movdqa %xmm1, %xmm2
167 ; SSSE3-NEXT: punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm0[2],xmm2[3],xmm0[3]
168 ; SSSE3-NEXT: psadbw %xmm0, %xmm2
169 ; SSSE3-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
170 ; SSSE3-NEXT: psadbw %xmm0, %xmm1
171 ; SSSE3-NEXT: packuswb %xmm2, %xmm1
172 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
175 ; SSE41-LABEL: testv4i32:
177 ; SSE41-NEXT: movdqa {{.*#+}} xmm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
178 ; SSE41-NEXT: movdqa %xmm0, %xmm3
179 ; SSE41-NEXT: pand %xmm2, %xmm3
180 ; SSE41-NEXT: movdqa {{.*#+}} xmm1 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
181 ; SSE41-NEXT: movdqa %xmm1, %xmm4
182 ; SSE41-NEXT: pshufb %xmm3, %xmm4
183 ; SSE41-NEXT: psrlw $4, %xmm0
184 ; SSE41-NEXT: pand %xmm2, %xmm0
185 ; SSE41-NEXT: pshufb %xmm0, %xmm1
186 ; SSE41-NEXT: paddb %xmm4, %xmm1
187 ; SSE41-NEXT: pxor %xmm0, %xmm0
188 ; SSE41-NEXT: movdqa %xmm1, %xmm2
189 ; SSE41-NEXT: punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm0[2],xmm2[3],xmm0[3]
190 ; SSE41-NEXT: psadbw %xmm0, %xmm2
191 ; SSE41-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
192 ; SSE41-NEXT: psadbw %xmm0, %xmm1
193 ; SSE41-NEXT: packuswb %xmm2, %xmm1
194 ; SSE41-NEXT: movdqa %xmm1, %xmm0
197 ; AVX-LABEL: testv4i32:
199 ; AVX-NEXT: vmovdqa {{.*#+}} xmm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
200 ; AVX-NEXT: vpand %xmm1, %xmm0, %xmm2
201 ; AVX-NEXT: vmovdqa {{.*#+}} xmm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
202 ; AVX-NEXT: vpshufb %xmm2, %xmm3, %xmm2
203 ; AVX-NEXT: vpsrlw $4, %xmm0, %xmm0
204 ; AVX-NEXT: vpand %xmm1, %xmm0, %xmm0
205 ; AVX-NEXT: vpshufb %xmm0, %xmm3, %xmm0
206 ; AVX-NEXT: vpaddb %xmm2, %xmm0, %xmm0
207 ; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
208 ; AVX-NEXT: vpunpckhdq {{.*#+}} xmm2 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
209 ; AVX-NEXT: vpsadbw %xmm2, %xmm1, %xmm2
210 ; AVX-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
211 ; AVX-NEXT: vpsadbw %xmm0, %xmm1, %xmm0
212 ; AVX-NEXT: vpackuswb %xmm2, %xmm0, %xmm0
214 %out = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> %in)
218 define <8 x i16> @testv8i16(<8 x i16> %in) nounwind {
219 ; SSE2-LABEL: testv8i16:
221 ; SSE2-NEXT: movdqa %xmm0, %xmm1
222 ; SSE2-NEXT: psrlw $1, %xmm1
223 ; SSE2-NEXT: pand {{.*}}(%rip), %xmm1
224 ; SSE2-NEXT: psubw %xmm1, %xmm0
225 ; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [13107,13107,13107,13107,13107,13107,13107,13107]
226 ; SSE2-NEXT: movdqa %xmm0, %xmm2
227 ; SSE2-NEXT: pand %xmm1, %xmm2
228 ; SSE2-NEXT: psrlw $2, %xmm0
229 ; SSE2-NEXT: pand %xmm1, %xmm0
230 ; SSE2-NEXT: paddw %xmm2, %xmm0
231 ; SSE2-NEXT: movdqa %xmm0, %xmm1
232 ; SSE2-NEXT: psrlw $4, %xmm1
233 ; SSE2-NEXT: paddw %xmm0, %xmm1
234 ; SSE2-NEXT: pand {{.*}}(%rip), %xmm1
235 ; SSE2-NEXT: movdqa %xmm1, %xmm0
236 ; SSE2-NEXT: psllw $8, %xmm0
237 ; SSE2-NEXT: paddb %xmm1, %xmm0
238 ; SSE2-NEXT: psrlw $8, %xmm0
241 ; SSE3-LABEL: testv8i16:
243 ; SSE3-NEXT: movdqa %xmm0, %xmm1
244 ; SSE3-NEXT: psrlw $1, %xmm1
245 ; SSE3-NEXT: pand {{.*}}(%rip), %xmm1
246 ; SSE3-NEXT: psubw %xmm1, %xmm0
247 ; SSE3-NEXT: movdqa {{.*#+}} xmm1 = [13107,13107,13107,13107,13107,13107,13107,13107]
248 ; SSE3-NEXT: movdqa %xmm0, %xmm2
249 ; SSE3-NEXT: pand %xmm1, %xmm2
250 ; SSE3-NEXT: psrlw $2, %xmm0
251 ; SSE3-NEXT: pand %xmm1, %xmm0
252 ; SSE3-NEXT: paddw %xmm2, %xmm0
253 ; SSE3-NEXT: movdqa %xmm0, %xmm1
254 ; SSE3-NEXT: psrlw $4, %xmm1
255 ; SSE3-NEXT: paddw %xmm0, %xmm1
256 ; SSE3-NEXT: pand {{.*}}(%rip), %xmm1
257 ; SSE3-NEXT: movdqa %xmm1, %xmm0
258 ; SSE3-NEXT: psllw $8, %xmm0
259 ; SSE3-NEXT: paddb %xmm1, %xmm0
260 ; SSE3-NEXT: psrlw $8, %xmm0
263 ; SSSE3-LABEL: testv8i16:
265 ; SSSE3-NEXT: movdqa {{.*#+}} xmm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
266 ; SSSE3-NEXT: movdqa %xmm0, %xmm2
267 ; SSSE3-NEXT: pand %xmm1, %xmm2
268 ; SSSE3-NEXT: movdqa {{.*#+}} xmm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
269 ; SSSE3-NEXT: movdqa %xmm3, %xmm4
270 ; SSSE3-NEXT: pshufb %xmm2, %xmm4
271 ; SSSE3-NEXT: psrlw $4, %xmm0
272 ; SSSE3-NEXT: pand %xmm1, %xmm0
273 ; SSSE3-NEXT: pshufb %xmm0, %xmm3
274 ; SSSE3-NEXT: paddb %xmm4, %xmm3
275 ; SSSE3-NEXT: movdqa %xmm3, %xmm0
276 ; SSSE3-NEXT: psllw $8, %xmm0
277 ; SSSE3-NEXT: paddb %xmm3, %xmm0
278 ; SSSE3-NEXT: psrlw $8, %xmm0
281 ; SSE41-LABEL: testv8i16:
283 ; SSE41-NEXT: movdqa {{.*#+}} xmm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
284 ; SSE41-NEXT: movdqa %xmm0, %xmm2
285 ; SSE41-NEXT: pand %xmm1, %xmm2
286 ; SSE41-NEXT: movdqa {{.*#+}} xmm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
287 ; SSE41-NEXT: movdqa %xmm3, %xmm4
288 ; SSE41-NEXT: pshufb %xmm2, %xmm4
289 ; SSE41-NEXT: psrlw $4, %xmm0
290 ; SSE41-NEXT: pand %xmm1, %xmm0
291 ; SSE41-NEXT: pshufb %xmm0, %xmm3
292 ; SSE41-NEXT: paddb %xmm4, %xmm3
293 ; SSE41-NEXT: movdqa %xmm3, %xmm0
294 ; SSE41-NEXT: psllw $8, %xmm0
295 ; SSE41-NEXT: paddb %xmm3, %xmm0
296 ; SSE41-NEXT: psrlw $8, %xmm0
299 ; AVX-LABEL: testv8i16:
301 ; AVX-NEXT: vmovdqa {{.*#+}} xmm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
302 ; AVX-NEXT: vpand %xmm1, %xmm0, %xmm2
303 ; AVX-NEXT: vmovdqa {{.*#+}} xmm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
304 ; AVX-NEXT: vpshufb %xmm2, %xmm3, %xmm2
305 ; AVX-NEXT: vpsrlw $4, %xmm0, %xmm0
306 ; AVX-NEXT: vpand %xmm1, %xmm0, %xmm0
307 ; AVX-NEXT: vpshufb %xmm0, %xmm3, %xmm0
308 ; AVX-NEXT: vpaddb %xmm2, %xmm0, %xmm0
309 ; AVX-NEXT: vpsllw $8, %xmm0, %xmm1
310 ; AVX-NEXT: vpaddb %xmm0, %xmm1, %xmm0
311 ; AVX-NEXT: vpsrlw $8, %xmm0, %xmm0
313 %out = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> %in)
317 define <16 x i8> @testv16i8(<16 x i8> %in) nounwind {
318 ; SSE2-LABEL: testv16i8:
320 ; SSE2-NEXT: movdqa %xmm0, %xmm1
321 ; SSE2-NEXT: psrlw $1, %xmm1
322 ; SSE2-NEXT: pand {{.*}}(%rip), %xmm1
323 ; SSE2-NEXT: psubb %xmm1, %xmm0
324 ; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [51,51,51,51,51,51,51,51,51,51,51,51,51,51,51,51]
325 ; SSE2-NEXT: movdqa %xmm0, %xmm2
326 ; SSE2-NEXT: pand %xmm1, %xmm2
327 ; SSE2-NEXT: psrlw $2, %xmm0
328 ; SSE2-NEXT: pand %xmm1, %xmm0
329 ; SSE2-NEXT: paddb %xmm2, %xmm0
330 ; SSE2-NEXT: movdqa %xmm0, %xmm1
331 ; SSE2-NEXT: psrlw $4, %xmm1
332 ; SSE2-NEXT: paddb %xmm0, %xmm1
333 ; SSE2-NEXT: pand {{.*}}(%rip), %xmm1
334 ; SSE2-NEXT: movdqa %xmm1, %xmm0
337 ; SSE3-LABEL: testv16i8:
339 ; SSE3-NEXT: movdqa %xmm0, %xmm1
340 ; SSE3-NEXT: psrlw $1, %xmm1
341 ; SSE3-NEXT: pand {{.*}}(%rip), %xmm1
342 ; SSE3-NEXT: psubb %xmm1, %xmm0
343 ; SSE3-NEXT: movdqa {{.*#+}} xmm1 = [51,51,51,51,51,51,51,51,51,51,51,51,51,51,51,51]
344 ; SSE3-NEXT: movdqa %xmm0, %xmm2
345 ; SSE3-NEXT: pand %xmm1, %xmm2
346 ; SSE3-NEXT: psrlw $2, %xmm0
347 ; SSE3-NEXT: pand %xmm1, %xmm0
348 ; SSE3-NEXT: paddb %xmm2, %xmm0
349 ; SSE3-NEXT: movdqa %xmm0, %xmm1
350 ; SSE3-NEXT: psrlw $4, %xmm1
351 ; SSE3-NEXT: paddb %xmm0, %xmm1
352 ; SSE3-NEXT: pand {{.*}}(%rip), %xmm1
353 ; SSE3-NEXT: movdqa %xmm1, %xmm0
356 ; SSSE3-LABEL: testv16i8:
358 ; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
359 ; SSSE3-NEXT: movdqa %xmm0, %xmm3
360 ; SSSE3-NEXT: pand %xmm2, %xmm3
361 ; SSSE3-NEXT: movdqa {{.*#+}} xmm1 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
362 ; SSSE3-NEXT: movdqa %xmm1, %xmm4
363 ; SSSE3-NEXT: pshufb %xmm3, %xmm4
364 ; SSSE3-NEXT: psrlw $4, %xmm0
365 ; SSSE3-NEXT: pand %xmm2, %xmm0
366 ; SSSE3-NEXT: pshufb %xmm0, %xmm1
367 ; SSSE3-NEXT: paddb %xmm4, %xmm1
368 ; SSSE3-NEXT: movdqa %xmm1, %xmm0
371 ; SSE41-LABEL: testv16i8:
373 ; SSE41-NEXT: movdqa {{.*#+}} xmm2 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
374 ; SSE41-NEXT: movdqa %xmm0, %xmm3
375 ; SSE41-NEXT: pand %xmm2, %xmm3
376 ; SSE41-NEXT: movdqa {{.*#+}} xmm1 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
377 ; SSE41-NEXT: movdqa %xmm1, %xmm4
378 ; SSE41-NEXT: pshufb %xmm3, %xmm4
379 ; SSE41-NEXT: psrlw $4, %xmm0
380 ; SSE41-NEXT: pand %xmm2, %xmm0
381 ; SSE41-NEXT: pshufb %xmm0, %xmm1
382 ; SSE41-NEXT: paddb %xmm4, %xmm1
383 ; SSE41-NEXT: movdqa %xmm1, %xmm0
386 ; AVX-LABEL: testv16i8:
388 ; AVX-NEXT: vmovdqa {{.*#+}} xmm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
389 ; AVX-NEXT: vpand %xmm1, %xmm0, %xmm2
390 ; AVX-NEXT: vmovdqa {{.*#+}} xmm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
391 ; AVX-NEXT: vpshufb %xmm2, %xmm3, %xmm2
392 ; AVX-NEXT: vpsrlw $4, %xmm0, %xmm0
393 ; AVX-NEXT: vpand %xmm1, %xmm0, %xmm0
394 ; AVX-NEXT: vpshufb %xmm0, %xmm3, %xmm0
395 ; AVX-NEXT: vpaddb %xmm2, %xmm0, %xmm0
397 %out = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> %in)
401 define <2 x i64> @foldv2i64() nounwind {
402 ; SSE-LABEL: foldv2i64:
404 ; SSE-NEXT: movaps {{.*#+}} xmm0 = [1,64]
407 ; AVX-LABEL: foldv2i64:
409 ; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [1,64]
411 %out = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> <i64 256, i64 -1>)
415 define <4 x i32> @foldv4i32() nounwind {
416 ; SSE-LABEL: foldv4i32:
418 ; SSE-NEXT: movaps {{.*#+}} xmm0 = [1,32,0,8]
421 ; AVX-LABEL: foldv4i32:
423 ; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [1,32,0,8]
425 %out = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> <i32 256, i32 -1, i32 0, i32 255>)
429 define <8 x i16> @foldv8i16() nounwind {
430 ; SSE-LABEL: foldv8i16:
432 ; SSE-NEXT: movaps {{.*#+}} xmm0 = [1,16,0,8,0,3,2,3]
435 ; AVX-LABEL: foldv8i16:
437 ; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [1,16,0,8,0,3,2,3]
439 %out = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> <i16 256, i16 -1, i16 0, i16 255, i16 -65536, i16 7, i16 24, i16 88>)
443 define <16 x i8> @foldv16i8() nounwind {
444 ; SSE-LABEL: foldv16i8:
446 ; SSE-NEXT: movaps {{.*#+}} xmm0 = [0,8,0,8,0,3,2,3,7,7,1,1,1,1,1,1]
449 ; AVX-LABEL: foldv16i8:
451 ; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [0,8,0,8,0,3,2,3,7,7,1,1,1,1,1,1]
453 %out = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> <i8 256, i8 -1, i8 0, i8 255, i8 -65536, i8 7, i8 24, i8 88, i8 -2, i8 254, i8 1, i8 2, i8 4, i8 8, i8 16, i8 32>)
457 declare <2 x i64> @llvm.ctpop.v2i64(<2 x i64>)
458 declare <4 x i32> @llvm.ctpop.v4i32(<4 x i32>)
459 declare <8 x i16> @llvm.ctpop.v8i16(<8 x i16>)
460 declare <16 x i8> @llvm.ctpop.v16i8(<16 x i8>)