avg-mask.ll
20 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw -mattr=+avx512vl | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512BWVL
define <16 x i8> @avg_v16i8_mask(<16 x i8> %a, <16 x i8> %b, <16 x i8> %src, i16 %mask) nounwind {
; AVX512F-LABEL: avg_v16i8_mask:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpavgb %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z}
; AVX512F-NEXT: vpmovdb %zmm1, %xmm1
; AVX512F-NEXT: vpblendvb %xmm1, %xmm0, %xmm2, %xmm0
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v16i8_mask:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovd %edi, %k1
; AVX512BWVL-NEXT: vpavgb %xmm1, %xmm0, %xmm2 {%k1}
; AVX512BWVL-NEXT: vmovdqa %xmm2, %xmm0
; AVX512BWVL-NEXT: retq
%za = zext <16 x i8> %a to <16 x i16>
%zb = zext <16 x i8> %b to <16 x i16>
%add = add nuw nsw <16 x i16> %za, %zb
%add1 = add nuw nsw <16 x i16> %add, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%lshr = lshr <16 x i16> %add1, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%trunc = trunc <16 x i16> %lshr to <16 x i8>
%mask1 = bitcast i16 %mask to <16 x i1>
%res = select <16 x i1> %mask1, <16 x i8> %trunc, <16 x i8> %src
ret <16 x i8> %res
}
define <16 x i8> @avg_v16i8_maskz(<16 x i8> %a, <16 x i8> %b, i16 %mask) nounwind {
; AVX512F-LABEL: avg_v16i8_maskz:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpavgb %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z}
; AVX512F-NEXT: vpmovdb %zmm1, %xmm1
; AVX512F-NEXT: vpand %xmm0, %xmm1, %xmm0
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v16i8_maskz:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovd %edi, %k1
; AVX512BWVL-NEXT: vpavgb %xmm1, %xmm0, %xmm0 {%k1} {z}
; AVX512BWVL-NEXT: retq
%za = zext <16 x i8> %a to <16 x i16>
%zb = zext <16 x i8> %b to <16 x i16>
%add = add nuw nsw <16 x i16> %za, %zb
%add1 = add nuw nsw <16 x i16> %add, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%lshr = lshr <16 x i16> %add1, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%trunc = trunc <16 x i16> %lshr to <16 x i8>
%mask1 = bitcast i16 %mask to <16 x i1>
%res = select <16 x i1> %mask1, <16 x i8> %trunc, <16 x i8> zeroinitializer
ret <16 x i8> %res
}
define <32 x i8> @avg_v32i8_mask(<32 x i8> %a, <32 x i8> %b, <32 x i8> %src, i32 %mask) nounwind {
; AVX512F-LABEL: avg_v32i8_mask:
; AVX512F: # %bb.0:
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: shrl $16, %edi
; AVX512F-NEXT: vpavgb %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: kmovw %edi, %k2
; AVX512F-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z}
; AVX512F-NEXT: vpmovdb %zmm1, %xmm1
; AVX512F-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k2} {z}
; AVX512F-NEXT: vpmovdb %zmm3, %xmm3
; AVX512F-NEXT: vinserti128 $1, %xmm3, %ymm1, %ymm1
; AVX512F-NEXT: vpblendvb %ymm1, %ymm0, %ymm2, %ymm0
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v32i8_mask:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovd %edi, %k1
; AVX512BWVL-NEXT: vpavgb %ymm1, %ymm0, %ymm2 {%k1}
; AVX512BWVL-NEXT: vmovdqa %ymm2, %ymm0
; AVX512BWVL-NEXT: retq
%za = zext <32 x i8> %a to <32 x i16>
%zb = zext <32 x i8> %b to <32 x i16>
%add = add nuw nsw <32 x i16> %za, %zb
%add1 = add nuw nsw <32 x i16> %add, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%lshr = lshr <32 x i16> %add1, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%trunc = trunc <32 x i16> %lshr to <32 x i8>
%mask1 = bitcast i32 %mask to <32 x i1>
%res = select <32 x i1> %mask1, <32 x i8> %trunc, <32 x i8> %src
ret <32 x i8> %res
}
define <32 x i8> @avg_v32i8_maskz(<32 x i8> %a, <32 x i8> %b, i32 %mask) nounwind {
; AVX512F-LABEL: avg_v32i8_maskz:
; AVX512F: # %bb.0:
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: shrl $16, %edi
; AVX512F-NEXT: vpavgb %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: kmovw %edi, %k2
; AVX512F-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z}
; AVX512F-NEXT: vpmovdb %zmm1, %xmm1
; AVX512F-NEXT: vpternlogd $255, %zmm2, %zmm2, %zmm2 {%k2} {z}
; AVX512F-NEXT: vpmovdb %zmm2, %xmm2
; AVX512F-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1
; AVX512F-NEXT: vpand %ymm0, %ymm1, %ymm0
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v32i8_maskz:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovd %edi, %k1
; AVX512BWVL-NEXT: vpavgb %ymm1, %ymm0, %ymm0 {%k1} {z}
; AVX512BWVL-NEXT: retq
%za = zext <32 x i8> %a to <32 x i16>
%zb = zext <32 x i8> %b to <32 x i16>
%add = add nuw nsw <32 x i16> %za, %zb
%add1 = add nuw nsw <32 x i16> %add, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%lshr = lshr <32 x i16> %add1, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%trunc = trunc <32 x i16> %lshr to <32 x i8>
%mask1 = bitcast i32 %mask to <32 x i1>
%res = select <32 x i1> %mask1, <32 x i8> %trunc, <32 x i8> zeroinitializer
ret <32 x i8> %res
}
define <64 x i8> @avg_v64i8_mask(<64 x i8> %a, <64 x i8> %b, <64 x i8> %src, i64 %mask) nounwind {
; AVX512F-LABEL: avg_v64i8_mask:
; AVX512F: # %bb.0:
; AVX512F-NEXT: movq %rdi, %rax
; AVX512F-NEXT: movl %edi, %ecx
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: shrq $32, %rdi
; AVX512F-NEXT: shrq $48, %rax
; AVX512F-NEXT: shrl $16, %ecx
; AVX512F-NEXT: vextracti64x4 $1, %zmm1, %ymm3
; AVX512F-NEXT: vextracti64x4 $1, %zmm0, %ymm4
; AVX512F-NEXT: vpavgb %ymm3, %ymm4, %ymm3
; AVX512F-NEXT: vpavgb %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: vinserti64x4 $1, %ymm3, %zmm0, %zmm1
; AVX512F-NEXT: kmovw %ecx, %k2
; AVX512F-NEXT: kmovw %eax, %k3
; AVX512F-NEXT: kmovw %edi, %k4
; AVX512F-NEXT: vpternlogd $255, %zmm0, %zmm0, %zmm0 {%k4} {z}
; AVX512F-NEXT: vpmovdb %zmm0, %xmm0
; AVX512F-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k3} {z}
; AVX512F-NEXT: vpmovdb %zmm3, %xmm3
; AVX512F-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm0
; AVX512F-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k1} {z}
; AVX512F-NEXT: vpmovdb %zmm3, %xmm3
; AVX512F-NEXT: vpternlogd $255, %zmm4, %zmm4, %zmm4 {%k2} {z}
; AVX512F-NEXT: vpmovdb %zmm4, %xmm4
; AVX512F-NEXT: vinserti128 $1, %xmm4, %ymm3, %ymm3
; AVX512F-NEXT: vinserti64x4 $1, %ymm0, %zmm3, %zmm0
; AVX512F-NEXT: vpternlogq $202, %zmm2, %zmm1, %zmm0
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v64i8_mask:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovq %rdi, %k1
; AVX512BWVL-NEXT: vpavgb %zmm1, %zmm0, %zmm2 {%k1}
; AVX512BWVL-NEXT: vmovdqa64 %zmm2, %zmm0
; AVX512BWVL-NEXT: retq
%za = zext <64 x i8> %a to <64 x i16>
%zb = zext <64 x i8> %b to <64 x i16>
%add = add nuw nsw <64 x i16> %za, %zb
%add1 = add nuw nsw <64 x i16> %add, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%lshr = lshr <64 x i16> %add1, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%trunc = trunc <64 x i16> %lshr to <64 x i8>
%mask1 = bitcast i64 %mask to <64 x i1>
%res = select <64 x i1> %mask1, <64 x i8> %trunc, <64 x i8> %src
ret <64 x i8> %res
}
define <64 x i8> @avg_v64i8_maskz(<64 x i8> %a, <64 x i8> %b, i64 %mask) nounwind {
; AVX512F-LABEL: avg_v64i8_maskz:
; AVX512F: # %bb.0:
; AVX512F-NEXT: movq %rdi, %rax
; AVX512F-NEXT: movl %edi, %ecx
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: shrq $32, %rdi
; AVX512F-NEXT: shrq $48, %rax
; AVX512F-NEXT: shrl $16, %ecx
; AVX512F-NEXT: vextracti64x4 $1, %zmm1, %ymm2
; AVX512F-NEXT: vextracti64x4 $1, %zmm0, %ymm3
; AVX512F-NEXT: vpavgb %ymm2, %ymm3, %ymm2
; AVX512F-NEXT: vpavgb %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: vinserti64x4 $1, %ymm2, %zmm0, %zmm0
; AVX512F-NEXT: kmovw %ecx, %k2
; AVX512F-NEXT: kmovw %eax, %k3
; AVX512F-NEXT: kmovw %edi, %k4
; AVX512F-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k4} {z}
; AVX512F-NEXT: vpmovdb %zmm1, %xmm1
; AVX512F-NEXT: vpternlogd $255, %zmm2, %zmm2, %zmm2 {%k3} {z}
; AVX512F-NEXT: vpmovdb %zmm2, %xmm2
; AVX512F-NEXT: vinserti128 $1, %xmm2, %ymm1, %ymm1
; AVX512F-NEXT: vpternlogd $255, %zmm2, %zmm2, %zmm2 {%k1} {z}
; AVX512F-NEXT: vpmovdb %zmm2, %xmm2
; AVX512F-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k2} {z}
; AVX512F-NEXT: vpmovdb %zmm3, %xmm3
; AVX512F-NEXT: vinserti128 $1, %xmm3, %ymm2, %ymm2
; AVX512F-NEXT: vinserti64x4 $1, %ymm1, %zmm2, %zmm1
; AVX512F-NEXT: vpandq %zmm0, %zmm1, %zmm0
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v64i8_maskz:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovq %rdi, %k1
; AVX512BWVL-NEXT: vpavgb %zmm1, %zmm0, %zmm0 {%k1} {z}
; AVX512BWVL-NEXT: retq
%za = zext <64 x i8> %a to <64 x i16>
%zb = zext <64 x i8> %b to <64 x i16>
%add = add nuw nsw <64 x i16> %za, %zb
%add1 = add nuw nsw <64 x i16> %add, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%lshr = lshr <64 x i16> %add1, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
%trunc = trunc <64 x i16> %lshr to <64 x i8>
%mask1 = bitcast i64 %mask to <64 x i1>
%res = select <64 x i1> %mask1, <64 x i8> %trunc, <64 x i8> zeroinitializer
ret <64 x i8> %res
}
define <8 x i16> @avg_v8i16_mask(<8 x i16> %a, <8 x i16> %b, <8 x i16> %src, i8 %mask) nounwind {
; AVX512F-LABEL: avg_v8i16_mask:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpavgw %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z}
; AVX512F-NEXT: vpmovdw %zmm1, %ymm1
; AVX512F-NEXT: vpblendvb %xmm1, %xmm0, %xmm2, %xmm0
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v8i16_mask:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovd %edi, %k1
; AVX512BWVL-NEXT: vpavgw %xmm1, %xmm0, %xmm2 {%k1}
; AVX512BWVL-NEXT: vmovdqa %xmm2, %xmm0
; AVX512BWVL-NEXT: retq
%za = zext <8 x i16> %a to <8 x i32>
%zb = zext <8 x i16> %b to <8 x i32>
%add = add nuw nsw <8 x i32> %za, %zb
%add1 = add nuw nsw <8 x i32> %add, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%lshr = lshr <8 x i32> %add1, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%trunc = trunc <8 x i32> %lshr to <8 x i16>
%mask1 = bitcast i8 %mask to <8 x i1>
%res = select <8 x i1> %mask1, <8 x i16> %trunc, <8 x i16> %src
ret <8 x i16> %res
}
define <8 x i16> @avg_v8i16_maskz(<8 x i16> %a, <8 x i16> %b, i8 %mask) nounwind {
; AVX512F-LABEL: avg_v8i16_maskz:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpavgw %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z}
; AVX512F-NEXT: vpmovdw %zmm1, %ymm1
; AVX512F-NEXT: vpand %xmm0, %xmm1, %xmm0
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v8i16_maskz:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovd %edi, %k1
; AVX512BWVL-NEXT: vpavgw %xmm1, %xmm0, %xmm0 {%k1} {z}
; AVX512BWVL-NEXT: retq
%za = zext <8 x i16> %a to <8 x i32>
%zb = zext <8 x i16> %b to <8 x i32>
%add = add nuw nsw <8 x i32> %za, %zb
%add1 = add nuw nsw <8 x i32> %add, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%lshr = lshr <8 x i32> %add1, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%trunc = trunc <8 x i32> %lshr to <8 x i16>
%mask1 = bitcast i8 %mask to <8 x i1>
%res = select <8 x i1> %mask1, <8 x i16> %trunc, <8 x i16> zeroinitializer
ret <8 x i16> %res
}
define <16 x i16> @avg_v16i16_mask(<16 x i16> %a, <16 x i16> %b, <16 x i16> %src, i16 %mask) nounwind {
; AVX512F-LABEL: avg_v16i16_mask:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpavgw %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z}
; AVX512F-NEXT: vpmovdw %zmm1, %ymm1
; AVX512F-NEXT: vpblendvb %ymm1, %ymm0, %ymm2, %ymm0
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v16i16_mask:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovd %edi, %k1
; AVX512BWVL-NEXT: vpavgw %ymm1, %ymm0, %ymm2 {%k1}
; AVX512BWVL-NEXT: vmovdqa %ymm2, %ymm0
; AVX512BWVL-NEXT: retq
%za = zext <16 x i16> %a to <16 x i32>
%zb = zext <16 x i16> %b to <16 x i32>
%add = add nuw nsw <16 x i32> %za, %zb
%add1 = add nuw nsw <16 x i32> %add, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%lshr = lshr <16 x i32> %add1, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%trunc = trunc <16 x i32> %lshr to <16 x i16>
%mask1 = bitcast i16 %mask to <16 x i1>
%res = select <16 x i1> %mask1, <16 x i16> %trunc, <16 x i16> %src
ret <16 x i16> %res
}
define <16 x i16> @avg_v16i16_maskz(<16 x i16> %a, <16 x i16> %b, i16 %mask) nounwind {
; AVX512F-LABEL: avg_v16i16_maskz:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpavgw %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z}
; AVX512F-NEXT: vpmovdw %zmm1, %ymm1
; AVX512F-NEXT: vpand %ymm0, %ymm1, %ymm0
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v16i16_maskz:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovd %edi, %k1
; AVX512BWVL-NEXT: vpavgw %ymm1, %ymm0, %ymm0 {%k1} {z}
; AVX512BWVL-NEXT: retq
%za = zext <16 x i16> %a to <16 x i32>
%zb = zext <16 x i16> %b to <16 x i32>
%add = add nuw nsw <16 x i32> %za, %zb
%add1 = add nuw nsw <16 x i32> %add, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%lshr = lshr <16 x i32> %add1, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%trunc = trunc <16 x i32> %lshr to <16 x i16>
%mask1 = bitcast i16 %mask to <16 x i1>
%res = select <16 x i1> %mask1, <16 x i16> %trunc, <16 x i16> zeroinitializer
ret <16 x i16> %res
}
define <32 x i16> @avg_v32i16_mask(<32 x i16> %a, <32 x i16> %b, <32 x i16> %src, i32 %mask) nounwind {
; AVX512F-LABEL: avg_v32i16_mask:
; AVX512F: # %bb.0:
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: shrl $16, %edi
; AVX512F-NEXT: vextracti64x4 $1, %zmm1, %ymm3
; AVX512F-NEXT: vextracti64x4 $1, %zmm0, %ymm4
; AVX512F-NEXT: vpavgw %ymm3, %ymm4, %ymm3
; AVX512F-NEXT: vpavgw %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: vinserti64x4 $1, %ymm3, %zmm0, %zmm1
; AVX512F-NEXT: kmovw %edi, %k2
; AVX512F-NEXT: vpternlogd $255, %zmm0, %zmm0, %zmm0 {%k1} {z}
; AVX512F-NEXT: vpmovdw %zmm0, %ymm0
; AVX512F-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 {%k2} {z}
; AVX512F-NEXT: vpmovdw %zmm3, %ymm3
; AVX512F-NEXT: vinserti64x4 $1, %ymm3, %zmm0, %zmm0
; AVX512F-NEXT: vpternlogq $202, %zmm2, %zmm1, %zmm0
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v32i16_mask:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovd %edi, %k1
; AVX512BWVL-NEXT: vpavgw %zmm1, %zmm0, %zmm2 {%k1}
; AVX512BWVL-NEXT: vmovdqa64 %zmm2, %zmm0
; AVX512BWVL-NEXT: retq
%za = zext <32 x i16> %a to <32 x i32>
%zb = zext <32 x i16> %b to <32 x i32>
%add = add nuw nsw <32 x i32> %za, %zb
%add1 = add nuw nsw <32 x i32> %add, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%lshr = lshr <32 x i32> %add1, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%trunc = trunc <32 x i32> %lshr to <32 x i16>
%mask1 = bitcast i32 %mask to <32 x i1>
%res = select <32 x i1> %mask1, <32 x i16> %trunc, <32 x i16> %src
ret <32 x i16> %res
}
define <32 x i16> @avg_v32i16_maskz(<32 x i16> %a, <32 x i16> %b, i32 %mask) nounwind {
; AVX512F-LABEL: avg_v32i16_maskz:
; AVX512F: # %bb.0:
; AVX512F-NEXT: kmovw %edi, %k1
; AVX512F-NEXT: shrl $16, %edi
; AVX512F-NEXT: vextracti64x4 $1, %zmm1, %ymm2
; AVX512F-NEXT: vextracti64x4 $1, %zmm0, %ymm3
; AVX512F-NEXT: vpavgw %ymm2, %ymm3, %ymm2
; AVX512F-NEXT: vpavgw %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: vinserti64x4 $1, %ymm2, %zmm0, %zmm0
; AVX512F-NEXT: kmovw %edi, %k2
; AVX512F-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 {%k1} {z}
; AVX512F-NEXT: vpmovdw %zmm1, %ymm1
; AVX512F-NEXT: vpternlogd $255, %zmm2, %zmm2, %zmm2 {%k2} {z}
; AVX512F-NEXT: vpmovdw %zmm2, %ymm2
; AVX512F-NEXT: vinserti64x4 $1, %ymm2, %zmm1, %zmm1
; AVX512F-NEXT: vpandq %zmm0, %zmm1, %zmm0
; AVX512F-NEXT: retq
;
; AVX512BWVL-LABEL: avg_v32i16_maskz:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: kmovd %edi, %k1
; AVX512BWVL-NEXT: vpavgw %zmm1, %zmm0, %zmm0 {%k1} {z}
; AVX512BWVL-NEXT: retq
%za = zext <32 x i16> %a to <32 x i32>
%zb = zext <32 x i16> %b to <32 x i32>
%add = add nuw nsw <32 x i32> %za, %zb
%add1 = add nuw nsw <32 x i32> %add, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%lshr = lshr <32 x i32> %add1, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
%trunc = trunc <32 x i32> %lshr to <32 x i16>
%mask1 = bitcast i32 %mask to <32 x i1>
%res = select <32 x i1> %mask1, <32 x i16> %trunc, <32 x i16> zeroinitializer
ret <32 x i16> %res
}