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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=sse2 | FileCheck %s --check-prefixes=SSE,SSE2
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=sse4.1 | FileCheck %s --check-prefixes=SSE,SSE41
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=avx2 | FileCheck %s --check-prefixes=AVX,AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=avx512f | FileCheck %s --check-prefixes=AVX,AVX512
; Verify that we don't scalarize a packed vector shift left of 16-bit
; signed integers if the amount is a constant build_vector.
; Check that we produce a SSE2 packed integer multiply (pmullw) instead.
define <8 x i16> @test1(<8 x i16> %a) {
; SSE-LABEL: test1:
; SSE: # %bb.0:
; SSE-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: test1:
; AVX: # %bb.0:
; AVX-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX-NEXT: retq
%shl = shl <8 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
ret <8 x i16> %shl
}
define <8 x i16> @test2(<8 x i16> %a) {
; SSE-LABEL: test2:
; SSE: # %bb.0:
; SSE-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: test2:
; AVX: # %bb.0:
; AVX-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX-NEXT: retq
%shl = shl <8 x i16> %a, <i16 0, i16 undef, i16 0, i16 0, i16 1, i16 undef, i16 -1, i16 1>
ret <8 x i16> %shl
}
; Verify that a vector shift left of 32-bit signed integers is simply expanded
; into a SSE4.1 pmulld (instead of cvttps2dq + pmulld) if the vector of shift
; counts is a constant build_vector.
define <4 x i32> @test3(<4 x i32> %a) {
; SSE2-LABEL: test3:
; SSE2: # %bb.0:
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm0, %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE2-NEXT: pmuludq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE2-NEXT: retq
;
; SSE41-LABEL: test3:
; SSE41: # %bb.0:
; SSE41-NEXT: pmulld {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE41-NEXT: retq
;
; AVX-LABEL: test3:
; AVX: # %bb.0:
; AVX-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX-NEXT: retq
%shl = shl <4 x i32> %a, <i32 1, i32 -1, i32 2, i32 -3>
ret <4 x i32> %shl
}
define <4 x i32> @test4(<4 x i32> %a) {
; SSE2-LABEL: test4:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: pslld $1, %xmm1
; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
; SSE2-NEXT: retq
;
; SSE41-LABEL: test4:
; SSE41: # %bb.0:
; SSE41-NEXT: movdqa %xmm0, %xmm1
; SSE41-NEXT: pslld $1, %xmm1
; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
; SSE41-NEXT: retq
;
; AVX-LABEL: test4:
; AVX: # %bb.0:
; AVX-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX-NEXT: retq
%shl = shl <4 x i32> %a, <i32 0, i32 0, i32 1, i32 1>
ret <4 x i32> %shl
}
; If we have AVX/SSE2 but not AVX2, verify that the following shift is split
; into two pmullw instructions. With AVX2, the test case below would produce
; a single vpmullw.
define <16 x i16> @test5(<16 x i16> %a) {
; SSE-LABEL: test5:
; SSE: # %bb.0:
; SSE-NEXT: movdqa {{.*#+}} xmm2 = [2,2,4,8,128,1,512,2048]
; SSE-NEXT: pmullw %xmm2, %xmm0
; SSE-NEXT: pmullw %xmm2, %xmm1
; SSE-NEXT: retq
;
; AVX-LABEL: test5:
; AVX: # %bb.0:
; AVX-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
; AVX-NEXT: retq
%shl = shl <16 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
ret <16 x i16> %shl
}
; If we have AVX/SSE4.1 but not AVX2, verify that the following shift is split
; into two pmulld instructions. With AVX2, the test case below would produce
; a single vpsllvd instead.
define <8 x i32> @test6(<8 x i32> %a) {
; SSE2-LABEL: test6:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [2,2,4,8]
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm2, %xmm0
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [2,2,8,8]
; SSE2-NEXT: pmuludq %xmm4, %xmm3
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
; SSE2-NEXT: pmuludq %xmm1, %xmm2
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm4, %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; SSE2-NEXT: movdqa %xmm2, %xmm1
; SSE2-NEXT: retq
;
; SSE41-LABEL: test6:
; SSE41: # %bb.0:
; SSE41-NEXT: movdqa {{.*#+}} xmm2 = [2,2,4,8]
; SSE41-NEXT: pmulld %xmm2, %xmm0
; SSE41-NEXT: pmulld %xmm2, %xmm1
; SSE41-NEXT: retq
;
; AVX-LABEL: test6:
; AVX: # %bb.0:
; AVX-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
; AVX-NEXT: retq
%shl = shl <8 x i32> %a, <i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3>
ret <8 x i32> %shl
}
; With AVX2 and AVX512, the test case below should produce a sequence of
; two vpmullw instructions. On SSE2 instead, we split the shift in four
; parts and then we convert each part into a pmullw.
define <32 x i16> @test7(<32 x i16> %a) {
; SSE-LABEL: test7:
; SSE: # %bb.0:
; SSE-NEXT: movdqa {{.*#+}} xmm4 = [2,2,4,8,128,1,512,2048]
; SSE-NEXT: pmullw %xmm4, %xmm0
; SSE-NEXT: pmullw %xmm4, %xmm1
; SSE-NEXT: pmullw %xmm4, %xmm2
; SSE-NEXT: pmullw %xmm4, %xmm3
; SSE-NEXT: retq
;
; AVX2-LABEL: test7:
; AVX2: # %bb.0:
; AVX2-NEXT: vbroadcasti128 {{.*#+}} ymm2 = [2,2,4,8,128,1,512,2048,2,2,4,8,128,1,512,2048]
; AVX2-NEXT: # ymm2 = mem[0,1,0,1]
; AVX2-NEXT: vpmullw %ymm2, %ymm0, %ymm0
; AVX2-NEXT: vpmullw %ymm2, %ymm1, %ymm1
; AVX2-NEXT: retq
;
; AVX512-LABEL: test7:
; AVX512: # %bb.0:
; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
; AVX512-NEXT: vbroadcasti128 {{.*#+}} ymm2 = [2,2,4,8,128,1,512,2048,2,2,4,8,128,1,512,2048]
; AVX512-NEXT: # ymm2 = mem[0,1,0,1]
; AVX512-NEXT: vpmullw %ymm2, %ymm1, %ymm1
; AVX512-NEXT: vpmullw %ymm2, %ymm0, %ymm0
; AVX512-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512-NEXT: retq
%shl = shl <32 x i16> %a, <i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11, i16 1, i16 1, i16 2, i16 3, i16 7, i16 0, i16 9, i16 11>
ret <32 x i16> %shl
}
; Similar to test7; the difference is that with AVX512 support
; we only produce a single vpsllvd/vpsllvq instead of a pair of vpsllvd/vpsllvq.
define <16 x i32> @test8(<16 x i32> %a) {
; SSE2-LABEL: test8:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [2,2,4,8]
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm0[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm4, %xmm0
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT: movdqa {{.*#+}} xmm6 = [2,2,8,8]
; SSE2-NEXT: pmuludq %xmm6, %xmm5
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1]
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm1[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm4, %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE2-NEXT: pmuludq %xmm6, %xmm5
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1]
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm2[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm4, %xmm2
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
; SSE2-NEXT: pmuludq %xmm6, %xmm5
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1]
; SSE2-NEXT: pmuludq %xmm3, %xmm4
; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[0,2,2,3]
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm6, %xmm3
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; SSE2-NEXT: movdqa %xmm4, %xmm3
; SSE2-NEXT: retq
;
; SSE41-LABEL: test8:
; SSE41: # %bb.0:
; SSE41-NEXT: movdqa {{.*#+}} xmm4 = [2,2,4,8]
; SSE41-NEXT: pmulld %xmm4, %xmm0
; SSE41-NEXT: pmulld %xmm4, %xmm1
; SSE41-NEXT: pmulld %xmm4, %xmm2
; SSE41-NEXT: pmulld %xmm4, %xmm3
; SSE41-NEXT: retq
;
; AVX2-LABEL: test8:
; AVX2: # %bb.0:
; AVX2-NEXT: vbroadcasti128 {{.*#+}} ymm2 = [1,1,2,3,1,1,2,3]
; AVX2-NEXT: # ymm2 = mem[0,1,0,1]
; AVX2-NEXT: vpsllvd %ymm2, %ymm0, %ymm0
; AVX2-NEXT: vpsllvd %ymm2, %ymm1, %ymm1
; AVX2-NEXT: retq
;
; AVX512-LABEL: test8:
; AVX512: # %bb.0:
; AVX512-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
; AVX512-NEXT: retq
%shl = shl <16 x i32> %a, <i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3, i32 1, i32 1, i32 2, i32 3>
ret <16 x i32> %shl
}
; The shift from 'test9' gets shifted separately and blended if we don't have AVX2/AVX512f support.
define <8 x i64> @test9(<8 x i64> %a) {
; SSE2-LABEL: test9:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa %xmm1, %xmm4
; SSE2-NEXT: psllq $2, %xmm4
; SSE2-NEXT: psllq $3, %xmm1
; SSE2-NEXT: movsd {{.*#+}} xmm1 = xmm4[0],xmm1[1]
; SSE2-NEXT: movdqa %xmm3, %xmm4
; SSE2-NEXT: psllq $2, %xmm4
; SSE2-NEXT: psllq $3, %xmm3
; SSE2-NEXT: movsd {{.*#+}} xmm3 = xmm4[0],xmm3[1]
; SSE2-NEXT: paddq %xmm0, %xmm0
; SSE2-NEXT: paddq %xmm2, %xmm2
; SSE2-NEXT: retq
;
; SSE41-LABEL: test9:
; SSE41: # %bb.0:
; SSE41-NEXT: movdqa %xmm1, %xmm4
; SSE41-NEXT: psllq $3, %xmm4
; SSE41-NEXT: psllq $2, %xmm1
; SSE41-NEXT: pblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm4[4,5,6,7]
; SSE41-NEXT: movdqa %xmm3, %xmm4
; SSE41-NEXT: psllq $3, %xmm4
; SSE41-NEXT: psllq $2, %xmm3
; SSE41-NEXT: pblendw {{.*#+}} xmm3 = xmm3[0,1,2,3],xmm4[4,5,6,7]
; SSE41-NEXT: paddq %xmm0, %xmm0
; SSE41-NEXT: paddq %xmm2, %xmm2
; SSE41-NEXT: retq
;
; AVX2-LABEL: test9:
; AVX2: # %bb.0:
; AVX2-NEXT: vmovdqa {{.*#+}} ymm2 = [1,1,2,3]
; AVX2-NEXT: vpsllvq %ymm2, %ymm0, %ymm0
; AVX2-NEXT: vpsllvq %ymm2, %ymm1, %ymm1
; AVX2-NEXT: retq
;
; AVX512-LABEL: test9:
; AVX512: # %bb.0:
; AVX512-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
; AVX512-NEXT: retq
%shl = shl <8 x i64> %a, <i64 1, i64 1, i64 2, i64 3, i64 1, i64 1, i64 2, i64 3>
ret <8 x i64> %shl
}
|
