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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-apple-darwin -mattr=+sse4.1 | FileCheck %s --check-prefix=SSE
; RUN: llc < %s -mtriple=i686-apple-darwin -mattr=+avx2 | FileCheck %s --check-prefix=AVX
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=+sse4.1 | FileCheck %s --check-prefix=SSE
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=+avx2 | FileCheck %s --check-prefix=AVX
define <4 x float> @test_unpacklo_hadd_v4f32(<4 x float> %0, <4 x float> %1, <4 x float> %2, <4 x float> %3) {
; SSE-LABEL: test_unpacklo_hadd_v4f32:
; SSE: ## %bb.0:
; SSE-NEXT: haddps %xmm2, %xmm0
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: test_unpacklo_hadd_v4f32:
; AVX: ## %bb.0:
; AVX-NEXT: vhaddps %xmm2, %xmm0, %xmm0
; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX-NEXT: ret{{[l|q]}}
%5 = tail call <4 x float> @llvm.x86.sse3.hadd.ps(<4 x float> %0, <4 x float> %1) #4
%6 = tail call <4 x float> @llvm.x86.sse3.hadd.ps(<4 x float> %2, <4 x float> %3) #4
%7 = shufflevector <4 x float> %5, <4 x float> %6, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
ret <4 x float> %7
}
define <4 x float> @test_unpackhi_hadd_v4f32(<4 x float> %0, <4 x float> %1, <4 x float> %2, <4 x float> %3) {
; SSE-LABEL: test_unpackhi_hadd_v4f32:
; SSE: ## %bb.0:
; SSE-NEXT: movaps %xmm1, %xmm0
; SSE-NEXT: haddps %xmm3, %xmm0
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: test_unpackhi_hadd_v4f32:
; AVX: ## %bb.0:
; AVX-NEXT: vhaddps %xmm3, %xmm1, %xmm0
; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX-NEXT: ret{{[l|q]}}
%5 = tail call <4 x float> @llvm.x86.sse3.hadd.ps(<4 x float> %0, <4 x float> %1) #4
%6 = tail call <4 x float> @llvm.x86.sse3.hadd.ps(<4 x float> %2, <4 x float> %3) #4
%7 = shufflevector <4 x float> %5, <4 x float> %6, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
ret <4 x float> %7
}
define <4 x float> @test_unpacklo_hsub_v4f32(<4 x float> %0, <4 x float> %1, <4 x float> %2, <4 x float> %3) {
; SSE-LABEL: test_unpacklo_hsub_v4f32:
; SSE: ## %bb.0:
; SSE-NEXT: hsubps %xmm2, %xmm0
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: test_unpacklo_hsub_v4f32:
; AVX: ## %bb.0:
; AVX-NEXT: vhsubps %xmm2, %xmm0, %xmm0
; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX-NEXT: ret{{[l|q]}}
%5 = tail call <4 x float> @llvm.x86.sse3.hsub.ps(<4 x float> %0, <4 x float> %1) #4
%6 = tail call <4 x float> @llvm.x86.sse3.hsub.ps(<4 x float> %2, <4 x float> %3) #4
%7 = shufflevector <4 x float> %5, <4 x float> %6, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
ret <4 x float> %7
}
define <4 x float> @test_unpackhi_hsub_v4f32(<4 x float> %0, <4 x float> %1, <4 x float> %2, <4 x float> %3) {
; SSE-LABEL: test_unpackhi_hsub_v4f32:
; SSE: ## %bb.0:
; SSE-NEXT: movaps %xmm1, %xmm0
; SSE-NEXT: hsubps %xmm3, %xmm0
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: test_unpackhi_hsub_v4f32:
; AVX: ## %bb.0:
; AVX-NEXT: vhsubps %xmm3, %xmm1, %xmm0
; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX-NEXT: ret{{[l|q]}}
%5 = tail call <4 x float> @llvm.x86.sse3.hsub.ps(<4 x float> %0, <4 x float> %1) #4
%6 = tail call <4 x float> @llvm.x86.sse3.hsub.ps(<4 x float> %2, <4 x float> %3) #4
%7 = shufflevector <4 x float> %5, <4 x float> %6, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
ret <4 x float> %7
}
define <4 x i32> @test_unpacklo_hadd_v4i32(<4 x i32> %0, <4 x i32> %1, <4 x i32> %2, <4 x i32> %3) {
; SSE-LABEL: test_unpacklo_hadd_v4i32:
; SSE: ## %bb.0:
; SSE-NEXT: phaddd %xmm2, %xmm0
; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: test_unpacklo_hadd_v4i32:
; AVX: ## %bb.0:
; AVX-NEXT: vphaddd %xmm2, %xmm0, %xmm0
; AVX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX-NEXT: ret{{[l|q]}}
%5 = tail call <4 x i32> @llvm.x86.ssse3.phadd.d.128(<4 x i32> %0, <4 x i32> %1) #5
%6 = tail call <4 x i32> @llvm.x86.ssse3.phadd.d.128(<4 x i32> %2, <4 x i32> %3) #5
%7 = shufflevector <4 x i32> %5, <4 x i32> %6, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
ret <4 x i32> %7
}
define <4 x i32> @test_unpackhi_hadd_v4i32(<4 x i32> %0, <4 x i32> %1, <4 x i32> %2, <4 x i32> %3) {
; SSE-LABEL: test_unpackhi_hadd_v4i32:
; SSE: ## %bb.0:
; SSE-NEXT: phaddd %xmm3, %xmm1
; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm1[0,2,1,3]
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: test_unpackhi_hadd_v4i32:
; AVX: ## %bb.0:
; AVX-NEXT: vphaddd %xmm3, %xmm1, %xmm0
; AVX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX-NEXT: ret{{[l|q]}}
%5 = tail call <4 x i32> @llvm.x86.ssse3.phadd.d.128(<4 x i32> %0, <4 x i32> %1) #5
%6 = tail call <4 x i32> @llvm.x86.ssse3.phadd.d.128(<4 x i32> %2, <4 x i32> %3) #5
%7 = shufflevector <4 x i32> %5, <4 x i32> %6, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
ret <4 x i32> %7
}
define <4 x i32> @test_unpacklo_hsub_v4i32(<4 x i32> %0, <4 x i32> %1, <4 x i32> %2, <4 x i32> %3) {
; SSE-LABEL: test_unpacklo_hsub_v4i32:
; SSE: ## %bb.0:
; SSE-NEXT: phsubd %xmm2, %xmm0
; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: test_unpacklo_hsub_v4i32:
; AVX: ## %bb.0:
; AVX-NEXT: vphsubd %xmm2, %xmm0, %xmm0
; AVX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX-NEXT: ret{{[l|q]}}
%5 = tail call <4 x i32> @llvm.x86.ssse3.phsub.d.128(<4 x i32> %0, <4 x i32> %1) #5
%6 = tail call <4 x i32> @llvm.x86.ssse3.phsub.d.128(<4 x i32> %2, <4 x i32> %3) #5
%7 = shufflevector <4 x i32> %5, <4 x i32> %6, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
ret <4 x i32> %7
}
define <4 x i32> @test_unpackhi_hsub_v4i32(<4 x i32> %0, <4 x i32> %1, <4 x i32> %2, <4 x i32> %3) {
; SSE-LABEL: test_unpackhi_hsub_v4i32:
; SSE: ## %bb.0:
; SSE-NEXT: phsubd %xmm3, %xmm1
; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm1[0,2,1,3]
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: test_unpackhi_hsub_v4i32:
; AVX: ## %bb.0:
; AVX-NEXT: vphsubd %xmm3, %xmm1, %xmm0
; AVX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX-NEXT: ret{{[l|q]}}
%5 = tail call <4 x i32> @llvm.x86.ssse3.phsub.d.128(<4 x i32> %0, <4 x i32> %1) #5
%6 = tail call <4 x i32> @llvm.x86.ssse3.phsub.d.128(<4 x i32> %2, <4 x i32> %3) #5
%7 = shufflevector <4 x i32> %5, <4 x i32> %6, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
ret <4 x i32> %7
}
;
; Special Case
;
define <4 x float> @test_unpacklo_hadd_v4f32_unary(<4 x float> %0) {
; SSE-LABEL: test_unpacklo_hadd_v4f32_unary:
; SSE: ## %bb.0:
; SSE-NEXT: haddps %xmm0, %xmm0
; SSE-NEXT: unpcklps {{.*#+}} xmm0 = xmm0[0,0,1,1]
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: test_unpacklo_hadd_v4f32_unary:
; AVX: ## %bb.0:
; AVX-NEXT: vhaddps %xmm0, %xmm0, %xmm0
; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,0,1,1]
; AVX-NEXT: ret{{[l|q]}}
%2 = tail call <4 x float> @llvm.x86.sse3.hadd.ps(<4 x float> %0, <4 x float> %0) #4
%3 = shufflevector <4 x float> %2, <4 x float> %2, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
ret <4 x float> %3
}
define <8 x i16> @PR51974(<8 x i16> %a0) {
; SSE-LABEL: PR51974:
; SSE: ## %bb.0:
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: phaddw %xmm0, %xmm1
; SSE-NEXT: phaddw %xmm0, %xmm1
; SSE-NEXT: movdqa %xmm1, %xmm0
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: PR51974:
; AVX: ## %bb.0:
; AVX-NEXT: vphaddw %xmm0, %xmm0, %xmm1
; AVX-NEXT: vphaddw %xmm0, %xmm1, %xmm0
; AVX-NEXT: ret{{[l|q]}}
%r0 = call <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16> %a0, <8 x i16> %a0)
%r1 = call <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16> %r0, <8 x i16> %a0)
ret <8 x i16> %r1
}
define <8 x i16> @PR52040(<8 x i16> %a0) {
; SSE-LABEL: PR52040:
; SSE: ## %bb.0:
; SSE-NEXT: phaddw %xmm0, %xmm0
; SSE-NEXT: movdqa %xmm0, %xmm1
; SSE-NEXT: phaddw %xmm0, %xmm1
; SSE-NEXT: phaddw %xmm0, %xmm1
; SSE-NEXT: movdqa %xmm1, %xmm0
; SSE-NEXT: ret{{[l|q]}}
;
; AVX-LABEL: PR52040:
; AVX: ## %bb.0:
; AVX-NEXT: vphaddw %xmm0, %xmm0, %xmm0
; AVX-NEXT: vphaddw %xmm0, %xmm0, %xmm1
; AVX-NEXT: vphaddw %xmm0, %xmm1, %xmm0
; AVX-NEXT: ret{{[l|q]}}
%r1 = tail call <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16> %a0, <8 x i16> %a0)
%r2 = tail call <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16> %r1, <8 x i16> %r1)
%r3 = tail call <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16> %r2, <8 x i16> %r1)
ret <8 x i16> %r3
}
declare <4 x float> @llvm.x86.sse3.hadd.ps(<4 x float>, <4 x float>)
declare <4 x float> @llvm.x86.sse3.hsub.ps(<4 x float>, <4 x float>)
declare <2 x double> @llvm.x86.sse3.hadd.pd(<2 x double>, <2 x double>)
declare <2 x double> @llvm.x86.sse3.hsub.pd(<2 x double>, <2 x double>)
declare <8 x i16> @llvm.x86.ssse3.phadd.w.128(<8 x i16>, <8 x i16>)
declare <4 x i32> @llvm.x86.ssse3.phadd.d.128(<4 x i32>, <4 x i32>)
declare <8 x i16> @llvm.x86.ssse3.phsub.w.128(<8 x i16>, <8 x i16>)
declare <4 x i32> @llvm.x86.ssse3.phsub.d.128(<4 x i32>, <4 x i32>)
declare <16 x i8> @llvm.x86.sse2.packsswb.128(<8 x i16>, <8 x i16>)
declare <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32>, <4 x i32>)
declare <16 x i8> @llvm.x86.sse2.packuswb.128(<8 x i16>, <8 x i16>)
declare <8 x i16> @llvm.x86.sse41.packusdw(<4 x i32>, <4 x i32>)
|
