Newer
Older
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
; RUN: llc -show-mc-encoding -march=arm -mcpu=cortex-a8 -mattr=+neon < %s | FileCheck %s
; CHECK: vmul_8xi8
define <8 x i8> @vmul_8xi8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
%tmp1 = load <8 x i8>* %A
%tmp2 = load <8 x i8>* %B
; CHECK: vmul.i8 d16, d16, d17 @ encoding: [0xb1,0x09,0x40,0xf2]
%tmp3 = mul <8 x i8> %tmp1, %tmp2
ret <8 x i8> %tmp3
}
; CHECK: vmul_4xi16
define <4 x i16> @vmul_4xi16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
%tmp1 = load <4 x i16>* %A
%tmp2 = load <4 x i16>* %B
; CHECK: vmul.i16 d16, d16, d17 @ encoding: [0xb1,0x09,0x50,0xf2]
%tmp3 = mul <4 x i16> %tmp1, %tmp2
ret <4 x i16> %tmp3
}
; CHECK: vmul_2xi32
define <2 x i32> @vmul_2xi32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
%tmp1 = load <2 x i32>* %A
%tmp2 = load <2 x i32>* %B
; CHECK: vmul.i32 d16, d16, d17 @ encoding: [0xb1,0x09,0x60,0xf2]
%tmp3 = mul <2 x i32> %tmp1, %tmp2
ret <2 x i32> %tmp3
}
; CHECK: vmul_2xfloat
define <2 x float> @vmul_2xfloat(<2 x float>* %A, <2 x float>* %B) nounwind {
%tmp1 = load <2 x float>* %A
%tmp2 = load <2 x float>* %B
; CHECK: vmul.f32 d16, d16, d17 @ encoding: [0xb1,0x0d,0x40,0xf3]
%tmp3 = fmul <2 x float> %tmp1, %tmp2
ret <2 x float> %tmp3
}
; CHECK: vmul_16xi8
define <16 x i8> @vmul_16xi8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
%tmp1 = load <16 x i8>* %A
%tmp2 = load <16 x i8>* %B
; CHECK: vmul.i8 q8, q8, q9 @ encoding: [0xf2,0x09,0x40,0xf2]
%tmp3 = mul <16 x i8> %tmp1, %tmp2
ret <16 x i8> %tmp3
}
; CHECK: vmul_8xi16
define <8 x i16> @vmul_8xi16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
%tmp1 = load <8 x i16>* %A
%tmp2 = load <8 x i16>* %B
; CHECK: vmul.i16 q8, q8, q9 @ encoding: [0xf2,0x09,0x50,0xf2]
%tmp3 = mul <8 x i16> %tmp1, %tmp2
ret <8 x i16> %tmp3
}
; CHECK: vmul_4xi32
define <4 x i32> @vmul_4xi32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
%tmp1 = load <4 x i32>* %A
%tmp2 = load <4 x i32>* %B
; CHECK: vmul.i32 q8, q8, q9 @ encoding: [0xf2,0x09,0x60,0xf2]
%tmp3 = mul <4 x i32> %tmp1, %tmp2
ret <4 x i32> %tmp3
}
; CHECK: vmul_4xfloat
define <4 x float> @vmul_4xfloat(<4 x float>* %A, <4 x float>* %B) nounwind {
%tmp1 = load <4 x float>* %A
%tmp2 = load <4 x float>* %B
; CHECK: vmul.f32 q8, q8, q9 @ encoding: [0xf2,0x0d,0x40,0xf3]
%tmp3 = fmul <4 x float> %tmp1, %tmp2
ret <4 x float> %tmp3
}
declare <8 x i8> @llvm.arm.neon.vmulp.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
declare <16 x i8> @llvm.arm.neon.vmulp.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
; CHECK: vmulp_8xi8
define <8 x i8> @vmulp_8xi8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
%tmp1 = load <8 x i8>* %A
%tmp2 = load <8 x i8>* %B
; CHECK: vmul.p8 d16, d16, d17 @ encoding: [0xb1,0x09,0x40,0xf3]
%tmp3 = call <8 x i8> @llvm.arm.neon.vmulp.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
ret <8 x i8> %tmp3
}
; CHECK: vmulp_16xi8
define <16 x i8> @vmulp_16xi8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
%tmp1 = load <16 x i8>* %A
%tmp2 = load <16 x i8>* %B
; CHECK: vmul.p8 q8, q8, q9 @ encoding: [0xf2,0x09,0x40,0xf3]
%tmp3 = call <16 x i8> @llvm.arm.neon.vmulp.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
ret <16 x i8> %tmp3
}
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
declare <4 x i16> @llvm.arm.neon.vqdmulh.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
declare <2 x i32> @llvm.arm.neon.vqdmulh.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
; CHECK: vqdmulh_4xi16
define <4 x i16> @vqdmulh_4xi16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
%tmp1 = load <4 x i16>* %A
%tmp2 = load <4 x i16>* %B
; CHECK: vqdmulh.s16 d16, d16, d17 @ encoding: [0xa1,0x0b,0x50,0xf2]
%tmp3 = call <4 x i16> @llvm.arm.neon.vqdmulh.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
ret <4 x i16> %tmp3
}
; CHECK: vqdmulh_2xi32
define <2 x i32> @vqdmulh_2xi32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
%tmp1 = load <2 x i32>* %A
%tmp2 = load <2 x i32>* %B
; CHECK: vqdmulh.s32 d16, d16, d17 @ encoding: [0xa1,0x0b,0x60,0xf2]
%tmp3 = call <2 x i32> @llvm.arm.neon.vqdmulh.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
ret <2 x i32> %tmp3
}
declare <8 x i16> @llvm.arm.neon.vqdmulh.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
declare <4 x i32> @llvm.arm.neon.vqdmulh.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
; CHECK: vqdmulh_8xi16
define <8 x i16> @vqdmulh_8xi16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
%tmp1 = load <8 x i16>* %A
%tmp2 = load <8 x i16>* %B
; CHECK: vqdmulh.s16 q8, q8, q9 @ encoding: [0xe2,0x0b,0x50,0xf2]
%tmp3 = call <8 x i16> @llvm.arm.neon.vqdmulh.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
ret <8 x i16> %tmp3
}
; CHECK: vqdmulh_4xi32
define <4 x i32> @vqdmulh_4xi32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
%tmp1 = load <4 x i32>* %A
%tmp2 = load <4 x i32>* %B
; CHECK: vqdmulh.s32 q8, q8, q9 @ encoding: [0xe2,0x0b,0x60,0xf2]
%tmp3 = call <4 x i32> @llvm.arm.neon.vqdmulh.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
ret <4 x i32> %tmp3
}
declare <2 x i32> @llvm.arm.neon.vqrdmulh.v2i32(<2 x i32>, <2 x i32>) nounwind readnone
declare <4 x i16> @llvm.arm.neon.vqrdmulh.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
; CHECK: vqrdmulh_4xi16
define <4 x i16> @vqrdmulh_4xi16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
%tmp1 = load <4 x i16>* %A
%tmp2 = load <4 x i16>* %B
; CHECK: vqrdmulh.s16 d16, d16, d17 @ encoding: [0xa1,0x0b,0x50,0xf3]
%tmp3 = call <4 x i16> @llvm.arm.neon.vqrdmulh.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
ret <4 x i16> %tmp3
}
; CHECK: vqrdmulh_2xi32
define <2 x i32> @vqrdmulh_2xi32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
%tmp1 = load <2 x i32>* %A
%tmp2 = load <2 x i32>* %B
; CHECK: vqrdmulh.s32 d16, d16, d17 @ encoding: [0xa1,0x0b,0x60,0xf3]
%tmp3 = call <2 x i32> @llvm.arm.neon.vqrdmulh.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
ret <2 x i32> %tmp3
}
declare <8 x i16> @llvm.arm.neon.vqrdmulh.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
declare <4 x i32> @llvm.arm.neon.vqrdmulh.v4i32(<4 x i32>, <4 x i32>) nounwind readnone
; CHECK: vqrdmulh_8xi16
define <8 x i16> @vqrdmulh_8xi16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
%tmp1 = load <8 x i16>* %A
%tmp2 = load <8 x i16>* %B
; CHECK: vqrdmulh.s16 q8, q8, q9 @ encoding: [0xe2,0x0b,0x50,0xf3]
%tmp3 = call <8 x i16> @llvm.arm.neon.vqrdmulh.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
ret <8 x i16> %tmp3
}
; CHECK: vqrdmulh_4xi32
define <4 x i32> @vqrdmulh_4xi32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
%tmp1 = load <4 x i32>* %A
%tmp2 = load <4 x i32>* %B
; CHECK: vqrdmulh.s32 q8, q8, q9 @ encoding: [0xe2,0x0b,0x60,0xf3]
%tmp3 = call <4 x i32> @llvm.arm.neon.vqrdmulh.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
ret <4 x i32> %tmp3
}
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
; CHECK: vmulls_8xi16
define <8 x i16> @vmulls_8xi16(<8 x i8>* %A, <8 x i8>* %B) nounwind {
%tmp1 = load <8 x i8>* %A
%tmp2 = load <8 x i8>* %B
%tmp3 = sext <8 x i8> %tmp1 to <8 x i16>
%tmp4 = sext <8 x i8> %tmp2 to <8 x i16>
; CHECK: vmull.s8 q8, d16, d17 @ encoding: [0xa1,0x0c,0xc0,0xf2]
%tmp5 = mul <8 x i16> %tmp3, %tmp4
ret <8 x i16> %tmp5
}
; CHECK: vmulls_4xi16
define <4 x i32> @vmulls_4xi16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
%tmp1 = load <4 x i16>* %A
%tmp2 = load <4 x i16>* %B
%tmp3 = sext <4 x i16> %tmp1 to <4 x i32>
%tmp4 = sext <4 x i16> %tmp2 to <4 x i32>
; CHECK: vmull.s16 q8, d16, d17 @ encoding: [0xa1,0x0c,0xd0,0xf2]
%tmp5 = mul <4 x i32> %tmp3, %tmp4
ret <4 x i32> %tmp5
}
; CHECK: vmulls_2xi32
define <2 x i64> @vmulls_2xi32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
%tmp1 = load <2 x i32>* %A
%tmp2 = load <2 x i32>* %B
%tmp3 = sext <2 x i32> %tmp1 to <2 x i64>
%tmp4 = sext <2 x i32> %tmp2 to <2 x i64>
; CHECK: vmull.s32 q8, d16, d17 @ encoding: [0xa1,0x0c,0xe0,0xf2]
%tmp5 = mul <2 x i64> %tmp3, %tmp4
ret <2 x i64> %tmp5
}
; CHECK: vmullu_8xi8
define <8 x i16> @vmullu_8xi8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
%tmp1 = load <8 x i8>* %A
%tmp2 = load <8 x i8>* %B
%tmp3 = zext <8 x i8> %tmp1 to <8 x i16>
%tmp4 = zext <8 x i8> %tmp2 to <8 x i16>
; CHECK: vmull.u8 q8, d16, d17 @ encoding: [0xa1,0x0c,0xc0,0xf3]
%tmp5 = mul <8 x i16> %tmp3, %tmp4
ret <8 x i16> %tmp5
}
; CHECK: vmullu_4xi16
define <4 x i32> @vmullu_4xi16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
%tmp1 = load <4 x i16>* %A
%tmp2 = load <4 x i16>* %B
%tmp3 = zext <4 x i16> %tmp1 to <4 x i32>
%tmp4 = zext <4 x i16> %tmp2 to <4 x i32>
; CHECK: vmull.u16 q8, d16, d17 @ encoding: [0xa1,0x0c,0xd0,0xf3]
%tmp5 = mul <4 x i32> %tmp3, %tmp4
ret <4 x i32> %tmp5
}
; CHECK: vmullu_2xi32
define <2 x i64> @vmullu_2xi32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
%tmp1 = load <2 x i32>* %A
%tmp2 = load <2 x i32>* %B
%tmp3 = zext <2 x i32> %tmp1 to <2 x i64>
%tmp4 = zext <2 x i32> %tmp2 to <2 x i64>
; CHECK: vmull.u32 q8, d16, d17 @ encoding: [0xa1,0x0c,0xe0,0xf3]
%tmp5 = mul <2 x i64> %tmp3, %tmp4
ret <2 x i64> %tmp5
}
declare <8 x i16> @llvm.arm.neon.vmullp.v8i16(<8 x i8>, <8 x i8>) nounwind readnone
; CHECK: vmullp_8xi8
define <8 x i16> @vmullp_8xi8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
%tmp1 = load <8 x i8>* %A
%tmp2 = load <8 x i8>* %B
; CHECK: vmull.p8 q8, d16, d17 @ encoding: [0xa1,0x0e,0xc0,0xf2]
%tmp3 = call <8 x i16> @llvm.arm.neon.vmullp.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp2)
ret <8 x i16> %tmp3
}
declare <4 x i32> @llvm.arm.neon.vqdmull.v4i32(<4 x i16>, <4 x i16>) nounwind readnone
declare <2 x i64> @llvm.arm.neon.vqdmull.v2i64(<2 x i32>, <2 x i32>) nounwind readnone
; CHECK: vqdmull_4xi16
define <4 x i32> @vqdmull_4xi16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
%tmp1 = load <4 x i16>* %A
%tmp2 = load <4 x i16>* %B
; CHECK: vqdmull.s16 q8, d16, d17 @ encoding: [0xa1,0x0d,0xd0,0xf2]
%tmp3 = call <4 x i32> @llvm.arm.neon.vqdmull.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
ret <4 x i32> %tmp3
}
; CHECK: vqdmull_2xi32
define <2 x i64> @vqdmull_2xi32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
%tmp1 = load <2 x i32>* %A
%tmp2 = load <2 x i32>* %B
; CHECK: vqdmull.s32 q8, d16, d17 @ encoding: [0xa1,0x0d,0xe0,0xf2]
%tmp3 = call <2 x i64> @llvm.arm.neon.vqdmull.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
ret <2 x i64> %tmp3
}