mve-gather-ind16-unscaled.ll
9.77 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
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=thumbv8.1m.main-arm-none-eabi -mattr=+mve.fp -enable-arm-maskedgatscat %s -o - | FileCheck %s
define arm_aapcs_vfpcc <8 x i16> @zext_unscaled_i8_i16(i8* %base, <8 x i16>* %offptr) {
; CHECK-LABEL: zext_unscaled_i8_i16:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r7, lr}
; CHECK-NEXT: push {r4, r5, r7, lr}
; CHECK-NEXT: vldrh.u32 q0, [r1]
; CHECK-NEXT: vldrh.u32 q1, [r1, #8]
; CHECK-NEXT: vadd.i32 q0, q0, r0
; CHECK-NEXT: vadd.i32 q1, q1, r0
; CHECK-NEXT: vmov r2, s3
; CHECK-NEXT: vmov r3, s1
; CHECK-NEXT: vmov r5, s0
; CHECK-NEXT: vmov r0, s4
; CHECK-NEXT: vmov r1, s5
; CHECK-NEXT: vmov r4, s7
; CHECK-NEXT: ldrb.w r12, [r2]
; CHECK-NEXT: vmov r2, s2
; CHECK-NEXT: ldrb.w lr, [r3]
; CHECK-NEXT: vmov r3, s6
; CHECK-NEXT: ldrb r5, [r5]
; CHECK-NEXT: ldrb r0, [r0]
; CHECK-NEXT: vmov.16 q0[0], r5
; CHECK-NEXT: ldrb r1, [r1]
; CHECK-NEXT: vmov.16 q0[1], lr
; CHECK-NEXT: ldrb r4, [r4]
; CHECK-NEXT: ldrb r2, [r2]
; CHECK-NEXT: ldrb r3, [r3]
; CHECK-NEXT: vmov.16 q0[2], r2
; CHECK-NEXT: vmov.16 q0[3], r12
; CHECK-NEXT: vmov.16 q0[4], r0
; CHECK-NEXT: vmov.16 q0[5], r1
; CHECK-NEXT: vmov.16 q0[6], r3
; CHECK-NEXT: vmov.16 q0[7], r4
; CHECK-NEXT: vmovlb.u8 q0, q0
; CHECK-NEXT: pop {r4, r5, r7, pc}
entry:
%offs = load <8 x i16>, <8 x i16>* %offptr, align 2
%offs.zext = zext <8 x i16> %offs to <8 x i32>
%ptrs = getelementptr inbounds i8, i8* %base, <8 x i32> %offs.zext
%gather = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> %ptrs, i32 1, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x i8> undef)
%gather.zext = zext <8 x i8> %gather to <8 x i16>
ret <8 x i16> %gather.zext
}
define arm_aapcs_vfpcc <8 x i16> @sext_unscaled_i8_i16(i8* %base, <8 x i16>* %offptr) {
; CHECK-LABEL: sext_unscaled_i8_i16:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r7, lr}
; CHECK-NEXT: push {r4, r5, r7, lr}
; CHECK-NEXT: vldrh.u32 q0, [r1]
; CHECK-NEXT: vldrh.u32 q1, [r1, #8]
; CHECK-NEXT: vadd.i32 q0, q0, r0
; CHECK-NEXT: vadd.i32 q1, q1, r0
; CHECK-NEXT: vmov r2, s3
; CHECK-NEXT: vmov r3, s1
; CHECK-NEXT: vmov r5, s0
; CHECK-NEXT: vmov r0, s4
; CHECK-NEXT: vmov r1, s5
; CHECK-NEXT: vmov r4, s7
; CHECK-NEXT: ldrb.w r12, [r2]
; CHECK-NEXT: vmov r2, s2
; CHECK-NEXT: ldrb.w lr, [r3]
; CHECK-NEXT: vmov r3, s6
; CHECK-NEXT: ldrb r5, [r5]
; CHECK-NEXT: ldrb r0, [r0]
; CHECK-NEXT: vmov.16 q0[0], r5
; CHECK-NEXT: ldrb r1, [r1]
; CHECK-NEXT: vmov.16 q0[1], lr
; CHECK-NEXT: ldrb r4, [r4]
; CHECK-NEXT: ldrb r2, [r2]
; CHECK-NEXT: ldrb r3, [r3]
; CHECK-NEXT: vmov.16 q0[2], r2
; CHECK-NEXT: vmov.16 q0[3], r12
; CHECK-NEXT: vmov.16 q0[4], r0
; CHECK-NEXT: vmov.16 q0[5], r1
; CHECK-NEXT: vmov.16 q0[6], r3
; CHECK-NEXT: vmov.16 q0[7], r4
; CHECK-NEXT: vmovlb.s8 q0, q0
; CHECK-NEXT: pop {r4, r5, r7, pc}
entry:
%offs = load <8 x i16>, <8 x i16>* %offptr, align 2
%offs.zext = zext <8 x i16> %offs to <8 x i32>
%ptrs = getelementptr inbounds i8, i8* %base, <8 x i32> %offs.zext
%gather = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> %ptrs, i32 1, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x i8> undef)
%gather.sext = sext <8 x i8> %gather to <8 x i16>
ret <8 x i16> %gather.sext
}
define arm_aapcs_vfpcc <8 x i16> @unscaled_i16_i16(i8* %base, <8 x i16>* %offptr) {
; CHECK-LABEL: unscaled_i16_i16:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vldrh.u16 q1, [r1]
; CHECK-NEXT: vldrh.u16 q0, [r0, q1]
; CHECK-NEXT: bx lr
entry:
%offs = load <8 x i16>, <8 x i16>* %offptr, align 2
%offs.zext = zext <8 x i16> %offs to <8 x i32>
%byte_ptrs = getelementptr inbounds i8, i8* %base, <8 x i32> %offs.zext
%ptrs = bitcast <8 x i8*> %byte_ptrs to <8 x i16*>
%gather = call <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*> %ptrs, i32 2, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x i16> undef)
ret <8 x i16> %gather
}
define arm_aapcs_vfpcc <8 x half> @unscaled_f16_i16(i8* %base, <8 x i16>* %offptr) {
; CHECK-LABEL: unscaled_f16_i16:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vldrh.u16 q1, [r1]
; CHECK-NEXT: vldrh.u16 q0, [r0, q1]
; CHECK-NEXT: bx lr
entry:
%offs = load <8 x i16>, <8 x i16>* %offptr, align 2
%offs.zext = zext <8 x i16> %offs to <8 x i32>
%byte_ptrs = getelementptr inbounds i8, i8* %base, <8 x i32> %offs.zext
%ptrs = bitcast <8 x i8*> %byte_ptrs to <8 x half*>
%gather = call <8 x half> @llvm.masked.gather.v8f16.v8p0f16(<8 x half*> %ptrs, i32 2, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x half> undef)
ret <8 x half> %gather
}
define arm_aapcs_vfpcc <8 x i16> @zext_unsigned_unscaled_i8_i8(i8* %base, <8 x i8>* %offptr) {
; CHECK-LABEL: zext_unsigned_unscaled_i8_i8:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r7, lr}
; CHECK-NEXT: push {r4, r5, r7, lr}
; CHECK-NEXT: vldrb.u32 q0, [r1]
; CHECK-NEXT: vldrb.u32 q1, [r1, #4]
; CHECK-NEXT: vadd.i32 q0, q0, r0
; CHECK-NEXT: vadd.i32 q1, q1, r0
; CHECK-NEXT: vmov r2, s3
; CHECK-NEXT: vmov r3, s1
; CHECK-NEXT: vmov r5, s0
; CHECK-NEXT: vmov r0, s4
; CHECK-NEXT: vmov r1, s5
; CHECK-NEXT: vmov r4, s7
; CHECK-NEXT: ldrb.w r12, [r2]
; CHECK-NEXT: vmov r2, s2
; CHECK-NEXT: ldrb.w lr, [r3]
; CHECK-NEXT: vmov r3, s6
; CHECK-NEXT: ldrb r5, [r5]
; CHECK-NEXT: ldrb r0, [r0]
; CHECK-NEXT: vmov.16 q0[0], r5
; CHECK-NEXT: ldrb r1, [r1]
; CHECK-NEXT: vmov.16 q0[1], lr
; CHECK-NEXT: ldrb r4, [r4]
; CHECK-NEXT: ldrb r2, [r2]
; CHECK-NEXT: ldrb r3, [r3]
; CHECK-NEXT: vmov.16 q0[2], r2
; CHECK-NEXT: vmov.16 q0[3], r12
; CHECK-NEXT: vmov.16 q0[4], r0
; CHECK-NEXT: vmov.16 q0[5], r1
; CHECK-NEXT: vmov.16 q0[6], r3
; CHECK-NEXT: vmov.16 q0[7], r4
; CHECK-NEXT: vmovlb.u8 q0, q0
; CHECK-NEXT: pop {r4, r5, r7, pc}
entry:
%offs = load <8 x i8>, <8 x i8>* %offptr, align 1
%offs.zext = zext <8 x i8> %offs to <8 x i32>
%ptrs = getelementptr inbounds i8, i8* %base, <8 x i32> %offs.zext
%gather = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> %ptrs, i32 1, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x i8> undef)
%gather.zext = zext <8 x i8> %gather to <8 x i16>
ret <8 x i16> %gather.zext
}
define arm_aapcs_vfpcc <8 x i16> @sext_unsigned_unscaled_i8_i8(i8* %base, <8 x i8>* %offptr) {
; CHECK-LABEL: sext_unsigned_unscaled_i8_i8:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r7, lr}
; CHECK-NEXT: push {r4, r5, r7, lr}
; CHECK-NEXT: vldrb.u32 q0, [r1]
; CHECK-NEXT: vldrb.u32 q1, [r1, #4]
; CHECK-NEXT: vadd.i32 q0, q0, r0
; CHECK-NEXT: vadd.i32 q1, q1, r0
; CHECK-NEXT: vmov r2, s3
; CHECK-NEXT: vmov r3, s1
; CHECK-NEXT: vmov r5, s0
; CHECK-NEXT: vmov r0, s4
; CHECK-NEXT: vmov r1, s5
; CHECK-NEXT: vmov r4, s7
; CHECK-NEXT: ldrb.w r12, [r2]
; CHECK-NEXT: vmov r2, s2
; CHECK-NEXT: ldrb.w lr, [r3]
; CHECK-NEXT: vmov r3, s6
; CHECK-NEXT: ldrb r5, [r5]
; CHECK-NEXT: ldrb r0, [r0]
; CHECK-NEXT: vmov.16 q0[0], r5
; CHECK-NEXT: ldrb r1, [r1]
; CHECK-NEXT: vmov.16 q0[1], lr
; CHECK-NEXT: ldrb r4, [r4]
; CHECK-NEXT: ldrb r2, [r2]
; CHECK-NEXT: ldrb r3, [r3]
; CHECK-NEXT: vmov.16 q0[2], r2
; CHECK-NEXT: vmov.16 q0[3], r12
; CHECK-NEXT: vmov.16 q0[4], r0
; CHECK-NEXT: vmov.16 q0[5], r1
; CHECK-NEXT: vmov.16 q0[6], r3
; CHECK-NEXT: vmov.16 q0[7], r4
; CHECK-NEXT: vmovlb.s8 q0, q0
; CHECK-NEXT: pop {r4, r5, r7, pc}
entry:
%offs = load <8 x i8>, <8 x i8>* %offptr, align 1
%offs.zext = zext <8 x i8> %offs to <8 x i32>
%ptrs = getelementptr inbounds i8, i8* %base, <8 x i32> %offs.zext
%gather = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> %ptrs, i32 1, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x i8> undef)
%gather.sext = sext <8 x i8> %gather to <8 x i16>
ret <8 x i16> %gather.sext
}
define arm_aapcs_vfpcc <8 x i16> @unsigned_unscaled_i16_i8(i8* %base, <8 x i8>* %offptr) {
; CHECK-LABEL: unsigned_unscaled_i16_i8:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vldrb.u16 q1, [r1]
; CHECK-NEXT: vldrh.u16 q0, [r0, q1]
; CHECK-NEXT: bx lr
entry:
%offs = load <8 x i8>, <8 x i8>* %offptr, align 1
%offs.zext = zext <8 x i8> %offs to <8 x i32>
%byte_ptrs = getelementptr inbounds i8, i8* %base, <8 x i32> %offs.zext
%ptrs = bitcast <8 x i8*> %byte_ptrs to <8 x i16*>
%gather = call <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*> %ptrs, i32 2, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x i16> undef)
ret <8 x i16> %gather
}
define arm_aapcs_vfpcc <8 x half> @unsigned_unscaled_f16_i8(i8* %base, <8 x i8>* %offptr) {
; CHECK-LABEL: unsigned_unscaled_f16_i8:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vldrb.u16 q1, [r1]
; CHECK-NEXT: vldrh.u16 q0, [r0, q1]
; CHECK-NEXT: bx lr
entry:
%offs = load <8 x i8>, <8 x i8>* %offptr, align 1
%offs.zext = zext <8 x i8> %offs to <8 x i32>
%byte_ptrs = getelementptr inbounds i8, i8* %base, <8 x i32> %offs.zext
%ptrs = bitcast <8 x i8*> %byte_ptrs to <8 x half*>
%gather = call <8 x half> @llvm.masked.gather.v8f16.v8p0f16(<8 x half*> %ptrs, i32 2, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x half> undef)
ret <8 x half> %gather
}
declare <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*>, i32, <8 x i1>, <8 x i8>) #1
declare <8 x i16> @llvm.masked.gather.v8i16.v8p0i16(<8 x i16*>, i32, <8 x i1>, <8 x i16>) #1
declare <8 x half> @llvm.masked.gather.v8f16.v8p0f16(<8 x half*>, i32, <8 x i1>, <8 x half>) #1