File: llvm.amdgcn.tensor.load.store.ll

package info (click to toggle)
llvm-toolchain-21 1%3A21.1.4-5
  • links: PTS, VCS
  • area: main
  • in suites: sid
  • size: 2,236,516 kB
  • sloc: cpp: 7,619,569; ansic: 1,433,956; asm: 1,058,748; python: 252,181; f90: 94,671; objc: 70,753; lisp: 42,813; pascal: 18,401; sh: 8,601; ml: 5,111; perl: 4,720; makefile: 3,585; awk: 3,523; javascript: 2,272; xml: 892; fortran: 770
file content (245 lines) | stat: -rw-r--r-- 12,129 bytes parent folder | download | duplicates (8) 
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
 
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx1250 < %s | FileCheck -check-prefixes=GFX1250,GFX1250-SDAG %s
; RUN: llc -global-isel=1 -mtriple=amdgcn -mcpu=gfx1250 < %s | FileCheck -check-prefixes=GFX1250,GFX1250-GISEL %s

declare void @llvm.amdgcn.tensor.load.to.lds(<4 x i32> %D0, <8 x i32> %D1, <4 x i32> %D2, <4 x i32> %D3, i32 %cpol)
declare void @llvm.amdgcn.tensor.load.to.lds.d2(<4 x i32> %D0, <8 x i32> %D1, i32 %cpol)
declare void @llvm.amdgcn.tensor.store.from.lds(<4 x i32> %D0, <8 x i32> %D1, <4 x i32> %D2, <4 x i32> %D3, i32 %cpol)
declare void @llvm.amdgcn.tensor.store.from.lds.d2(<4 x i32> %D0, <8 x i32> %D1, i32 %cpol)

define amdgpu_ps void @tensor_load_to_lds(<4 x i32> inreg %D0, <8 x i32> inreg %D1, <4 x i32> inreg %D2, <4 x i32> inreg %D3) {
; GFX1250-LABEL: tensor_load_to_lds:
; GFX1250:       ; %bb.0: ; %entry
; GFX1250-NEXT:    tensor_load_to_lds s[0:3], s[4:11], s[12:15], s[16:19]
; GFX1250-NEXT:    s_endpgm
entry:
  call void @llvm.amdgcn.tensor.load.to.lds(<4 x i32> %D0, <8 x i32> %D1, <4 x i32> %D2, <4 x i32> %D3, i32 0)
  ret void
}

define amdgpu_ps void @tensor_load_to_lds_vector(<4 x i32> %D0, <8 x i32> %D1, <4 x i32> %D2, <4 x i32> %D3) {
; GFX1250-SDAG-LABEL: tensor_load_to_lds_vector:
; GFX1250-SDAG:       ; %bb.0: ; %entry
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s0, v4
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s1, v5
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s2, v6
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s3, v7
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s4, v8
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s8, v0
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s9, v1
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s10, v2
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s11, v3
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s5, v9
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s6, v10
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s7, v11
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s12, v12
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s13, v13
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s14, v14
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s15, v15
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s16, v16
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s17, v17
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s18, v18
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s19, v19
; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-SDAG-NEXT:    tensor_load_to_lds s[8:11], s[0:7], s[12:15], s[16:19]
; GFX1250-SDAG-NEXT:    s_endpgm
;
; GFX1250-GISEL-LABEL: tensor_load_to_lds_vector:
; GFX1250-GISEL:       ; %bb.0: ; %entry
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s8, v0
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s9, v1
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s10, v2
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s11, v3
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s0, v4
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s1, v5
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s2, v6
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s3, v7
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s4, v8
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s5, v9
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s6, v10
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s7, v11
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s12, v12
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s13, v13
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s14, v14
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s15, v15
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s16, v16
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s17, v17
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s18, v18
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s19, v19
; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-GISEL-NEXT:    tensor_load_to_lds s[8:11], s[0:7], s[12:15], s[16:19]
; GFX1250-GISEL-NEXT:    s_endpgm
entry:
  call void @llvm.amdgcn.tensor.load.to.lds(<4 x i32> %D0, <8 x i32> %D1, <4 x i32> %D2, <4 x i32> %D3, i32 0)
  ret void
}


define amdgpu_ps void @tensor_load_to_lds_d2(<4 x i32> inreg %D0, <8 x i32> inreg %D1) {
; GFX1250-LABEL: tensor_load_to_lds_d2:
; GFX1250:       ; %bb.0: ; %entry
; GFX1250-NEXT:    tensor_load_to_lds s[0:3], s[4:11] th:TH_LOAD_BYPASS scope:SCOPE_SYS
; GFX1250-NEXT:    s_endpgm
entry:
  call void @llvm.amdgcn.tensor.load.to.lds.d2(<4 x i32> %D0, <8 x i32> %D1, i32 27)
  ret void
}

define amdgpu_ps void @tensor_load_to_lds_d2_vector(<4 x i32> %D0, <8 x i32> %D1) {
; GFX1250-SDAG-LABEL: tensor_load_to_lds_d2_vector:
; GFX1250-SDAG:       ; %bb.0: ; %entry
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s0, v4
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s8, v0
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s9, v1
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s10, v2
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s11, v3
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s1, v5
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s2, v6
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s3, v7
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s4, v8
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s5, v9
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s6, v10
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s7, v11
; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-SDAG-NEXT:    tensor_load_to_lds s[8:11], s[0:7] th:TH_LOAD_BYPASS scope:SCOPE_SYS
; GFX1250-SDAG-NEXT:    s_endpgm
;
; GFX1250-GISEL-LABEL: tensor_load_to_lds_d2_vector:
; GFX1250-GISEL:       ; %bb.0: ; %entry
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s8, v0
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s9, v1
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s10, v2
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s11, v3
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s0, v4
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s1, v5
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s2, v6
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s3, v7
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s4, v8
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s5, v9
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s6, v10
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s7, v11
; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-GISEL-NEXT:    tensor_load_to_lds s[8:11], s[0:7] th:TH_LOAD_BYPASS scope:SCOPE_SYS
; GFX1250-GISEL-NEXT:    s_endpgm
entry:
  call void @llvm.amdgcn.tensor.load.to.lds.d2(<4 x i32> %D0, <8 x i32> %D1, i32 27)
  ret void
}

define amdgpu_ps void @tensor_store_from_lds(<4 x i32> inreg %D0, <8 x i32> inreg %D1, <4 x i32> inreg %D2, <4 x i32> inreg %D3) {
; GFX1250-LABEL: tensor_store_from_lds:
; GFX1250:       ; %bb.0: ; %entry
; GFX1250-NEXT:    tensor_store_from_lds s[0:3], s[4:11], s[12:15], s[16:19] th:TH_STORE_NT_HT scope:SCOPE_DEV
; GFX1250-NEXT:    s_endpgm
entry:
  call void @llvm.amdgcn.tensor.store.from.lds(<4 x i32> %D0, <8 x i32> %D1, <4 x i32> %D2, <4 x i32> %D3, i32 22)
  ret void
}

define amdgpu_ps void @tensor_store_from_lds_vector(<4 x i32> %D0, <8 x i32> %D1, <4 x i32> %D2, <4 x i32> %D3) {
; GFX1250-SDAG-LABEL: tensor_store_from_lds_vector:
; GFX1250-SDAG:       ; %bb.0: ; %entry
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s0, v4
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s1, v5
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s2, v6
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s3, v7
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s4, v8
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s8, v0
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s9, v1
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s10, v2
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s11, v3
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s5, v9
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s6, v10
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s7, v11
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s12, v12
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s13, v13
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s14, v14
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s15, v15
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s16, v16
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s17, v17
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s18, v18
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s19, v19
; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-SDAG-NEXT:    tensor_store_from_lds s[8:11], s[0:7], s[12:15], s[16:19] th:TH_STORE_NT_HT scope:SCOPE_DEV
; GFX1250-SDAG-NEXT:    s_endpgm
;
; GFX1250-GISEL-LABEL: tensor_store_from_lds_vector:
; GFX1250-GISEL:       ; %bb.0: ; %entry
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s8, v0
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s9, v1
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s10, v2
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s11, v3
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s0, v4
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s1, v5
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s2, v6
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s3, v7
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s4, v8
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s5, v9
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s6, v10
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s7, v11
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s12, v12
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s13, v13
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s14, v14
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s15, v15
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s16, v16
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s17, v17
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s18, v18
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s19, v19
; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-GISEL-NEXT:    tensor_store_from_lds s[8:11], s[0:7], s[12:15], s[16:19] th:TH_STORE_NT_HT scope:SCOPE_DEV
; GFX1250-GISEL-NEXT:    s_endpgm
entry:
  call void @llvm.amdgcn.tensor.store.from.lds(<4 x i32> %D0, <8 x i32> %D1, <4 x i32> %D2, <4 x i32> %D3, i32 22)
  ret void
}

define amdgpu_ps void @tensor_store_from_lds_d2(<4 x i32> inreg %D0, <8 x i32> inreg %D1) {
; GFX1250-LABEL: tensor_store_from_lds_d2:
; GFX1250:       ; %bb.0: ; %entry
; GFX1250-NEXT:    tensor_store_from_lds s[0:3], s[4:11]
; GFX1250-NEXT:    s_endpgm
entry:
  call void @llvm.amdgcn.tensor.store.from.lds.d2(<4 x i32> %D0, <8 x i32> %D1, i32 0)
  ret void
}

define amdgpu_ps void @tensor_store_from_lds_d2_vector(<4 x i32> %D0, <8 x i32> %D1) {
; GFX1250-SDAG-LABEL: tensor_store_from_lds_d2_vector:
; GFX1250-SDAG:       ; %bb.0: ; %entry
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s0, v4
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s8, v0
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s9, v1
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s10, v2
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s11, v3
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s1, v5
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s2, v6
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s3, v7
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s4, v8
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s5, v9
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s6, v10
; GFX1250-SDAG-NEXT:    v_readfirstlane_b32 s7, v11
; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-SDAG-NEXT:    tensor_store_from_lds s[8:11], s[0:7]
; GFX1250-SDAG-NEXT:    s_endpgm
;
; GFX1250-GISEL-LABEL: tensor_store_from_lds_d2_vector:
; GFX1250-GISEL:       ; %bb.0: ; %entry
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s8, v0
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s9, v1
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s10, v2
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s11, v3
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s0, v4
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s1, v5
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s2, v6
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s3, v7
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s4, v8
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s5, v9
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s6, v10
; GFX1250-GISEL-NEXT:    v_readfirstlane_b32 s7, v11
; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-GISEL-NEXT:    tensor_store_from_lds s[8:11], s[0:7]
; GFX1250-GISEL-NEXT:    s_endpgm
entry:
  call void @llvm.amdgcn.tensor.store.from.lds.d2(<4 x i32> %D0, <8 x i32> %D1, i32 0)
  ret void
}