File: buffer-atomic-fadd.v2f16-no-rtn.ll

package info (click to toggle)
llvm-toolchain-17 1%3A17.0.6-22
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid, trixie
  • size: 1,799,624 kB
  • sloc: cpp: 6,428,607; ansic: 1,383,196; asm: 793,408; python: 223,504; objc: 75,364; f90: 60,502; lisp: 33,869; pascal: 15,282; sh: 9,684; perl: 7,453; ml: 4,937; awk: 3,523; makefile: 2,889; javascript: 2,149; xml: 888; fortran: 619; cs: 573
file content (278 lines) | stat: -rw-r--r-- 20,953 bytes parent folder | download | duplicates (2) 
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
 
; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -global-isel -march=amdgcn -mcpu=gfx908 -verify-machineinstrs -stop-after=instruction-select < %s | FileCheck -check-prefix=GFX908 %s
; RUN: llc -global-isel -march=amdgcn -mcpu=gfx90a -verify-machineinstrs -stop-after=instruction-select < %s | FileCheck -check-prefix=GFX90A_GFX940 %s
; RUN: llc -global-isel -march=amdgcn -mcpu=gfx940 -verify-machineinstrs -stop-after=instruction-select < %s | FileCheck -check-prefix=GFX90A_GFX940 %s

define amdgpu_ps void @buffer_atomic_fadd_v2f16_offset_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
  ; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_offset_no_rtn
  ; GFX908: bb.1 (%ir-block.0):
  ; GFX908-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
  ; GFX908-NEXT: {{  $}}
  ; GFX908-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX908-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX908-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX908-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX908-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX908-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX908-NEXT:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX908-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
  ; GFX908-NEXT:   S_ENDPGM 0
  ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offset_no_rtn
  ; GFX90A_GFX940: bb.1 (%ir-block.0):
  ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
  ; GFX90A_GFX940-NEXT: {{  $}}
  ; GFX90A_GFX940-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX90A_GFX940-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX90A_GFX940-NEXT:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
  ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
  %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0)
  ret void
}

define amdgpu_ps void @buffer_atomic_fadd_v2f16_offen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
  ; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_offen_no_rtn
  ; GFX908: bb.1 (%ir-block.0):
  ; GFX908-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
  ; GFX908-NEXT: {{  $}}
  ; GFX908-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX908-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX908-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX908-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX908-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX908-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX908-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX908-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX908-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
  ; GFX908-NEXT:   S_ENDPGM 0
  ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offen_no_rtn
  ; GFX90A_GFX940: bb.1 (%ir-block.0):
  ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
  ; GFX90A_GFX940-NEXT: {{  $}}
  ; GFX90A_GFX940-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX90A_GFX940-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX90A_GFX940-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
  ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
  %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
  ret void
}

define amdgpu_ps void @buffer_atomic_fadd_v2f16_idxen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
  ; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_idxen_no_rtn
  ; GFX908: bb.1 (%ir-block.0):
  ; GFX908-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
  ; GFX908-NEXT: {{  $}}
  ; GFX908-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX908-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX908-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX908-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX908-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX908-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX908-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX908-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX908-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
  ; GFX908-NEXT:   S_ENDPGM 0
  ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_idxen_no_rtn
  ; GFX90A_GFX940: bb.1 (%ir-block.0):
  ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
  ; GFX90A_GFX940-NEXT: {{  $}}
  ; GFX90A_GFX940-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX90A_GFX940-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX90A_GFX940-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
  ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
  %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
  ret void
}

define amdgpu_ps void @buffer_atomic_fadd_v2f16_bothen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
  ; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_bothen_no_rtn
  ; GFX908: bb.1 (%ir-block.0):
  ; GFX908-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
  ; GFX908-NEXT: {{  $}}
  ; GFX908-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX908-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX908-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX908-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX908-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX908-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX908-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX908-NEXT:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; GFX908-NEXT:   [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX908-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
  ; GFX908-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
  ; GFX908-NEXT:   S_ENDPGM 0
  ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_bothen_no_rtn
  ; GFX90A_GFX940: bb.1 (%ir-block.0):
  ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
  ; GFX90A_GFX940-NEXT: {{  $}}
  ; GFX90A_GFX940-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX90A_GFX940-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX90A_GFX940-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; GFX90A_GFX940-NEXT:   [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
  ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
  ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
  %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
  ret void
}

define amdgpu_ps void @buffer_ptr_atomic_fadd_v2f16_offset_no_rtn(<2 x half> %val, ptr addrspace(8) inreg %rsrc, i32 inreg %soffset) {
  ; GFX908-LABEL: name: buffer_ptr_atomic_fadd_v2f16_offset_no_rtn
  ; GFX908: bb.1 (%ir-block.0):
  ; GFX908-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
  ; GFX908-NEXT: {{  $}}
  ; GFX908-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX908-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX908-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX908-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX908-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX908-NEXT:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX908-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX908-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>) on %ir.rsrc, align 1, addrspace 8)
  ; GFX908-NEXT:   S_ENDPGM 0
  ; GFX90A_GFX940-LABEL: name: buffer_ptr_atomic_fadd_v2f16_offset_no_rtn
  ; GFX90A_GFX940: bb.1 (%ir-block.0):
  ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
  ; GFX90A_GFX940-NEXT: {{  $}}
  ; GFX90A_GFX940-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX90A_GFX940-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX90A_GFX940-NEXT:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>) on %ir.rsrc, align 1, addrspace 8)
  ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
  %ret = call <2 x half> @llvm.amdgcn.raw.ptr.buffer.atomic.fadd.v2f16(<2 x half> %val, ptr addrspace(8) %rsrc, i32 4095, i32 %soffset, i32 0)
  ret void
}

define amdgpu_ps void @buffer_ptr_atomic_fadd_v2f16_offen_no_rtn(<2 x half> %val, ptr addrspace(8) inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
  ; GFX908-LABEL: name: buffer_ptr_atomic_fadd_v2f16_offen_no_rtn
  ; GFX908: bb.1 (%ir-block.0):
  ; GFX908-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
  ; GFX908-NEXT: {{  $}}
  ; GFX908-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX908-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX908-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX908-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX908-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX908-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX908-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX908-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX908-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>) on %ir.rsrc, align 1, addrspace 8)
  ; GFX908-NEXT:   S_ENDPGM 0
  ; GFX90A_GFX940-LABEL: name: buffer_ptr_atomic_fadd_v2f16_offen_no_rtn
  ; GFX90A_GFX940: bb.1 (%ir-block.0):
  ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
  ; GFX90A_GFX940-NEXT: {{  $}}
  ; GFX90A_GFX940-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX90A_GFX940-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX90A_GFX940-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>) on %ir.rsrc, align 1, addrspace 8)
  ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
  %ret = call <2 x half> @llvm.amdgcn.raw.ptr.buffer.atomic.fadd.v2f16(<2 x half> %val, ptr addrspace(8) %rsrc, i32 %voffset, i32 %soffset, i32 0)
  ret void
}

define amdgpu_ps void @buffer_ptr_atomic_fadd_v2f16_idxen_no_rtn(<2 x half> %val, ptr addrspace(8) inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
  ; GFX908-LABEL: name: buffer_ptr_atomic_fadd_v2f16_idxen_no_rtn
  ; GFX908: bb.1 (%ir-block.0):
  ; GFX908-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
  ; GFX908-NEXT: {{  $}}
  ; GFX908-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX908-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX908-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX908-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX908-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX908-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX908-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX908-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX908-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>) on %ir.rsrc, align 1, addrspace 8)
  ; GFX908-NEXT:   S_ENDPGM 0
  ; GFX90A_GFX940-LABEL: name: buffer_ptr_atomic_fadd_v2f16_idxen_no_rtn
  ; GFX90A_GFX940: bb.1 (%ir-block.0):
  ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
  ; GFX90A_GFX940-NEXT: {{  $}}
  ; GFX90A_GFX940-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX90A_GFX940-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX90A_GFX940-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>) on %ir.rsrc, align 1, addrspace 8)
  ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
  %ret = call <2 x half> @llvm.amdgcn.struct.ptr.buffer.atomic.fadd.v2f16(<2 x half> %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
  ret void
}

define amdgpu_ps void @buffer_ptr_atomic_fadd_v2f16_bothen_no_rtn(<2 x half> %val, ptr addrspace(8) inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
  ; GFX908-LABEL: name: buffer_ptr_atomic_fadd_v2f16_bothen_no_rtn
  ; GFX908: bb.1 (%ir-block.0):
  ; GFX908-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
  ; GFX908-NEXT: {{  $}}
  ; GFX908-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX908-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX908-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX908-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX908-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX908-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX908-NEXT:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; GFX908-NEXT:   [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX908-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX908-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
  ; GFX908-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (<2 x s16>) on %ir.rsrc, align 1, addrspace 8)
  ; GFX908-NEXT:   S_ENDPGM 0
  ; GFX90A_GFX940-LABEL: name: buffer_ptr_atomic_fadd_v2f16_bothen_no_rtn
  ; GFX90A_GFX940: bb.1 (%ir-block.0):
  ; GFX90A_GFX940-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
  ; GFX90A_GFX940-NEXT: {{  $}}
  ; GFX90A_GFX940-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
  ; GFX90A_GFX940-NEXT:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; GFX90A_GFX940-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; GFX90A_GFX940-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; GFX90A_GFX940-NEXT:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; GFX90A_GFX940-NEXT:   [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; GFX90A_GFX940-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
  ; GFX90A_GFX940-NEXT:   BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (<2 x s16>) on %ir.rsrc, align 1, addrspace 8)
  ; GFX90A_GFX940-NEXT:   S_ENDPGM 0
  %ret = call <2 x half> @llvm.amdgcn.struct.ptr.buffer.atomic.fadd.v2f16(<2 x half> %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
  ret void
}

declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg)
declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg)

declare <2 x half> @llvm.amdgcn.raw.ptr.buffer.atomic.fadd.v2f16(<2 x half>, ptr addrspace(8), i32, i32, i32 immarg)
declare <2 x half> @llvm.amdgcn.struct.ptr.buffer.atomic.fadd.v2f16(<2 x half>, ptr addrspace(8), i32, i32, i32, i32 immarg)