File: llvm.amdgcn.image.atomic.pk.add.ll

package info (click to toggle)
llvm-toolchain-19 1%3A19.1.7-3~deb12u1
  • links: PTS, VCS
  • area: main
  • in suites: bookworm-proposed-updates
  • size: 1,998,492 kB
  • sloc: cpp: 6,951,680; ansic: 1,486,157; asm: 913,598; python: 232,024; f90: 80,126; objc: 75,281; lisp: 37,276; pascal: 16,990; sh: 10,009; ml: 5,058; perl: 4,724; awk: 3,523; makefile: 3,167; javascript: 2,504; xml: 892; fortran: 664; cs: 573
file content (245 lines) | stat: -rw-r--r-- 12,922 bytes parent folder | download | duplicates (4) 
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 -mtriple=amdgcn -global-isel=0 -mcpu=gfx1200 -verify-machineinstrs < %s | FileCheck -check-prefix=GFX12-SDAG %s
; RUN: llc -mtriple=amdgcn -global-isel=1 -mcpu=gfx1200 -verify-machineinstrs < %s | FileCheck -check-prefix=GFX12-GISEL %s

define amdgpu_ps float @atomic_pk_add_f16_1d_v2(<8 x i32> inreg %rsrc, <2 x half> %data, i32 %s) {
; GFX12-SDAG-LABEL: atomic_pk_add_f16_1d_v2:
; GFX12-SDAG:       ; %bb.0: ; %main_body
; GFX12-SDAG-NEXT:    image_atomic_pk_add_f16 v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-SDAG-NEXT:    s_wait_loadcnt 0x0
; GFX12-SDAG-NEXT:    ; return to shader part epilog
;
; GFX12-GISEL-LABEL: atomic_pk_add_f16_1d_v2:
; GFX12-GISEL:       ; %bb.0: ; %main_body
; GFX12-GISEL-NEXT:    image_atomic_pk_add_f16 v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-GISEL-NEXT:    s_wait_loadcnt 0x0
; GFX12-GISEL-NEXT:    ; return to shader part epilog
main_body:
  %out = call <2 x half> @llvm.amdgcn.image.atomic.pk.add.f16.1d.v2f16.v2f16(<2 x half> %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
  %out_float = bitcast <2 x half> %out to float
  ret float %out_float
}

define amdgpu_ps float @atomic_pk_add_f16_1d_v2_nt(<8 x i32> inreg %rsrc, <2 x half> %data, i32 %s) {
; GFX12-SDAG-LABEL: atomic_pk_add_f16_1d_v2_nt:
; GFX12-SDAG:       ; %bb.0: ; %main_body
; GFX12-SDAG-NEXT:    image_atomic_pk_add_f16 v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_NT_RETURN
; GFX12-SDAG-NEXT:    s_wait_loadcnt 0x0
; GFX12-SDAG-NEXT:    ; return to shader part epilog
;
; GFX12-GISEL-LABEL: atomic_pk_add_f16_1d_v2_nt:
; GFX12-GISEL:       ; %bb.0: ; %main_body
; GFX12-GISEL-NEXT:    image_atomic_pk_add_f16 v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_NT_RETURN
; GFX12-GISEL-NEXT:    s_wait_loadcnt 0x0
; GFX12-GISEL-NEXT:    ; return to shader part epilog
main_body:
  %out = call <2 x half> @llvm.amdgcn.image.atomic.pk.add.f16.1d.v2f16.v2f16(<2 x half> %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 2)
  %out_float = bitcast <2 x half> %out to float
  ret float %out_float
}

define amdgpu_ps float @atomic_pk_add_f16_1d_v2_noret(<8 x i32> inreg %rsrc, <2 x half> %data, i32 %s) {
; GFX12-SDAG-LABEL: atomic_pk_add_f16_1d_v2_noret:
; GFX12-SDAG:       ; %bb.0: ; %main_body
; GFX12-SDAG-NEXT:    image_atomic_pk_add_f16 v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-SDAG-NEXT:    s_wait_loadcnt 0x0
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-SDAG-NEXT:    ; return to shader part epilog
;
; GFX12-GISEL-LABEL: atomic_pk_add_f16_1d_v2_noret:
; GFX12-GISEL:       ; %bb.0: ; %main_body
; GFX12-GISEL-NEXT:    image_atomic_pk_add_f16 v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-GISEL-NEXT:    s_wait_loadcnt 0x0
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-GISEL-NEXT:    ; return to shader part epilog
main_body:
  %unused = call <2 x half> @llvm.amdgcn.image.atomic.pk.add.f16.1d.v2f16.v2f16(<2 x half> %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
  ret float 1.0
}

define amdgpu_ps float @atomic_pk_add_f16_1d_v4(<8 x i32> inreg %rsrc, <4 x half> %data, i32 %s) {
; GFX12-SDAG-LABEL: atomic_pk_add_f16_1d_v4:
; GFX12-SDAG:       ; %bb.0: ; %main_body
; GFX12-SDAG-NEXT:    image_atomic_pk_add_f16 v[0:1], v2, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-SDAG-NEXT:    s_wait_loadcnt 0x0
; GFX12-SDAG-NEXT:    ; return to shader part epilog
;
; GFX12-GISEL-LABEL: atomic_pk_add_f16_1d_v4:
; GFX12-GISEL:       ; %bb.0: ; %main_body
; GFX12-GISEL-NEXT:    image_atomic_pk_add_f16 v[0:1], v2, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-GISEL-NEXT:    s_wait_loadcnt 0x0
; GFX12-GISEL-NEXT:    ; return to shader part epilog
main_body:
  %out = call <4 x half> @llvm.amdgcn.image.atomic.pk.add.f16.1d.v4f16.v4f16(<4 x half> %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
  %out_v2f32 = bitcast <4 x half> %out to <2 x float>
  %out0 = extractelement <2 x float> %out_v2f32, i32 0
  ret float %out0
}

define amdgpu_ps float @atomic_pk_add_f16_1d_v4_noret(<8 x i32> inreg %rsrc, <4 x half> %data, i32 %s) {
; GFX12-SDAG-LABEL: atomic_pk_add_f16_1d_v4_noret:
; GFX12-SDAG:       ; %bb.0: ; %main_body
; GFX12-SDAG-NEXT:    image_atomic_pk_add_f16 v[0:1], v2, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-SDAG-NEXT:    s_wait_loadcnt 0x0
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-SDAG-NEXT:    ; return to shader part epilog
;
; GFX12-GISEL-LABEL: atomic_pk_add_f16_1d_v4_noret:
; GFX12-GISEL:       ; %bb.0: ; %main_body
; GFX12-GISEL-NEXT:    image_atomic_pk_add_f16 v[0:1], v2, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-GISEL-NEXT:    s_wait_loadcnt 0x0
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-GISEL-NEXT:    ; return to shader part epilog
main_body:
  %unused = call <4 x half> @llvm.amdgcn.image.atomic.pk.add.f16.1d.v4f16.v4f16(<4 x half> %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
  ret float 1.0
}

define amdgpu_ps float @atomic_pk_add_bf16_1d_v2(<8 x i32> inreg %rsrc, <2 x bfloat> %data, i32 %s) {
; GFX12-SDAG-LABEL: atomic_pk_add_bf16_1d_v2:
; GFX12-SDAG:       ; %bb.0: ; %main_body
; GFX12-SDAG-NEXT:    image_atomic_pk_add_bf16 v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v1, 0
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v2, 0
; GFX12-SDAG-NEXT:    s_wait_loadcnt 0x0
; GFX12-SDAG-NEXT:    flat_store_b32 v[1:2], v0
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-SDAG-NEXT:    s_wait_dscnt 0x0
; GFX12-SDAG-NEXT:    ; return to shader part epilog
;
; GFX12-GISEL-LABEL: atomic_pk_add_bf16_1d_v2:
; GFX12-GISEL:       ; %bb.0: ; %main_body
; GFX12-GISEL-NEXT:    image_atomic_pk_add_bf16 v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v1, 0
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v2, 0
; GFX12-GISEL-NEXT:    s_wait_loadcnt 0x0
; GFX12-GISEL-NEXT:    flat_store_b32 v[1:2], v0
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-GISEL-NEXT:    s_wait_dscnt 0x0
; GFX12-GISEL-NEXT:    ; return to shader part epilog
main_body:
  %out = call <2 x bfloat> @llvm.amdgcn.image.atomic.pk.add.bf16.1d.v2bf16.v2bf16(<2 x bfloat> %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
  store <2 x bfloat> %out, ptr null
  ret float 1.0
}

define amdgpu_ps float @atomic_pk_add_bf16_1d_v2_noret(<8 x i32> inreg %rsrc, <2 x bfloat> %data, i32 %s) {
; GFX12-SDAG-LABEL: atomic_pk_add_bf16_1d_v2_noret:
; GFX12-SDAG:       ; %bb.0: ; %main_body
; GFX12-SDAG-NEXT:    image_atomic_pk_add_bf16 v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-SDAG-NEXT:    s_wait_loadcnt 0x0
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-SDAG-NEXT:    ; return to shader part epilog
;
; GFX12-GISEL-LABEL: atomic_pk_add_bf16_1d_v2_noret:
; GFX12-GISEL:       ; %bb.0: ; %main_body
; GFX12-GISEL-NEXT:    image_atomic_pk_add_bf16 v0, v1, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-GISEL-NEXT:    s_wait_loadcnt 0x0
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-GISEL-NEXT:    ; return to shader part epilog
main_body:
  %unused = call <2 x bfloat> @llvm.amdgcn.image.atomic.pk.add.bf16.1d.v2bf16.v2bf16(<2 x bfloat> %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
  ret float 1.0
}

define amdgpu_ps float @atomic_pk_add_bf16_1d_v4(<8 x i32> inreg %rsrc, <4 x bfloat> %data, i32 %s) {
; GFX12-SDAG-LABEL: atomic_pk_add_bf16_1d_v4:
; GFX12-SDAG:       ; %bb.0: ; %main_body
; GFX12-SDAG-NEXT:    image_atomic_pk_add_bf16 v[0:1], v2, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v2, 0
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v3, 0
; GFX12-SDAG-NEXT:    s_wait_loadcnt 0x0
; GFX12-SDAG-NEXT:    flat_store_b64 v[2:3], v[0:1]
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-SDAG-NEXT:    s_wait_dscnt 0x0
; GFX12-SDAG-NEXT:    ; return to shader part epilog
;
; GFX12-GISEL-LABEL: atomic_pk_add_bf16_1d_v4:
; GFX12-GISEL:       ; %bb.0: ; %main_body
; GFX12-GISEL-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
; GFX12-GISEL-NEXT:    v_lshrrev_b32_e32 v4, 16, v1
; GFX12-GISEL-NEXT:    v_and_b32_e32 v0, 0xffff, v0
; GFX12-GISEL-NEXT:    v_and_b32_e32 v1, 0xffff, v1
; GFX12-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4)
; GFX12-GISEL-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
; GFX12-GISEL-NEXT:    v_lshlrev_b32_e32 v4, 16, v4
; GFX12-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX12-GISEL-NEXT:    v_or_b32_e32 v0, v3, v0
; GFX12-GISEL-NEXT:    v_or_b32_e32 v1, v4, v1
; GFX12-GISEL-NEXT:    image_atomic_pk_add_bf16 v[0:1], v2, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v2, 0
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v3, 0
; GFX12-GISEL-NEXT:    s_wait_loadcnt 0x0
; GFX12-GISEL-NEXT:    flat_store_b64 v[2:3], v[0:1]
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-GISEL-NEXT:    s_wait_dscnt 0x0
; GFX12-GISEL-NEXT:    ; return to shader part epilog
main_body:
  %out = call <4 x bfloat> @llvm.amdgcn.image.atomic.pk.add.bf16.1d.v4bf16.v4bf16(<4 x bfloat> %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
  store <4 x bfloat> %out, ptr null
  ret float 1.0
}

define amdgpu_ps float @atomic_pk_add_bf16_1d_v4_noret(<8 x i32> inreg %rsrc, <4 x bfloat> %data, i32 %s) {
; GFX12-SDAG-LABEL: atomic_pk_add_bf16_1d_v4_noret:
; GFX12-SDAG:       ; %bb.0: ; %main_body
; GFX12-SDAG-NEXT:    image_atomic_pk_add_bf16 v[0:1], v2, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-SDAG-NEXT:    s_wait_loadcnt 0x0
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-SDAG-NEXT:    ; return to shader part epilog
;
; GFX12-GISEL-LABEL: atomic_pk_add_bf16_1d_v4_noret:
; GFX12-GISEL:       ; %bb.0: ; %main_body
; GFX12-GISEL-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
; GFX12-GISEL-NEXT:    v_lshrrev_b32_e32 v4, 16, v1
; GFX12-GISEL-NEXT:    v_and_b32_e32 v0, 0xffff, v0
; GFX12-GISEL-NEXT:    v_and_b32_e32 v1, 0xffff, v1
; GFX12-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4)
; GFX12-GISEL-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
; GFX12-GISEL-NEXT:    v_lshlrev_b32_e32 v4, 16, v4
; GFX12-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX12-GISEL-NEXT:    v_or_b32_e32 v0, v3, v0
; GFX12-GISEL-NEXT:    v_or_b32_e32 v1, v4, v1
; GFX12-GISEL-NEXT:    image_atomic_pk_add_bf16 v[0:1], v2, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_RETURN
; GFX12-GISEL-NEXT:    s_wait_loadcnt 0x0
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-GISEL-NEXT:    ; return to shader part epilog
main_body:
  %unused = call <4 x bfloat> @llvm.amdgcn.image.atomic.pk.add.bf16.1d.v4bf16.v4bf16(<4 x bfloat> %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
  ret float 1.0
}

define amdgpu_ps float @atomic_pk_add_bf16_1d_v4_nt(<8 x i32> inreg %rsrc, <4 x bfloat> %data, i32 %s) {
; GFX12-SDAG-LABEL: atomic_pk_add_bf16_1d_v4_nt:
; GFX12-SDAG:       ; %bb.0: ; %main_body
; GFX12-SDAG-NEXT:    image_atomic_pk_add_bf16 v[0:1], v2, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_NT_RETURN
; GFX12-SDAG-NEXT:    s_wait_loadcnt 0x0
; GFX12-SDAG-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-SDAG-NEXT:    ; return to shader part epilog
;
; GFX12-GISEL-LABEL: atomic_pk_add_bf16_1d_v4_nt:
; GFX12-GISEL:       ; %bb.0: ; %main_body
; GFX12-GISEL-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
; GFX12-GISEL-NEXT:    v_lshrrev_b32_e32 v4, 16, v1
; GFX12-GISEL-NEXT:    v_and_b32_e32 v0, 0xffff, v0
; GFX12-GISEL-NEXT:    v_and_b32_e32 v1, 0xffff, v1
; GFX12-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4)
; GFX12-GISEL-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
; GFX12-GISEL-NEXT:    v_lshlrev_b32_e32 v4, 16, v4
; GFX12-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX12-GISEL-NEXT:    v_or_b32_e32 v0, v3, v0
; GFX12-GISEL-NEXT:    v_or_b32_e32 v1, v4, v1
; GFX12-GISEL-NEXT:    image_atomic_pk_add_bf16 v[0:1], v2, s[0:7] dmask:0x3 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_NT_RETURN
; GFX12-GISEL-NEXT:    s_wait_loadcnt 0x0
; GFX12-GISEL-NEXT:    v_mov_b32_e32 v0, 1.0
; GFX12-GISEL-NEXT:    ; return to shader part epilog
main_body:
  %unused = call <4 x bfloat> @llvm.amdgcn.image.atomic.pk.add.bf16.1d.v4bf16.v4bf16(<4 x bfloat> %data, i32 %s, <8 x i32> %rsrc, i32 0, i32 2)
  ret float 1.0
}


declare <2 x half> @llvm.amdgcn.image.atomic.pk.add.f16.1d.v2f16.v2f16(<2 x half>, i32, <8 x i32>, i32, i32)
declare <4 x half> @llvm.amdgcn.image.atomic.pk.add.f16.1d.v4f16.v4f16(<4 x half>, i32, <8 x i32>, i32, i32)
declare <2 x bfloat> @llvm.amdgcn.image.atomic.pk.add.bf16.1d.v2bf16.v2bf16(<2 x bfloat>, i32, <8 x i32>, i32, i32)
declare <4 x bfloat> @llvm.amdgcn.image.atomic.pk.add.bf16.1d.v4bf16.v4bf16(<4 x bfloat>, i32, <8 x i32>, i32, i32)