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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=kaveri < %s | FileCheck -check-prefixes=GCN,GFX7 %s
; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=fiji < %s | FileCheck -check-prefixes=GCN,GFX8 %s
; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 < %s | FileCheck -check-prefixes=GCN,GFX9 %s
define i8 @atomic_load_flat_monotonic_i8(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_i8:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_ubyte v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i8, ptr %ptr monotonic, align 1
ret i8 %load
}
define i32 @atomic_load_flat_monotonic_i8_zext_to_i32(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_i8_zext_to_i32:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_ubyte v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i8, ptr %ptr monotonic, align 1
%ext = zext i8 %load to i32
ret i32 %ext
}
define i32 @atomic_load_flat_monotonic_i8_sext_to_i32(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_i8_sext_to_i32:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_sbyte v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i8, ptr %ptr monotonic, align 1
%ext = sext i8 %load to i32
ret i32 %ext
}
define i16 @atomic_load_flat_monotonic_i8_zext_to_i16(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_i8_zext_to_i16:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_ubyte v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i8, ptr %ptr monotonic, align 1
%ext = zext i8 %load to i16
ret i16 %ext
}
define i16 @atomic_load_flat_monotonic_i8_sext_to_i16(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_i8_sext_to_i16:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_sbyte v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i8, ptr %ptr monotonic, align 1
%ext = sext i8 %load to i16
ret i16 %ext
}
define i16 @atomic_load_flat_monotonic_i16(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_i16:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_ushort v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i16, ptr %ptr monotonic, align 2
ret i16 %load
}
define i32 @atomic_load_flat_monotonic_i16_zext_to_i32(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_i16_zext_to_i32:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_ushort v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i16, ptr %ptr monotonic, align 2
%ext = zext i16 %load to i32
ret i32 %ext
}
define i32 @atomic_load_flat_monotonic_i16_sext_to_i32(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_i16_sext_to_i32:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_sshort v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i16, ptr %ptr monotonic, align 2
%ext = sext i16 %load to i32
ret i32 %ext
}
define half @atomic_load_flat_monotonic_f16(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_f16:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_ushort v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic half, ptr %ptr monotonic, align 2
ret half %load
}
define bfloat @atomic_load_flat_monotonic_bf16(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_bf16:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_ushort v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic bfloat, ptr %ptr monotonic, align 2
ret bfloat %load
}
define i32 @atomic_load_flat_monotonic_f16_zext_to_i32(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_f16_zext_to_i32:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_ushort v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic half, ptr %ptr monotonic, align 2
%cast = bitcast half %load to i16
%ext = zext i16 %cast to i32
ret i32 %ext
}
define i32 @atomic_load_flat_monotonic_bf16_zext_to_i32(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_bf16_zext_to_i32:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_ushort v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic bfloat, ptr %ptr monotonic, align 2
%cast = bitcast bfloat %load to i16
%ext = zext i16 %cast to i32
ret i32 %ext
}
define i32 @atomic_load_flat_monotonic_i16_d16_hi_shift(ptr %ptr) {
; GCN-LABEL: atomic_load_flat_monotonic_i16_d16_hi_shift:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: flat_load_ushort v0, v[0:1] glc
; GCN-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT: v_lshlrev_b32_e32 v0, 16, v0
; GCN-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i16, ptr %ptr monotonic, align 2
%ext = zext i16 %load to i32
%shl = shl i32 %ext, 16
ret i32 %shl
}
define <2 x i16> @atomic_load_flat_monotonic_i16_d16_hi_vector_insert(ptr %ptr, <2 x i16> %vec) {
; GFX7-LABEL: atomic_load_flat_monotonic_i16_d16_hi_vector_insert:
; GFX7: ; %bb.0:
; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX7-NEXT: flat_load_ushort v0, v[0:1] glc
; GFX7-NEXT: v_lshlrev_b32_e32 v1, 16, v3
; GFX7-NEXT: v_and_b32_e32 v2, 0xffff, v2
; GFX7-NEXT: v_or_b32_e32 v1, v1, v2
; GFX7-NEXT: v_and_b32_e32 v1, 0xffff, v1
; GFX7-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX7-NEXT: v_lshlrev_b32_e32 v0, 16, v0
; GFX7-NEXT: v_or_b32_e32 v0, v1, v0
; GFX7-NEXT: v_lshrrev_b32_e32 v1, 16, v0
; GFX7-NEXT: s_setpc_b64 s[30:31]
;
; GFX8-LABEL: atomic_load_flat_monotonic_i16_d16_hi_vector_insert:
; GFX8: ; %bb.0:
; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX8-NEXT: flat_load_ushort v0, v[0:1] glc
; GFX8-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX8-NEXT: v_lshlrev_b32_e32 v0, 16, v0
; GFX8-NEXT: v_or_b32_sdwa v0, v2, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
; GFX8-NEXT: s_setpc_b64 s[30:31]
;
; GFX9-LABEL: atomic_load_flat_monotonic_i16_d16_hi_vector_insert:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT: flat_load_ushort v0, v[0:1] glc
; GFX9-NEXT: v_mov_b32_e32 v1, 0xffff
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 16, v0
; GFX9-NEXT: v_and_or_b32 v0, v2, v1, v0
; GFX9-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i16, ptr %ptr monotonic, align 2
%insert = insertelement <2 x i16> %vec, i16 %load, i32 1
ret <2 x i16> %insert
}
define i32 @atomic_load_flat_monotonic_i16_d16_lo_or(ptr %ptr, i16 %high) {
; GFX7-LABEL: atomic_load_flat_monotonic_i16_d16_lo_or:
; GFX7: ; %bb.0:
; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX7-NEXT: flat_load_ushort v0, v[0:1] glc
; GFX7-NEXT: v_and_b32_e32 v1, 0xffff, v2
; GFX7-NEXT: v_lshlrev_b32_e32 v1, 16, v1
; GFX7-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX7-NEXT: v_or_b32_e32 v0, v1, v0
; GFX7-NEXT: s_setpc_b64 s[30:31]
;
; GFX8-LABEL: atomic_load_flat_monotonic_i16_d16_lo_or:
; GFX8: ; %bb.0:
; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX8-NEXT: flat_load_ushort v0, v[0:1] glc
; GFX8-NEXT: v_and_b32_e32 v1, 0xffff, v2
; GFX8-NEXT: v_lshlrev_b32_e32 v1, 16, v1
; GFX8-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX8-NEXT: v_or_b32_e32 v0, v1, v0
; GFX8-NEXT: s_setpc_b64 s[30:31]
;
; GFX9-LABEL: atomic_load_flat_monotonic_i16_d16_lo_or:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT: flat_load_ushort v0, v[0:1] glc
; GFX9-NEXT: v_and_b32_e32 v1, 0xffff, v2
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX9-NEXT: v_lshl_or_b32 v0, v1, 16, v0
; GFX9-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i16, ptr %ptr monotonic, align 2
%ext = zext i16 %load to i32
%high.ext = zext i16 %high to i32
%shl = shl i32 %high.ext, 16
%or = or i32 %shl, %ext
ret i32 %or
}
define <2 x i16> @atomic_load_flat_monotonic_i16_d16_lo_vector_insert(ptr %ptr, <2 x i16> %vec) {
; GFX7-LABEL: atomic_load_flat_monotonic_i16_d16_lo_vector_insert:
; GFX7: ; %bb.0:
; GFX7-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX7-NEXT: flat_load_ushort v0, v[0:1] glc
; GFX7-NEXT: v_lshlrev_b32_e32 v1, 16, v3
; GFX7-NEXT: v_and_b32_e32 v2, 0xffff, v2
; GFX7-NEXT: v_or_b32_e32 v1, v1, v2
; GFX7-NEXT: v_and_b32_e32 v1, 0xffff0000, v1
; GFX7-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX7-NEXT: v_or_b32_e32 v0, v1, v0
; GFX7-NEXT: v_lshrrev_b32_e32 v1, 16, v0
; GFX7-NEXT: s_setpc_b64 s[30:31]
;
; GFX8-LABEL: atomic_load_flat_monotonic_i16_d16_lo_vector_insert:
; GFX8: ; %bb.0:
; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX8-NEXT: flat_load_ushort v0, v[0:1] glc
; GFX8-NEXT: v_and_b32_e32 v1, 0xffff0000, v2
; GFX8-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX8-NEXT: v_or_b32_e32 v0, v1, v0
; GFX8-NEXT: s_setpc_b64 s[30:31]
;
; GFX9-LABEL: atomic_load_flat_monotonic_i16_d16_lo_vector_insert:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT: flat_load_ushort v0, v[0:1] glc
; GFX9-NEXT: v_mov_b32_e32 v1, 0xffff0000
; GFX9-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX9-NEXT: v_and_or_b32 v0, v2, v1, v0
; GFX9-NEXT: s_setpc_b64 s[30:31]
%load = load atomic i16, ptr %ptr monotonic, align 2
%insert = insertelement <2 x i16> %vec, i16 %load, i32 0
ret <2 x i16> %insert
}
|
