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
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
|
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
// REQUIRES: amdgpu-registered-target
// RUN: %clang_cc1 %s -x hip -fcuda-is-device -emit-llvm -O0 -o - \
// RUN: -triple=amdgcn-amd-amdhsa | FileCheck %s
// CHECK-LABEL: @_Z29test_non_volatile_parameter32Pj(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[PTR_ADDR:%.*]] = alloca ptr, align 8, addrspace(5)
// CHECK-NEXT: [[RES:%.*]] = alloca i32, align 4, addrspace(5)
// CHECK-NEXT: [[PTR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[PTR_ADDR]] to ptr
// CHECK-NEXT: [[RES_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RES]] to ptr
// CHECK-NEXT: store ptr [[PTR:%.*]], ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw uinc_wrap ptr [[TMP0]], i32 [[TMP2]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP3]], ptr [[RES_ASCAST]], align 4
// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK-NEXT: [[TMP7:%.*]] = atomicrmw udec_wrap ptr [[TMP4]], i32 [[TMP6]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP7]], ptr [[RES_ASCAST]], align 4
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_non_volatile_parameter32(__UINT32_TYPE__ *ptr) {
__UINT32_TYPE__ res;
res = __builtin_amdgcn_atomic_inc32(ptr, *ptr, __ATOMIC_SEQ_CST, "workgroup");
res = __builtin_amdgcn_atomic_dec32(ptr, *ptr, __ATOMIC_SEQ_CST, "workgroup");
}
// CHECK-LABEL: @_Z29test_non_volatile_parameter64Py(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[PTR_ADDR:%.*]] = alloca ptr, align 8, addrspace(5)
// CHECK-NEXT: [[RES:%.*]] = alloca i64, align 8, addrspace(5)
// CHECK-NEXT: [[PTR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[PTR_ADDR]] to ptr
// CHECK-NEXT: [[RES_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RES]] to ptr
// CHECK-NEXT: store ptr [[PTR:%.*]], ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP2:%.*]] = load i64, ptr [[TMP1]], align 8
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw uinc_wrap ptr [[TMP0]], i64 [[TMP2]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP3]], ptr [[RES_ASCAST]], align 8
// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP6:%.*]] = load i64, ptr [[TMP5]], align 8
// CHECK-NEXT: [[TMP7:%.*]] = atomicrmw udec_wrap ptr [[TMP4]], i64 [[TMP6]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP7]], ptr [[RES_ASCAST]], align 8
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_non_volatile_parameter64(__UINT64_TYPE__ *ptr) {
__UINT64_TYPE__ res;
res = __builtin_amdgcn_atomic_inc64(ptr, *ptr, __ATOMIC_SEQ_CST, "workgroup");
res = __builtin_amdgcn_atomic_dec64(ptr, *ptr, __ATOMIC_SEQ_CST, "workgroup");
}
// CHECK-LABEL: @_Z25test_volatile_parameter32PVj(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[PTR_ADDR:%.*]] = alloca ptr, align 8, addrspace(5)
// CHECK-NEXT: [[RES:%.*]] = alloca i32, align 4, addrspace(5)
// CHECK-NEXT: [[PTR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[PTR_ADDR]] to ptr
// CHECK-NEXT: [[RES_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RES]] to ptr
// CHECK-NEXT: store ptr [[PTR:%.*]], ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP2:%.*]] = load volatile i32, ptr [[TMP1]], align 4
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw volatile uinc_wrap ptr [[TMP0]], i32 [[TMP2]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP3]], ptr [[RES_ASCAST]], align 4
// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP6:%.*]] = load volatile i32, ptr [[TMP5]], align 4
// CHECK-NEXT: [[TMP7:%.*]] = atomicrmw volatile udec_wrap ptr [[TMP4]], i32 [[TMP6]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP7]], ptr [[RES_ASCAST]], align 4
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_volatile_parameter32(volatile __UINT32_TYPE__ *ptr) {
__UINT32_TYPE__ res;
res = __builtin_amdgcn_atomic_inc32(ptr, *ptr, __ATOMIC_SEQ_CST, "workgroup");
res = __builtin_amdgcn_atomic_dec32(ptr, *ptr, __ATOMIC_SEQ_CST, "workgroup");
}
// CHECK-LABEL: @_Z25test_volatile_parameter64PVy(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[PTR_ADDR:%.*]] = alloca ptr, align 8, addrspace(5)
// CHECK-NEXT: [[RES:%.*]] = alloca i64, align 8, addrspace(5)
// CHECK-NEXT: [[PTR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[PTR_ADDR]] to ptr
// CHECK-NEXT: [[RES_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RES]] to ptr
// CHECK-NEXT: store ptr [[PTR:%.*]], ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP2:%.*]] = load volatile i64, ptr [[TMP1]], align 8
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw volatile uinc_wrap ptr [[TMP0]], i64 [[TMP2]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP3]], ptr [[RES_ASCAST]], align 8
// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
// CHECK-NEXT: [[TMP6:%.*]] = load volatile i64, ptr [[TMP5]], align 8
// CHECK-NEXT: [[TMP7:%.*]] = atomicrmw volatile udec_wrap ptr [[TMP4]], i64 [[TMP6]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP7]], ptr [[RES_ASCAST]], align 8
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_volatile_parameter64(volatile __UINT64_TYPE__ *ptr) {
__UINT64_TYPE__ res;
res = __builtin_amdgcn_atomic_inc64(ptr, *ptr, __ATOMIC_SEQ_CST, "workgroup");
res = __builtin_amdgcn_atomic_dec64(ptr, *ptr, __ATOMIC_SEQ_CST, "workgroup");
}
// CHECK-LABEL: @_Z13test_shared32v(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP1:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared32vE3val to ptr), i32 [[TMP0]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP1]], ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared32vE3val to ptr), i32 [[TMP2]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP3]], ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared32vE3val to ptr), align 4
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_shared32() {
__attribute__((shared)) __UINT32_TYPE__ val;
val = __builtin_amdgcn_atomic_inc32(&val, val, __ATOMIC_SEQ_CST, "workgroup");
val = __builtin_amdgcn_atomic_dec32(&val, val, __ATOMIC_SEQ_CST, "workgroup");
}
// CHECK-LABEL: @_Z13test_shared64v(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP1:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared64vE3val to ptr), i64 [[TMP0]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP1]], ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP2:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared64vE3val to ptr), i64 [[TMP2]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP3]], ptr addrspacecast (ptr addrspace(3) @_ZZ13test_shared64vE3val to ptr), align 8
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_shared64() {
__attribute__((shared)) __UINT64_TYPE__ val;
val = __builtin_amdgcn_atomic_inc64(&val, val, __ATOMIC_SEQ_CST, "workgroup");
val = __builtin_amdgcn_atomic_dec64(&val, val, __ATOMIC_SEQ_CST, "workgroup");
}
__attribute__((device)) __UINT32_TYPE__ global_val32;
// CHECK-LABEL: @_Z13test_global32v(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspacecast (ptr addrspace(1) @global_val32 to ptr), align 4
// CHECK-NEXT: [[TMP1:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(1) @global_val32 to ptr), i32 [[TMP0]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP1]], ptr addrspacecast (ptr addrspace(1) @global_val32 to ptr), align 4
// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr addrspacecast (ptr addrspace(1) @global_val32 to ptr), align 4
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(1) @global_val32 to ptr), i32 [[TMP2]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP3]], ptr addrspacecast (ptr addrspace(1) @global_val32 to ptr), align 4
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_global32() {
global_val32 = __builtin_amdgcn_atomic_inc32(&global_val32, global_val32, __ATOMIC_SEQ_CST, "workgroup");
global_val32 = __builtin_amdgcn_atomic_dec32(&global_val32, global_val32, __ATOMIC_SEQ_CST, "workgroup");
}
__attribute__((device)) __UINT64_TYPE__ global_val64;
// CHECK-LABEL: @_Z13test_global64v(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr addrspacecast (ptr addrspace(1) @global_val64 to ptr), align 8
// CHECK-NEXT: [[TMP1:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(1) @global_val64 to ptr), i64 [[TMP0]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP1]], ptr addrspacecast (ptr addrspace(1) @global_val64 to ptr), align 8
// CHECK-NEXT: [[TMP2:%.*]] = load i64, ptr addrspacecast (ptr addrspace(1) @global_val64 to ptr), align 8
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(1) @global_val64 to ptr), i64 [[TMP2]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP3]], ptr addrspacecast (ptr addrspace(1) @global_val64 to ptr), align 8
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_global64() {
global_val64 = __builtin_amdgcn_atomic_inc64(&global_val64, global_val64, __ATOMIC_SEQ_CST, "workgroup");
global_val64 = __builtin_amdgcn_atomic_dec64(&global_val64, global_val64, __ATOMIC_SEQ_CST, "workgroup");
}
__attribute__((constant)) __UINT32_TYPE__ cval32;
// CHECK-LABEL: @_Z15test_constant32v(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[LOCAL_VAL:%.*]] = alloca i32, align 4, addrspace(5)
// CHECK-NEXT: [[LOCAL_VAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[LOCAL_VAL]] to ptr
// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspacecast (ptr addrspace(4) @cval32 to ptr), align 4
// CHECK-NEXT: [[TMP1:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(4) @cval32 to ptr), i32 [[TMP0]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP1]], ptr [[LOCAL_VAL_ASCAST]], align 4
// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr addrspacecast (ptr addrspace(4) @cval32 to ptr), align 4
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(4) @cval32 to ptr), i32 [[TMP2]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP3]], ptr [[LOCAL_VAL_ASCAST]], align 4
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_constant32() {
__UINT32_TYPE__ local_val;
local_val = __builtin_amdgcn_atomic_inc32(&cval32, cval32, __ATOMIC_SEQ_CST, "workgroup");
local_val = __builtin_amdgcn_atomic_dec32(&cval32, cval32, __ATOMIC_SEQ_CST, "workgroup");
}
__attribute__((constant)) __UINT64_TYPE__ cval64;
// CHECK-LABEL: @_Z15test_constant64v(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[LOCAL_VAL:%.*]] = alloca i64, align 8, addrspace(5)
// CHECK-NEXT: [[LOCAL_VAL_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[LOCAL_VAL]] to ptr
// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr addrspacecast (ptr addrspace(4) @cval64 to ptr), align 8
// CHECK-NEXT: [[TMP1:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(4) @cval64 to ptr), i64 [[TMP0]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP1]], ptr [[LOCAL_VAL_ASCAST]], align 8
// CHECK-NEXT: [[TMP2:%.*]] = load i64, ptr addrspacecast (ptr addrspace(4) @cval64 to ptr), align 8
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(4) @cval64 to ptr), i64 [[TMP2]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP3]], ptr [[LOCAL_VAL_ASCAST]], align 8
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_constant64() {
__UINT64_TYPE__ local_val;
local_val = __builtin_amdgcn_atomic_inc64(&cval64, cval64, __ATOMIC_SEQ_CST, "workgroup");
local_val = __builtin_amdgcn_atomic_dec64(&cval64, cval64, __ATOMIC_SEQ_CST, "workgroup");
}
// CHECK-LABEL: @_Z12test_order32v(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP1:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), i32 [[TMP0]] syncscope("workgroup") monotonic, align 4
// CHECK-NEXT: store i32 [[TMP1]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), i32 [[TMP2]] syncscope("workgroup") acquire, align 4
// CHECK-NEXT: store i32 [[TMP3]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP5:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), i32 [[TMP4]] syncscope("workgroup") acquire, align 4
// CHECK-NEXT: store i32 [[TMP5]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP7:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), i32 [[TMP6]] syncscope("workgroup") release, align 4
// CHECK-NEXT: store i32 [[TMP7]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP9:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), i32 [[TMP8]] syncscope("workgroup") acq_rel, align 4
// CHECK-NEXT: store i32 [[TMP9]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP11:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), i32 [[TMP10]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP11]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order32vE3val to ptr), align 4
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_order32() {
__attribute__((shared)) __UINT32_TYPE__ val;
val = __builtin_amdgcn_atomic_inc32(&val, val, __ATOMIC_RELAXED, "workgroup");
val = __builtin_amdgcn_atomic_inc32(&val, val, __ATOMIC_CONSUME, "workgroup");
val = __builtin_amdgcn_atomic_inc32(&val, val, __ATOMIC_ACQUIRE, "workgroup");
val = __builtin_amdgcn_atomic_dec32(&val, val, __ATOMIC_RELEASE, "workgroup");
val = __builtin_amdgcn_atomic_dec32(&val, val, __ATOMIC_ACQ_REL, "workgroup");
val = __builtin_amdgcn_atomic_dec32(&val, val, __ATOMIC_SEQ_CST, "workgroup");
}
// CHECK-LABEL: @_Z12test_order64v(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP1:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), i64 [[TMP0]] syncscope("workgroup") monotonic, align 8
// CHECK-NEXT: store i64 [[TMP1]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP2:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), i64 [[TMP2]] syncscope("workgroup") acquire, align 8
// CHECK-NEXT: store i64 [[TMP3]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP4:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP5:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), i64 [[TMP4]] syncscope("workgroup") acquire, align 8
// CHECK-NEXT: store i64 [[TMP5]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP6:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP7:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), i64 [[TMP6]] syncscope("workgroup") release, align 8
// CHECK-NEXT: store i64 [[TMP7]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP8:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP9:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), i64 [[TMP8]] syncscope("workgroup") acq_rel, align 8
// CHECK-NEXT: store i64 [[TMP9]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP10:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP11:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), i64 [[TMP10]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP11]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_order64vE3val to ptr), align 8
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_order64() {
__attribute__((shared)) __UINT64_TYPE__ val;
val = __builtin_amdgcn_atomic_inc64(&val, val, __ATOMIC_RELAXED, "workgroup");
val = __builtin_amdgcn_atomic_dec64(&val, val, __ATOMIC_CONSUME, "workgroup");
val = __builtin_amdgcn_atomic_inc64(&val, val, __ATOMIC_ACQUIRE, "workgroup");
val = __builtin_amdgcn_atomic_dec64(&val, val, __ATOMIC_RELEASE, "workgroup");
val = __builtin_amdgcn_atomic_dec64(&val, val, __ATOMIC_ACQ_REL, "workgroup");
val = __builtin_amdgcn_atomic_dec64(&val, val, __ATOMIC_SEQ_CST, "workgroup");
}
// CHECK-LABEL: @_Z12test_scope32v(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP1:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), i32 [[TMP0]] seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP1]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), i32 [[TMP2]] syncscope("workgroup") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP3]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP5:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), i32 [[TMP4]] syncscope("agent") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP5]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), align 4
// CHECK-NEXT: [[TMP7:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), i32 [[TMP6]] syncscope("wavefront") seq_cst, align 4
// CHECK-NEXT: store i32 [[TMP7]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope32vE3val to ptr), align 4
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_scope32() {
__attribute__((shared)) __UINT32_TYPE__ val;
val = __builtin_amdgcn_atomic_inc32(&val, val, __ATOMIC_SEQ_CST, "");
val = __builtin_amdgcn_atomic_dec32(&val, val, __ATOMIC_SEQ_CST, "workgroup");
val = __builtin_amdgcn_atomic_dec32(&val, val, __ATOMIC_SEQ_CST, "agent");
val = __builtin_amdgcn_atomic_dec32(&val, val, __ATOMIC_SEQ_CST, "wavefront");
}
// CHECK-LABEL: @_Z12test_scope64v(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP1:%.*]] = atomicrmw uinc_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), i64 [[TMP0]] seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP1]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP2:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), i64 [[TMP2]] syncscope("workgroup") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP3]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP4:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP5:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), i64 [[TMP4]] syncscope("agent") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP5]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP6:%.*]] = load i64, ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), align 8
// CHECK-NEXT: [[TMP7:%.*]] = atomicrmw udec_wrap ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), i64 [[TMP6]] syncscope("wavefront") seq_cst, align 8
// CHECK-NEXT: store i64 [[TMP7]], ptr addrspacecast (ptr addrspace(3) @_ZZ12test_scope64vE3val to ptr), align 8
// CHECK-NEXT: ret void
//
__attribute__((device)) void test_scope64() {
__attribute__((shared)) __UINT64_TYPE__ val;
val = __builtin_amdgcn_atomic_inc64(&val, val, __ATOMIC_SEQ_CST, "");
val = __builtin_amdgcn_atomic_dec64(&val, val, __ATOMIC_SEQ_CST, "workgroup");
val = __builtin_amdgcn_atomic_dec64(&val, val, __ATOMIC_SEQ_CST, "agent");
val = __builtin_amdgcn_atomic_dec64(&val, val, __ATOMIC_SEQ_CST, "wavefront");
}
|
