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
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc -global-isel=0 -mtriple=amdgcn--amdpal -mcpu=gfx1200 -mattr=+real-true16 < %s | FileCheck %s --check-prefixes=CHECK,DAGISEL
; RUN: llc -global-isel=0 -mtriple=amdgcn--amdpal -mcpu=gfx1200 -mattr=-real-true16 < %s | FileCheck %s --check-prefixes=CHECK,DAGISEL
; RUN: llc -global-isel=1 -mtriple=amdgcn--amdpal -mcpu=gfx1200 -mattr=+real-true16 < %s | FileCheck %s --check-prefixes=CHECK,GISEL
; RUN: llc -global-isel=1 -mtriple=amdgcn--amdpal -mcpu=gfx1200 -mattr=-real-true16 < %s | FileCheck %s --check-prefixes=CHECK,GISEL
; Test that in dynamic VGPR mode, the return of sponentry points after the area reserved for CWSR.
define amdgpu_cs ptr addrspace(5) @sponentry_cs_dvgpr_16(i32 %val) #0 {
; CHECK-LABEL: sponentry_cs_dvgpr_16:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_getreg_b32 s33, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; CHECK-NEXT: s_getreg_b32 s0, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; CHECK-NEXT: s_cmp_lg_u32 0, s33
; CHECK-NEXT: s_cmovk_i32 s33, 0x1c0
; CHECK-NEXT: s_cmp_lg_u32 0, s0
; CHECK-NEXT: scratch_store_b32 off, v0, s33 scope:SCOPE_SYS
; CHECK-NEXT: s_wait_storecnt 0x0
; CHECK-NEXT: s_cmovk_i32 s0, 0x1c0
; CHECK-NEXT: ; return to shader part epilog
%local = alloca i32, addrspace(5)
store volatile i32 %val, ptr addrspace(5) %local
%stack.base = call ptr addrspace(5) @llvm.sponentry()
ret ptr addrspace(5) %stack.base
}
; CHECK: ScratchSize: 8
define amdgpu_cs ptr addrspace(5) @sponentry_cs_dvgpr_32(i32 %val) #1 {
; CHECK-LABEL: sponentry_cs_dvgpr_32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_getreg_b32 s33, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; CHECK-NEXT: s_getreg_b32 s0, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; CHECK-NEXT: s_cmp_lg_u32 0, s33
; CHECK-NEXT: s_cmovk_i32 s33, 0x380
; CHECK-NEXT: s_cmp_lg_u32 0, s0
; CHECK-NEXT: scratch_store_b32 off, v0, s33 scope:SCOPE_SYS
; CHECK-NEXT: s_wait_storecnt 0x0
; CHECK-NEXT: s_cmovk_i32 s0, 0x380
; CHECK-NEXT: ; return to shader part epilog
%local = alloca i32, addrspace(5)
store volatile i32 %val, ptr addrspace(5) %local
%stack.base = call ptr addrspace(5) @llvm.sponentry()
ret ptr addrspace(5) %stack.base
}
; CHECK: ScratchSize: 8
; If we're not in dynamic VGPR mode, then sponentry can just return 0.
define amdgpu_cs ptr addrspace(5) @sponentry_cs_no_dvgpr(i32 %val) #2 {
; CHECK-LABEL: sponentry_cs_no_dvgpr:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_mov_b32 s0, 0
; CHECK-NEXT: scratch_store_b32 off, v0, off scope:SCOPE_SYS
; CHECK-NEXT: s_wait_storecnt 0x0
; CHECK-NEXT: ; return to shader part epilog
%local = alloca i32, addrspace(5)
store volatile i32 %val, ptr addrspace(5) %local
%stack.base = call ptr addrspace(5) @llvm.sponentry()
ret ptr addrspace(5) %stack.base
}
; CHECK: ScratchSize: 8
define amdgpu_cs ptr addrspace(5) @sponentry_cs_dvgpr_control_flow(i32 %val, ptr addrspace(5) %ptr) #0 {
; CHECK-LABEL: sponentry_cs_dvgpr_control_flow:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_getreg_b32 s33, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; CHECK-NEXT: s_mov_b32 s0, exec_lo
; CHECK-NEXT: s_cmp_lg_u32 0, s33
; CHECK-NEXT: s_cmovk_i32 s33, 0x1c0
; CHECK-NEXT: scratch_store_b32 off, v0, s33 scope:SCOPE_SYS
; CHECK-NEXT: s_wait_storecnt 0x0
; CHECK-NEXT: v_cmpx_gt_i32_e32 0x43, v0
; CHECK-NEXT: ; %bb.1: ; %if.then
; CHECK-NEXT: s_getreg_b32 s1, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
; CHECK-NEXT: s_cmp_lg_u32 0, s1
; CHECK-NEXT: s_cmovk_i32 s1, 0x1c0
; CHECK-NEXT: v_mov_b32_e32 v1, s1
; CHECK-NEXT: ; %bb.2: ; %if.end
; CHECK-NEXT: s_or_b32 exec_lo, exec_lo, s0
; CHECK-NEXT: s_delay_alu instid0(VALU_DEP_1)
; CHECK-NEXT: v_readfirstlane_b32 s0, v1
; CHECK-NEXT: s_wait_alu depctr_va_sdst(0)
; CHECK-NEXT: ; return to shader part epilog
entry:
%local = alloca i32, addrspace(5)
store volatile i32 %val, ptr addrspace(5) %local
%which = icmp slt i32 %val, 67
br i1 %which, label %if.then, label %if.end
if.then:
%stack.base = call ptr addrspace(5) @llvm.sponentry()
br label %if.end
if.end:
%ret = phi ptr addrspace(5) [ %stack.base, %if.then ], [ %ptr, %entry ]
ret ptr addrspace(5) %ret
}
; CHECK: ScratchSize: 8
declare amdgpu_gfx i32 @callee()
define amdgpu_cs ptr addrspace(5) @sponentry_cs_dvgpr_calls(i32 %val) #0 {
; DAGISEL-LABEL: sponentry_cs_dvgpr_calls:
; DAGISEL: ; %bb.0:
; DAGISEL-NEXT: s_getreg_b32 s33, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; DAGISEL-NEXT: s_mov_b32 s1, callee@abs32@hi
; DAGISEL-NEXT: s_cmp_lg_u32 0, s33
; DAGISEL-NEXT: s_mov_b32 s0, callee@abs32@lo
; DAGISEL-NEXT: s_cmovk_i32 s33, 0x1c0
; DAGISEL-NEXT: s_cselect_b32 s32, 0x1d0, 16
; DAGISEL-NEXT: s_swappc_b64 s[30:31], s[0:1]
; DAGISEL-NEXT: s_getreg_b32 s0, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; DAGISEL-NEXT: s_wait_storecnt 0x0
; DAGISEL-NEXT: scratch_store_b32 off, v0, s33 scope:SCOPE_SYS
; DAGISEL-NEXT: s_wait_storecnt 0x0
; DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; DAGISEL-NEXT: s_cmp_lg_u32 0, s0
; DAGISEL-NEXT: s_cmovk_i32 s0, 0x1c0
; DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; DAGISEL-NEXT: ; return to shader part epilog
;
; GISEL-LABEL: sponentry_cs_dvgpr_calls:
; GISEL: ; %bb.0:
; GISEL-NEXT: s_getreg_b32 s33, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; GISEL-NEXT: s_mov_b32 s0, callee@abs32@lo
; GISEL-NEXT: s_cmp_lg_u32 0, s33
; GISEL-NEXT: s_mov_b32 s1, callee@abs32@hi
; GISEL-NEXT: s_cmovk_i32 s33, 0x1c0
; GISEL-NEXT: s_cselect_b32 s32, 0x1d0, 16
; GISEL-NEXT: s_swappc_b64 s[30:31], s[0:1]
; GISEL-NEXT: s_getreg_b32 s0, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; GISEL-NEXT: s_wait_storecnt 0x0
; GISEL-NEXT: scratch_store_b32 off, v0, s33 scope:SCOPE_SYS
; GISEL-NEXT: s_wait_storecnt 0x0
; GISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GISEL-NEXT: s_cmp_lg_u32 0, s0
; GISEL-NEXT: s_cmovk_i32 s0, 0x1c0
; GISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GISEL-NEXT: ; return to shader part epilog
%local = alloca i32, addrspace(5)
%res = call amdgpu_gfx i32 @callee()
store volatile i32 %res, ptr addrspace(5) %local
%stack.base = call ptr addrspace(5) @llvm.sponentry()
ret ptr addrspace(5) %stack.base
}
; CHECK: ScratchSize: 16
define amdgpu_cs ptr addrspace(5) @sponentry_cs_dvgpr_realign(i32 %val) #0 {
; CHECK-LABEL: sponentry_cs_dvgpr_realign:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_getreg_b32 s33, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; CHECK-NEXT: s_getreg_b32 s0, hwreg(HW_REG_WAVE_HW_ID2, 8, 2)
; CHECK-NEXT: s_cmp_lg_u32 0, s33
; CHECK-NEXT: s_cmovk_i32 s33, 0x200
; CHECK-NEXT: s_cmp_lg_u32 0, s0
; CHECK-NEXT: scratch_store_b32 off, v0, s33 scope:SCOPE_SYS
; CHECK-NEXT: s_wait_storecnt 0x0
; CHECK-NEXT: s_cmovk_i32 s0, 0x200
; CHECK-NEXT: ; return to shader part epilog
%boop = alloca i32, addrspace(5)
%local = alloca i32, align 128, addrspace(5)
store volatile i32 %val, ptr addrspace(5) %local
%stack.base = call ptr addrspace(5) @llvm.sponentry()
ret ptr addrspace(5) %stack.base
}
; CHECK: ScratchSize: 128
define amdgpu_gfx ptr addrspace(5) @sponentry_gfx(i32 %val, ptr addrspace(5) %ptr) #0 {
; DAGISEL-LABEL: sponentry_gfx:
; DAGISEL: ; %bb.0: ; %entry
; DAGISEL-NEXT: s_wait_loadcnt_dscnt 0x0
; DAGISEL-NEXT: s_wait_expcnt 0x0
; DAGISEL-NEXT: s_wait_samplecnt 0x0
; DAGISEL-NEXT: s_wait_bvhcnt 0x0
; DAGISEL-NEXT: s_wait_kmcnt 0x0
; DAGISEL-NEXT: s_mov_b32 s0, exec_lo
; DAGISEL-NEXT: s_wait_storecnt 0x0
; DAGISEL-NEXT: scratch_store_b32 off, v0, s32 offset:4 scope:SCOPE_SYS
; DAGISEL-NEXT: s_wait_storecnt 0x0
; DAGISEL-NEXT: v_cmpx_gt_i32_e32 0x43, v0
; DAGISEL-NEXT: ; %bb.1: ; %if.then
; DAGISEL-NEXT: v_mov_b32_e32 v1, s32
; DAGISEL-NEXT: ; %bb.2: ; %if.end
; DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; DAGISEL-NEXT: v_mov_b32_e32 v0, v1
; DAGISEL-NEXT: s_setpc_b64 s[30:31]
;
; GISEL-LABEL: sponentry_gfx:
; GISEL: ; %bb.0: ; %entry
; GISEL-NEXT: s_wait_loadcnt_dscnt 0x0
; GISEL-NEXT: s_wait_expcnt 0x0
; GISEL-NEXT: s_wait_samplecnt 0x0
; GISEL-NEXT: s_wait_bvhcnt 0x0
; GISEL-NEXT: s_wait_kmcnt 0x0
; GISEL-NEXT: v_mov_b32_e32 v2, v0
; GISEL-NEXT: v_mov_b32_e32 v0, v1
; GISEL-NEXT: s_mov_b32 s0, exec_lo
; GISEL-NEXT: s_wait_storecnt 0x0
; GISEL-NEXT: scratch_store_b32 off, v2, s32 offset:4 scope:SCOPE_SYS
; GISEL-NEXT: s_wait_storecnt 0x0
; GISEL-NEXT: v_cmpx_gt_i32_e32 0x43, v2
; GISEL-NEXT: ; %bb.1: ; %if.then
; GISEL-NEXT: s_mov_b32 s1, s32
; GISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GISEL-NEXT: v_mov_b32_e32 v0, s1
; GISEL-NEXT: ; %bb.2: ; %if.end
; GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GISEL-NEXT: s_setpc_b64 s[30:31]
entry:
%local = alloca i32, addrspace(5)
store volatile i32 %val, ptr addrspace(5) %local
%which = icmp slt i32 %val, 67
br i1 %which, label %if.then, label %if.end
if.then:
%stack.base = call ptr addrspace(5) @llvm.sponentry()
br label %if.end
if.end:
%ret = phi ptr addrspace(5) [ %stack.base, %if.then ], [ %ptr, %entry ]
ret ptr addrspace(5) %ret
}
; FIXME: Optimize away the 4 bytes for the sponentry frame index.
; CHECK: ScratchSize: 12
define amdgpu_gfx ptr addrspace(5) @sponentry_gfx_dvgpr_realign(i32 %val) #0 {
; CHECK-LABEL: sponentry_gfx_dvgpr_realign:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_wait_loadcnt_dscnt 0x0
; CHECK-NEXT: s_wait_expcnt 0x0
; CHECK-NEXT: s_wait_samplecnt 0x0
; CHECK-NEXT: s_wait_bvhcnt 0x0
; CHECK-NEXT: s_wait_kmcnt 0x0
; CHECK-NEXT: s_mov_b32 s0, s33
; CHECK-NEXT: s_add_co_i32 s33, s32, 0x7f
; CHECK-NEXT: s_mov_b32 s1, s34
; CHECK-NEXT: s_wait_alu depctr_sa_sdst(0)
; CHECK-NEXT: s_and_b32 s33, s33, 0xffffff80
; CHECK-NEXT: s_mov_b32 s34, s32
; CHECK-NEXT: s_wait_storecnt 0x0
; CHECK-NEXT: scratch_store_b32 off, v0, s33 offset:128 scope:SCOPE_SYS
; CHECK-NEXT: s_wait_storecnt 0x0
; CHECK-NEXT: s_wait_alu depctr_sa_sdst(0)
; CHECK-NEXT: v_mov_b32_e32 v0, s34
; CHECK-NEXT: s_addk_co_i32 s32, 0x180
; CHECK-NEXT: s_mov_b32 s32, s34
; CHECK-NEXT: s_mov_b32 s34, s1
; CHECK-NEXT: s_mov_b32 s33, s0
; CHECK-NEXT: s_wait_alu depctr_sa_sdst(0)
; CHECK-NEXT: s_setpc_b64 s[30:31]
%boop = alloca i32, addrspace(5)
%local = alloca i32, align 128, addrspace(5)
store volatile i32 %val, ptr addrspace(5) %local
%stack.base = call ptr addrspace(5) @llvm.sponentry()
ret ptr addrspace(5) %stack.base
}
; FIXME: Optimize away the sponentry fixed object.
; CHECK: ScratchSize: 384
define amdgpu_gfx ptr addrspace(5) @sponentry_gfx_stack_args(<32 x i32> %fill.sgprs, i32 %val, ptr addrspace(5) %ptr) #0 {
; DAGISEL-LABEL: sponentry_gfx_stack_args:
; DAGISEL: ; %bb.0:
; DAGISEL-NEXT: s_wait_loadcnt_dscnt 0x0
; DAGISEL-NEXT: s_wait_expcnt 0x0
; DAGISEL-NEXT: s_wait_samplecnt 0x0
; DAGISEL-NEXT: s_wait_bvhcnt 0x0
; DAGISEL-NEXT: s_wait_kmcnt 0x0
; DAGISEL-NEXT: s_clause 0x1
; DAGISEL-NEXT: scratch_load_b32 v0, off, s32 offset:4
; DAGISEL-NEXT: scratch_load_b32 v1, off, s32
; DAGISEL-NEXT: s_wait_loadcnt 0x0
; DAGISEL-NEXT: s_wait_storecnt 0x0
; DAGISEL-NEXT: scratch_store_b32 v0, v1, off scope:SCOPE_SYS
; DAGISEL-NEXT: s_wait_storecnt 0x0
; DAGISEL-NEXT: v_mov_b32_e32 v0, s32
; DAGISEL-NEXT: s_setpc_b64 s[30:31]
;
; GISEL-LABEL: sponentry_gfx_stack_args:
; GISEL: ; %bb.0:
; GISEL-NEXT: s_wait_loadcnt_dscnt 0x0
; GISEL-NEXT: s_wait_expcnt 0x0
; GISEL-NEXT: s_wait_samplecnt 0x0
; GISEL-NEXT: s_wait_bvhcnt 0x0
; GISEL-NEXT: s_wait_kmcnt 0x0
; GISEL-NEXT: s_clause 0x1
; GISEL-NEXT: scratch_load_b32 v0, off, s32
; GISEL-NEXT: scratch_load_b32 v1, off, s32 offset:4
; GISEL-NEXT: s_wait_loadcnt 0x0
; GISEL-NEXT: s_wait_storecnt 0x0
; GISEL-NEXT: scratch_store_b32 v1, v0, off scope:SCOPE_SYS
; GISEL-NEXT: s_wait_storecnt 0x0
; GISEL-NEXT: v_mov_b32_e32 v0, s32
; GISEL-NEXT: s_setpc_b64 s[30:31]
store volatile i32 %val, ptr addrspace(5) %ptr
%stack.base = call ptr addrspace(5) @llvm.sponentry()
ret ptr addrspace(5) %stack.base
}
; CHECK: ScratchSize: 12
define amdgpu_gfx ptr addrspace(5) @sponentry_gfx_dyn_alloc(i32 %val) #0 {
; DAGISEL-LABEL: sponentry_gfx_dyn_alloc:
; DAGISEL: ; %bb.0:
; DAGISEL-NEXT: s_wait_loadcnt_dscnt 0x0
; DAGISEL-NEXT: s_wait_expcnt 0x0
; DAGISEL-NEXT: s_wait_samplecnt 0x0
; DAGISEL-NEXT: s_wait_bvhcnt 0x0
; DAGISEL-NEXT: s_wait_kmcnt 0x0
; DAGISEL-NEXT: v_lshl_add_u32 v1, v0, 2, 15
; DAGISEL-NEXT: s_mov_b32 s34, s33
; DAGISEL-NEXT: s_mov_b32 s1, exec_lo
; DAGISEL-NEXT: s_mov_b32 s0, 0
; DAGISEL-NEXT: s_mov_b32 s33, s32
; DAGISEL-NEXT: v_and_b32_e32 v1, -16, v1
; DAGISEL-NEXT: s_add_co_i32 s32, s32, 16
; DAGISEL-NEXT: .LBB9_1: ; =>This Inner Loop Header: Depth=1
; DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; DAGISEL-NEXT: s_ctz_i32_b32 s2, s1
; DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; DAGISEL-NEXT: v_readlane_b32 s3, v1, s2
; DAGISEL-NEXT: s_bitset0_b32 s1, s2
; DAGISEL-NEXT: s_max_u32 s0, s0, s3
; DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; DAGISEL-NEXT: s_cmp_lg_u32 s1, 0
; DAGISEL-NEXT: s_cbranch_scc1 .LBB9_1
; DAGISEL-NEXT: ; %bb.2:
; DAGISEL-NEXT: s_mov_b32 s1, s32
; DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; DAGISEL-NEXT: v_lshl_add_u32 v1, s0, 5, s1
; DAGISEL-NEXT: s_wait_storecnt 0x0
; DAGISEL-NEXT: scratch_store_b32 off, v0, s1 scope:SCOPE_SYS
; DAGISEL-NEXT: s_wait_storecnt 0x0
; DAGISEL-NEXT: v_mov_b32_e32 v0, s33
; DAGISEL-NEXT: v_readfirstlane_b32 s32, v1
; DAGISEL-NEXT: s_mov_b32 s32, s33
; DAGISEL-NEXT: s_mov_b32 s33, s34
; DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; DAGISEL-NEXT: s_setpc_b64 s[30:31]
;
; GISEL-LABEL: sponentry_gfx_dyn_alloc:
; GISEL: ; %bb.0:
; GISEL-NEXT: s_wait_loadcnt_dscnt 0x0
; GISEL-NEXT: s_wait_expcnt 0x0
; GISEL-NEXT: s_wait_samplecnt 0x0
; GISEL-NEXT: s_wait_bvhcnt 0x0
; GISEL-NEXT: s_wait_kmcnt 0x0
; GISEL-NEXT: v_lshl_add_u32 v1, v0, 2, 15
; GISEL-NEXT: s_mov_b32 s34, s33
; GISEL-NEXT: s_mov_b32 s1, exec_lo
; GISEL-NEXT: s_mov_b32 s0, 0
; GISEL-NEXT: s_mov_b32 s33, s32
; GISEL-NEXT: v_and_b32_e32 v1, -16, v1
; GISEL-NEXT: s_add_co_i32 s32, s32, 16
; GISEL-NEXT: .LBB9_1: ; =>This Inner Loop Header: Depth=1
; GISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GISEL-NEXT: s_ctz_i32_b32 s2, s1
; GISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GISEL-NEXT: v_readlane_b32 s3, v1, s2
; GISEL-NEXT: s_bitset0_b32 s1, s2
; GISEL-NEXT: s_max_u32 s0, s0, s3
; GISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GISEL-NEXT: s_cmp_lg_u32 s1, 0
; GISEL-NEXT: s_cbranch_scc1 .LBB9_1
; GISEL-NEXT: ; %bb.2:
; GISEL-NEXT: s_mov_b32 s1, s32
; GISEL-NEXT: s_lshl_b32 s0, s0, 5
; GISEL-NEXT: s_wait_storecnt 0x0
; GISEL-NEXT: scratch_store_b32 off, v0, s1 scope:SCOPE_SYS
; GISEL-NEXT: s_wait_storecnt 0x0
; GISEL-NEXT: v_mov_b32_e32 v0, s33
; GISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GISEL-NEXT: s_add_co_u32 s32, s1, s0
; GISEL-NEXT: s_mov_b32 s32, s33
; GISEL-NEXT: s_mov_b32 s33, s34
; GISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GISEL-NEXT: s_setpc_b64 s[30:31]
%local = alloca i32, i32 %val, addrspace(5)
store volatile i32 %val, ptr addrspace(5) %local
%stack.base = call ptr addrspace(5) @llvm.sponentry()
ret ptr addrspace(5) %stack.base
}
; CHECK: ScratchSize: 16
attributes #0 = { nounwind "amdgpu-dynamic-vgpr-block-size"="16" }
attributes #1 = { nounwind "amdgpu-dynamic-vgpr-block-size"="32" }
attributes #2 = { nounwind "amdgpu-dynamic-vgpr-block-size"="0" }
|
