221
|
1 ; RUN: llc -march=amdgcn -mcpu=gfx1030 -verify-machineinstrs -asm-verbose=0 < %s | FileCheck --check-prefixes=GCN,GFX10,GFX10-ASM %s
|
|
2 ; RUN: llc -march=amdgcn -mcpu=gfx1030 -verify-machineinstrs < %s -filetype=obj | llvm-objdump -d --arch-name=amdgcn --mcpu=gfx1030 - | FileCheck --check-prefixes=GCN,GFX10,GFX10-DIS %s
|
|
3 ; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck --check-prefix=GFX8 %s
|
|
4
|
|
5 ; GFX8-NOT: s_inst_prefetch
|
|
6 ; GFX8-NOT: .palign 6
|
|
7
|
|
8 ; GCN-LABEL: test_loop_64
|
|
9 ; GFX10: s_movk_i32 s{{[0-9]+}}, 0x400
|
|
10 ; GFX10-DIS-NEXT: {{^$}}
|
|
11 ; GFX10-ASM-NEXT: [[L1:BB[0-9_]+]]:
|
|
12 ; GFX10-DIS-NEXT: <[[L1:BB[0-9_]+]]>:
|
|
13 ; GFX10: s_sleep 0
|
|
14 ; GFX10: s_cbranch_scc0 [[L1]]
|
|
15 ; GFX10-NEXT: s_endpgm
|
|
16 define amdgpu_kernel void @test_loop_64(i32 addrspace(1)* nocapture %arg) {
|
|
17 bb:
|
|
18 br label %bb2
|
|
19
|
|
20 bb1: ; preds = %bb2
|
|
21 ret void
|
|
22
|
|
23 bb2: ; preds = %bb2, %bb
|
|
24 %tmp1 = phi i32 [ 0, %bb ], [ %tmp2, %bb2 ]
|
|
25 %tmp2 = add nuw nsw i32 %tmp1, 1
|
|
26 %tmp3 = icmp eq i32 %tmp2, 1024
|
|
27 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
28 br i1 %tmp3, label %bb1, label %bb2
|
|
29 }
|
|
30
|
|
31 ; GCN-LABEL: test_loop_128
|
|
32 ; GFX10: s_movk_i32 s{{[0-9]+}}, 0x400
|
|
33 ; GFX10-ASM-NEXT: .p2align 6
|
|
34 ; GFX10-DIS-NEXT: s_nop 0
|
|
35 ; GFX10-NOT: s_inst_prefetch
|
|
36 ; GFX10-ASM: [[L1:BB[0-9_]+]]:
|
|
37 ; GFX10-DIS: <[[L1:BB[0-9_]+]]>:
|
|
38 ; GFX10: s_sleep 0
|
|
39 ; GFX10: s_cbranch_scc0 [[L1]]
|
|
40 ; GFX10-NEXT: s_endpgm
|
|
41 define amdgpu_kernel void @test_loop_128(i32 addrspace(1)* nocapture %arg) {
|
|
42 bb:
|
|
43 br label %bb2
|
|
44
|
|
45 bb1: ; preds = %bb2
|
|
46 ret void
|
|
47
|
|
48 bb2: ; preds = %bb2, %bb
|
|
49 %tmp1 = phi i32 [ 0, %bb ], [ %tmp2, %bb2 ]
|
|
50 %tmp2 = add nuw nsw i32 %tmp1, 1
|
|
51 %tmp3 = icmp eq i32 %tmp2, 1024
|
|
52 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
53 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
54 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
55 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
56 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
57 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
58 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
59 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
60 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
61 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
62 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
63 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
64 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
65 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
66 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
67 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
68 br i1 %tmp3, label %bb1, label %bb2
|
|
69 }
|
|
70
|
|
71 ; GCN-LABEL: test_loop_192
|
|
72 ; GFX10: s_movk_i32 s{{[0-9]+}}, 0x400
|
|
73 ; GFX10-NEXT: s_inst_prefetch 0x1
|
|
74 ; GFX10-ASM-NEXT: .p2align 6
|
|
75 ; GFX10-DIS-NEXT: s_nop 0
|
|
76 ; GFX10-NOT: s_inst_prefetch
|
|
77 ; GFX10-ASM: [[L1:BB[0-9_]+]]:
|
|
78 ; GFX10-DIS: <[[L1:BB[0-9_]+]]>:
|
|
79 ; GFX10: s_sleep 0
|
|
80 ; GFX10: s_cbranch_scc0 [[L1]]
|
|
81 ; GFX10-NEXT: s_inst_prefetch 0x2
|
|
82 ; GFX10-NEXT: s_endpgm
|
|
83 define amdgpu_kernel void @test_loop_192(i32 addrspace(1)* nocapture %arg) {
|
|
84 bb:
|
|
85 br label %bb2
|
|
86
|
|
87 bb1: ; preds = %bb2
|
|
88 ret void
|
|
89
|
|
90 bb2: ; preds = %bb2, %bb
|
|
91 %tmp1 = phi i32 [ 0, %bb ], [ %tmp2, %bb2 ]
|
|
92 %tmp2 = add nuw nsw i32 %tmp1, 1
|
|
93 %tmp3 = icmp eq i32 %tmp2, 1024
|
|
94 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
95 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
96 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
97 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
98 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
99 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
100 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
101 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
102 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
103 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
104 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
105 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
106 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
107 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
108 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
109 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
110 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
111 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
112 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
113 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
114 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
115 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
116 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
117 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
118 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
119 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
120 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
121 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
122 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
123 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
124 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
125 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
126 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
127 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
128 br i1 %tmp3, label %bb1, label %bb2
|
|
129 }
|
|
130
|
|
131 ; GCN-LABEL: test_loop_256
|
|
132 ; GFX10: s_movk_i32 s{{[0-9]+}}, 0x400
|
|
133 ; GFX10-DIS-NEXT: {{^$}}
|
|
134 ; GFX10-ASM-NEXT: [[L1:BB[0-9_]+]]:
|
|
135 ; GFX10-DIS-NEXT: <[[L1:BB[0-9_]+]]>:
|
|
136 ; GFX10: s_sleep 0
|
|
137 ; GFX10: s_cbranch_scc0 [[L1]]
|
|
138 ; GFX10-NEXT: s_endpgm
|
|
139 define amdgpu_kernel void @test_loop_256(i32 addrspace(1)* nocapture %arg) {
|
|
140 bb:
|
|
141 br label %bb2
|
|
142
|
|
143 bb1: ; preds = %bb2
|
|
144 ret void
|
|
145
|
|
146 bb2: ; preds = %bb2, %bb
|
|
147 %tmp1 = phi i32 [ 0, %bb ], [ %tmp2, %bb2 ]
|
|
148 %tmp2 = add nuw nsw i32 %tmp1, 1
|
|
149 %tmp3 = icmp eq i32 %tmp2, 1024
|
|
150 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
151 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
152 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
153 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
154 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
155 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
156 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
157 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
158 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
159 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
160 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
161 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
162 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
163 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
164 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
165 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
166 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
167 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
168 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
169 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
170 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
171 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
172 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
173 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
174 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
175 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
176 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
177 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
178 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
179 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
180 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
181 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
182 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
183 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
184 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
185 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
186 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
187 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
188 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
189 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
190 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
191 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
192 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
193 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
194 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
195 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
196 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
197 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
198 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
199 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
200 br i1 %tmp3, label %bb1, label %bb2
|
|
201 }
|
|
202
|
|
203 ; GCN-LABEL: test_loop_prefetch_inner_outer
|
|
204 ; GFX10: s_inst_prefetch 0x1
|
|
205 ; GFX10-ASM-NEXT: .p2align 6
|
|
206 ; GFX10-DIS-NEXT: s_nop 0
|
|
207 ; GFX10-NOT: s_inst_prefetch
|
|
208 ; GFX10-ASM: [[L1:BB[0-9_]+]]:
|
|
209 ; GFX10-DIS: <[[L1:BB[0-9_]+]]>:
|
|
210 ; GFX10-NOT: s_inst_prefetch
|
|
211 ; GFX10-ASM: .p2align 6
|
|
212 ; GFX10-DIS: s_nop 0
|
|
213 ; GFX10-NOT: s_inst_prefetch
|
|
214 ; GFX10-ASM: [[L2:BB[0-9_]+]]:
|
|
215 ; GFX10-DIS: <[[L2:BB[0-9_]+]]>:
|
|
216 ; GFX10-NOT: s_inst_prefetch
|
|
217 ; GFX10: s_sleep 0
|
|
218 ; GFX10: s_cbranch_scc{{[01]}} [[L2]]
|
|
219 ; GFX10-NOT: s_inst_prefetch
|
|
220 ; GFX10: s_cbranch_scc{{[01]}} [[L1]]
|
|
221 ; GFX10-NEXT: s_inst_prefetch 0x2
|
|
222 ; GFX10-NEXT: s_endpgm
|
|
223 define amdgpu_kernel void @test_loop_prefetch_inner_outer(i32 addrspace(1)* nocapture %arg) {
|
|
224 bb:
|
|
225 br label %bb2
|
|
226
|
|
227 bb1:
|
|
228 ret void
|
|
229
|
|
230 bb2:
|
|
231 %tmp1 = phi i32 [ 0, %bb ], [ %tmp2, %bb4 ]
|
|
232 %tmp2 = add nuw nsw i32 %tmp1, 1
|
|
233 %tmp3 = icmp eq i32 %tmp2, 1024
|
|
234 br label %bb3
|
|
235
|
|
236 bb3:
|
|
237 %tmp4 = phi i32 [ 0, %bb2 ], [ %tmp5, %bb3 ]
|
|
238 %tmp5 = add nuw nsw i32 %tmp4, 1
|
|
239 %tmp6 = icmp eq i32 %tmp5, 1024
|
|
240 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
241 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
242 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
243 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
244 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
245 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
246 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
247 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
248 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
249 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
250 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
251 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
252 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
253 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
254 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
255 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
256 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
257 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
258 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
259 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
260 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
261 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
262 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
263 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
264 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
265 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
266 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
267 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
268 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
269 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
270 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
271 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
272 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
273 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
274 br i1 %tmp6, label %bb4, label %bb3
|
|
275
|
|
276 bb4:
|
|
277 br i1 %tmp3, label %bb1, label %bb2
|
|
278 }
|
|
279
|
|
280 ; GCN-LABEL: test_loop_prefetch_inner_outer_noouter
|
|
281 ; GFX10-NOT: .p2align 6
|
|
282 ; GFX10-NOT: s_nop
|
|
283 ; GFX10-NOT: s_inst_prefetch
|
|
284 ; GFX10-ASM: [[L0:BB[0-9_]+]]:
|
|
285 ; GFX10-DIS: <[[L0:BB[0-9_]+]]>:
|
|
286 ; GFX10: s_inst_prefetch 0x1
|
|
287 ; GFX10-ASM-NEXT: .p2align 6
|
|
288 ; GFX10-DIS-NEXT: s_nop 0
|
|
289 ; GFX10-NOT: s_inst_prefetch
|
|
290 ; GFX10-ASM: [[L1:BB[0-9_]+]]:
|
|
291 ; GFX10-DIS: <[[L1:BB[0-9_]+]]>:
|
|
292 ; GFX10-NOT: s_inst_prefetch
|
|
293 ; GFX10-ASM: .p2align 6
|
|
294 ; GFX10-DIS: s_nop 0
|
|
295 ; GFX10-NOT: s_inst_prefetch
|
|
296 ; GFX10-ASM: [[L2:BB[0-9_]+]]:
|
|
297 ; GFX10-DIS: <[[L2:BB[0-9_]+]]>:
|
|
298 ; GFX10-NOT: s_inst_prefetch
|
|
299 ; GFX10: s_sleep 0
|
|
300 ; GFX10: s_cbranch_scc{{[01]}} [[L2]]
|
|
301 ; GFX10-NOT: s_inst_prefetch
|
|
302 ; GFX10: s_cbranch_scc{{[01]}} [[L1]]
|
|
303 ; GFX10-NEXT: s_inst_prefetch 0x2
|
|
304 ; GFX10: s_cbranch_scc{{[01]}} [[L0]]
|
|
305 ; GFX10-NEXT: s_endpgm
|
|
306 define amdgpu_kernel void @test_loop_prefetch_inner_outer_noouter(i32 addrspace(1)* nocapture %arg) {
|
|
307 bb:
|
|
308 br label %bb2
|
|
309
|
|
310 bb1:
|
|
311 ret void
|
|
312
|
|
313 bb2:
|
|
314 %tmp1 = phi i32 [ 0, %bb ], [ %tmp2, %bb6 ]
|
|
315 %tmp2 = add nuw nsw i32 %tmp1, 1
|
|
316 %tmp3 = icmp eq i32 %tmp2, 1024
|
|
317 br label %bb3
|
|
318
|
|
319 bb3:
|
|
320 %tmp4 = phi i32 [ 0, %bb2 ], [ %tmp5, %bb5 ]
|
|
321 %tmp5 = add nuw nsw i32 %tmp4, 1
|
|
322 %tmp6 = icmp eq i32 %tmp5, 1024
|
|
323 br label %bb4
|
|
324
|
|
325 bb4:
|
|
326 %tmp7 = phi i32 [ 0, %bb3 ], [ %tmp8, %bb4 ]
|
|
327 %tmp8 = add nuw nsw i32 %tmp7, 1
|
|
328 %tmp9 = icmp eq i32 %tmp8, 1024
|
|
329 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
330 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
331 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
332 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
333 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
334 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
335 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
336 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
337 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
338 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
339 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
340 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
341 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
342 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
343 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
344 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
345 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
346 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
347 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
348 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
349 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
350 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
351 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
352 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
353 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
354 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
355 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
356 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
357 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
358 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
359 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
360 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
361 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
362 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
363 br i1 %tmp9, label %bb5, label %bb4
|
|
364
|
|
365 bb5:
|
|
366 br i1 %tmp6, label %bb6, label %bb3
|
|
367
|
|
368 bb6:
|
|
369 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
370 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
371 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
372 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
373 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
374 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
375 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
376 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
377 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
378 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
379 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
380 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
381 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
382 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
383 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
384 tail call void @llvm.amdgcn.s.sleep(i32 0)
|
|
385 br i1 %tmp3, label %bb1, label %bb2
|
|
386 }
|
|
387
|
|
388 declare void @llvm.amdgcn.s.sleep(i32)
|