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
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -march=amdgcn -mcpu=gfx902 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
; RUN: llc -march=amdgcn -mcpu=gfx1030 -amdgpu-enable-flat-scratch -verify-machineinstrs < %s | FileCheck -check-prefix=GCN-SCRATCH %s
define amdgpu_kernel void @vector_clause(<4 x i32> addrspace(1)* noalias nocapture readonly %arg, <4 x i32> addrspace(1)* noalias nocapture %arg1) {
; GCN-LABEL: vector_clause:
; GCN: ; %bb.0: ; %bb
; GCN-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
; GCN-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x2c
; GCN-NEXT: v_lshlrev_b32_e32 v16, 4, v0
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: global_load_dwordx4 v[0:3], v16, s[2:3]
; GCN-NEXT: global_load_dwordx4 v[4:7], v16, s[2:3] offset:16
; GCN-NEXT: global_load_dwordx4 v[8:11], v16, s[2:3] offset:32
; GCN-NEXT: global_load_dwordx4 v[12:15], v16, s[2:3] offset:48
; GCN-NEXT: s_waitcnt vmcnt(3)
; GCN-NEXT: global_store_dwordx4 v16, v[0:3], s[4:5]
; GCN-NEXT: s_waitcnt vmcnt(3)
; GCN-NEXT: global_store_dwordx4 v16, v[4:7], s[4:5] offset:16
; GCN-NEXT: s_waitcnt vmcnt(3)
; GCN-NEXT: global_store_dwordx4 v16, v[8:11], s[4:5] offset:32
; GCN-NEXT: s_waitcnt vmcnt(3)
; GCN-NEXT: global_store_dwordx4 v16, v[12:15], s[4:5] offset:48
; GCN-NEXT: s_endpgm
;
; GCN-SCRATCH-LABEL: vector_clause:
; GCN-SCRATCH: ; %bb.0: ; %bb
; GCN-SCRATCH-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
; GCN-SCRATCH-NEXT: v_lshlrev_b32_e32 v16, 4, v0
; GCN-SCRATCH-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x2c
; GCN-SCRATCH-NEXT: s_waitcnt lgkmcnt(0)
; GCN-SCRATCH-NEXT: s_clause 0x3
; GCN-SCRATCH-NEXT: global_load_dwordx4 v[0:3], v16, s[2:3]
; GCN-SCRATCH-NEXT: global_load_dwordx4 v[4:7], v16, s[2:3] offset:16
; GCN-SCRATCH-NEXT: global_load_dwordx4 v[8:11], v16, s[2:3] offset:32
; GCN-SCRATCH-NEXT: global_load_dwordx4 v[12:15], v16, s[2:3] offset:48
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(3)
; GCN-SCRATCH-NEXT: global_store_dwordx4 v16, v[0:3], s[0:1]
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(2)
; GCN-SCRATCH-NEXT: global_store_dwordx4 v16, v[4:7], s[0:1] offset:16
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(1)
; GCN-SCRATCH-NEXT: global_store_dwordx4 v16, v[8:11], s[0:1] offset:32
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0)
; GCN-SCRATCH-NEXT: global_store_dwordx4 v16, v[12:15], s[0:1] offset:48
; GCN-SCRATCH-NEXT: s_endpgm
bb:
%tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
%tmp2 = zext i32 %tmp to i64
%tmp3 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg, i64 %tmp2
%tmp4 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp3, align 16
%tmp5 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg1, i64 %tmp2
%tmp6 = add nuw nsw i64 %tmp2, 1
%tmp7 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg, i64 %tmp6
%tmp8 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp7, align 16
%tmp9 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg1, i64 %tmp6
%tmp10 = add nuw nsw i64 %tmp2, 2
%tmp11 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg, i64 %tmp10
%tmp12 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp11, align 16
%tmp13 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg1, i64 %tmp10
%tmp14 = add nuw nsw i64 %tmp2, 3
%tmp15 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg, i64 %tmp14
%tmp16 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp15, align 16
%tmp17 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg1, i64 %tmp14
store <4 x i32> %tmp4, <4 x i32> addrspace(1)* %tmp5, align 16
store <4 x i32> %tmp8, <4 x i32> addrspace(1)* %tmp9, align 16
store <4 x i32> %tmp12, <4 x i32> addrspace(1)* %tmp13, align 16
store <4 x i32> %tmp16, <4 x i32> addrspace(1)* %tmp17, align 16
ret void
}
define amdgpu_kernel void @scalar_clause(<4 x i32> addrspace(1)* noalias nocapture readonly %arg, <4 x i32> addrspace(1)* noalias nocapture %arg1) {
; GCN-LABEL: scalar_clause:
; GCN: ; %bb.0: ; %bb
; GCN-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
; GCN-NEXT: s_load_dwordx2 s[16:17], s[0:1], 0x2c
; GCN-NEXT: v_mov_b32_e32 v16, 0
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: s_load_dwordx16 s[0:15], s[2:3], 0x0
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: v_mov_b32_e32 v0, s0
; GCN-NEXT: v_mov_b32_e32 v1, s1
; GCN-NEXT: v_mov_b32_e32 v2, s2
; GCN-NEXT: v_mov_b32_e32 v3, s3
; GCN-NEXT: v_mov_b32_e32 v4, s4
; GCN-NEXT: v_mov_b32_e32 v5, s5
; GCN-NEXT: v_mov_b32_e32 v6, s6
; GCN-NEXT: v_mov_b32_e32 v7, s7
; GCN-NEXT: v_mov_b32_e32 v8, s8
; GCN-NEXT: v_mov_b32_e32 v9, s9
; GCN-NEXT: v_mov_b32_e32 v10, s10
; GCN-NEXT: v_mov_b32_e32 v11, s11
; GCN-NEXT: v_mov_b32_e32 v12, s12
; GCN-NEXT: v_mov_b32_e32 v13, s13
; GCN-NEXT: v_mov_b32_e32 v14, s14
; GCN-NEXT: v_mov_b32_e32 v15, s15
; GCN-NEXT: global_store_dwordx4 v16, v[0:3], s[16:17]
; GCN-NEXT: global_store_dwordx4 v16, v[4:7], s[16:17] offset:16
; GCN-NEXT: global_store_dwordx4 v16, v[8:11], s[16:17] offset:32
; GCN-NEXT: global_store_dwordx4 v16, v[12:15], s[16:17] offset:48
; GCN-NEXT: s_endpgm
;
; GCN-SCRATCH-LABEL: scalar_clause:
; GCN-SCRATCH: ; %bb.0: ; %bb
; GCN-SCRATCH-NEXT: s_clause 0x1
; GCN-SCRATCH-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
; GCN-SCRATCH-NEXT: s_load_dwordx2 s[16:17], s[0:1], 0x2c
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v16, 0
; GCN-SCRATCH-NEXT: s_waitcnt lgkmcnt(0)
; GCN-SCRATCH-NEXT: s_load_dwordx16 s[0:15], s[2:3], 0x0
; GCN-SCRATCH-NEXT: s_waitcnt lgkmcnt(0)
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v0, s0
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v1, s1
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v2, s2
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v3, s3
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v4, s4
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v5, s5
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v6, s6
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v7, s7
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v8, s8
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v9, s9
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v10, s10
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v11, s11
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v12, s12
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v13, s13
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v14, s14
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v15, s15
; GCN-SCRATCH-NEXT: global_store_dwordx4 v16, v[0:3], s[16:17]
; GCN-SCRATCH-NEXT: global_store_dwordx4 v16, v[4:7], s[16:17] offset:16
; GCN-SCRATCH-NEXT: global_store_dwordx4 v16, v[8:11], s[16:17] offset:32
; GCN-SCRATCH-NEXT: global_store_dwordx4 v16, v[12:15], s[16:17] offset:48
; GCN-SCRATCH-NEXT: s_endpgm
bb:
%tmp = load <4 x i32>, <4 x i32> addrspace(1)* %arg, align 16
%tmp2 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg, i64 1
%tmp3 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp2, align 16
%tmp4 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg1, i64 1
%tmp5 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg, i64 2
%tmp6 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp5, align 16
%tmp7 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg1, i64 2
%tmp8 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg, i64 3
%tmp9 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp8, align 16
%tmp10 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg1, i64 3
store <4 x i32> %tmp, <4 x i32> addrspace(1)* %arg1, align 16
store <4 x i32> %tmp3, <4 x i32> addrspace(1)* %tmp4, align 16
store <4 x i32> %tmp6, <4 x i32> addrspace(1)* %tmp7, align 16
store <4 x i32> %tmp9, <4 x i32> addrspace(1)* %tmp10, align 16
ret void
}
define void @mubuf_clause(<4 x i32> addrspace(5)* noalias nocapture readonly %arg, <4 x i32> addrspace(5)* noalias nocapture %arg1) {
; GCN-LABEL: mubuf_clause:
; GCN: ; %bb.0: ; %bb
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_and_b32_e32 v2, 0x3ff, v31
; GCN-NEXT: v_lshlrev_b32_e32 v2, 4, v2
; GCN-NEXT: v_add_u32_e32 v0, v0, v2
; GCN-NEXT: buffer_load_dword v3, v0, s[0:3], 0 offen offset:12
; GCN-NEXT: buffer_load_dword v4, v0, s[0:3], 0 offen offset:8
; GCN-NEXT: buffer_load_dword v5, v0, s[0:3], 0 offen offset:4
; GCN-NEXT: buffer_load_dword v6, v0, s[0:3], 0 offen
; GCN-NEXT: buffer_load_dword v7, v0, s[0:3], 0 offen offset:28
; GCN-NEXT: buffer_load_dword v8, v0, s[0:3], 0 offen offset:24
; GCN-NEXT: buffer_load_dword v9, v0, s[0:3], 0 offen offset:20
; GCN-NEXT: buffer_load_dword v10, v0, s[0:3], 0 offen offset:16
; GCN-NEXT: buffer_load_dword v11, v0, s[0:3], 0 offen offset:44
; GCN-NEXT: buffer_load_dword v12, v0, s[0:3], 0 offen offset:40
; GCN-NEXT: buffer_load_dword v13, v0, s[0:3], 0 offen offset:36
; GCN-NEXT: buffer_load_dword v14, v0, s[0:3], 0 offen offset:32
; GCN-NEXT: buffer_load_dword v15, v0, s[0:3], 0 offen offset:60
; GCN-NEXT: buffer_load_dword v16, v0, s[0:3], 0 offen offset:56
; GCN-NEXT: buffer_load_dword v17, v0, s[0:3], 0 offen offset:52
; GCN-NEXT: s_nop 0
; GCN-NEXT: buffer_load_dword v0, v0, s[0:3], 0 offen offset:48
; GCN-NEXT: v_add_u32_e32 v1, v1, v2
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v3, v1, s[0:3], 0 offen offset:12
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v4, v1, s[0:3], 0 offen offset:8
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v5, v1, s[0:3], 0 offen offset:4
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v6, v1, s[0:3], 0 offen
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v7, v1, s[0:3], 0 offen offset:28
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v8, v1, s[0:3], 0 offen offset:24
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v9, v1, s[0:3], 0 offen offset:20
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v10, v1, s[0:3], 0 offen offset:16
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v11, v1, s[0:3], 0 offen offset:44
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v12, v1, s[0:3], 0 offen offset:40
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v13, v1, s[0:3], 0 offen offset:36
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v14, v1, s[0:3], 0 offen offset:32
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v15, v1, s[0:3], 0 offen offset:60
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v16, v1, s[0:3], 0 offen offset:56
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v17, v1, s[0:3], 0 offen offset:52
; GCN-NEXT: s_waitcnt vmcnt(15)
; GCN-NEXT: buffer_store_dword v0, v1, s[0:3], 0 offen offset:48
; GCN-NEXT: s_waitcnt vmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GCN-SCRATCH-LABEL: mubuf_clause:
; GCN-SCRATCH: ; %bb.0: ; %bb
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-SCRATCH-NEXT: s_waitcnt_vscnt null, 0x0
; GCN-SCRATCH-NEXT: v_and_b32_e32 v2, 0x3ff, v31
; GCN-SCRATCH-NEXT: v_lshlrev_b32_e32 v18, 4, v2
; GCN-SCRATCH-NEXT: v_add_nc_u32_e32 v0, v0, v18
; GCN-SCRATCH-NEXT: s_clause 0x3
; GCN-SCRATCH-NEXT: scratch_load_dwordx4 v[2:5], v0, off
; GCN-SCRATCH-NEXT: scratch_load_dwordx4 v[6:9], v0, off offset:16
; GCN-SCRATCH-NEXT: scratch_load_dwordx4 v[10:13], v0, off offset:32
; GCN-SCRATCH-NEXT: scratch_load_dwordx4 v[14:17], v0, off offset:48
; GCN-SCRATCH-NEXT: v_add_nc_u32_e32 v0, v1, v18
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(3)
; GCN-SCRATCH-NEXT: scratch_store_dwordx4 v0, v[2:5], off
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(2)
; GCN-SCRATCH-NEXT: scratch_store_dwordx4 v0, v[6:9], off offset:16
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(1)
; GCN-SCRATCH-NEXT: scratch_store_dwordx4 v0, v[10:13], off offset:32
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0)
; GCN-SCRATCH-NEXT: scratch_store_dwordx4 v0, v[14:17], off offset:48
; GCN-SCRATCH-NEXT: s_waitcnt_vscnt null, 0x0
; GCN-SCRATCH-NEXT: s_setpc_b64 s[30:31]
bb:
%tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
%tmp2 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(5)* %arg, i32 %tmp
%tmp3 = load <4 x i32>, <4 x i32> addrspace(5)* %tmp2, align 16
%tmp4 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(5)* %arg1, i32 %tmp
%tmp5 = add nuw nsw i32 %tmp, 1
%tmp6 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(5)* %arg, i32 %tmp5
%tmp7 = load <4 x i32>, <4 x i32> addrspace(5)* %tmp6, align 16
%tmp8 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(5)* %arg1, i32 %tmp5
%tmp9 = add nuw nsw i32 %tmp, 2
%tmp10 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(5)* %arg, i32 %tmp9
%tmp11 = load <4 x i32>, <4 x i32> addrspace(5)* %tmp10, align 16
%tmp12 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(5)* %arg1, i32 %tmp9
%tmp13 = add nuw nsw i32 %tmp, 3
%tmp14 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(5)* %arg, i32 %tmp13
%tmp15 = load <4 x i32>, <4 x i32> addrspace(5)* %tmp14, align 16
%tmp16 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(5)* %arg1, i32 %tmp13
store <4 x i32> %tmp3, <4 x i32> addrspace(5)* %tmp4, align 16
store <4 x i32> %tmp7, <4 x i32> addrspace(5)* %tmp8, align 16
store <4 x i32> %tmp11, <4 x i32> addrspace(5)* %tmp12, align 16
store <4 x i32> %tmp15, <4 x i32> addrspace(5)* %tmp16, align 16
ret void
}
define amdgpu_kernel void @vector_clause_indirect(i64 addrspace(1)* noalias nocapture readonly %arg, <4 x i32> addrspace(1)* noalias nocapture readnone %arg1, <4 x i32> addrspace(1)* noalias nocapture %arg2) {
; GCN-LABEL: vector_clause_indirect:
; GCN: ; %bb.0: ; %bb
; GCN-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
; GCN-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x34
; GCN-NEXT: v_lshlrev_b32_e32 v0, 3, v0
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: global_load_dwordx2 v[8:9], v0, s[2:3]
; GCN-NEXT: s_waitcnt vmcnt(0)
; GCN-NEXT: global_load_dwordx4 v[0:3], v[8:9], off
; GCN-NEXT: global_load_dwordx4 v[4:7], v[8:9], off offset:16
; GCN-NEXT: v_mov_b32_e32 v8, 0
; GCN-NEXT: s_waitcnt vmcnt(1)
; GCN-NEXT: global_store_dwordx4 v8, v[0:3], s[4:5]
; GCN-NEXT: s_waitcnt vmcnt(1)
; GCN-NEXT: global_store_dwordx4 v8, v[4:7], s[4:5] offset:16
; GCN-NEXT: s_endpgm
;
; GCN-SCRATCH-LABEL: vector_clause_indirect:
; GCN-SCRATCH: ; %bb.0: ; %bb
; GCN-SCRATCH-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
; GCN-SCRATCH-NEXT: v_lshlrev_b32_e32 v0, 3, v0
; GCN-SCRATCH-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v8, 0
; GCN-SCRATCH-NEXT: s_waitcnt lgkmcnt(0)
; GCN-SCRATCH-NEXT: global_load_dwordx2 v[4:5], v0, s[2:3]
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0)
; GCN-SCRATCH-NEXT: s_clause 0x1
; GCN-SCRATCH-NEXT: global_load_dwordx4 v[0:3], v[4:5], off
; GCN-SCRATCH-NEXT: global_load_dwordx4 v[4:7], v[4:5], off offset:16
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(1)
; GCN-SCRATCH-NEXT: global_store_dwordx4 v8, v[0:3], s[0:1]
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0)
; GCN-SCRATCH-NEXT: global_store_dwordx4 v8, v[4:7], s[0:1] offset:16
; GCN-SCRATCH-NEXT: s_endpgm
bb:
%tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
%tmp3 = zext i32 %tmp to i64
%tmp4 = getelementptr inbounds i64, i64 addrspace(1)* %arg, i64 %tmp3
%tmp5 = bitcast i64 addrspace(1)* %tmp4 to <4 x i32> addrspace(1)* addrspace(1)*
%tmp6 = load <4 x i32> addrspace(1)*, <4 x i32> addrspace(1)* addrspace(1)* %tmp5, align 8
%tmp7 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp6, align 16
%tmp8 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %tmp6, i64 1
%tmp9 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp8, align 16
store <4 x i32> %tmp7, <4 x i32> addrspace(1)* %arg2, align 16
%tmp10 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg2, i64 1
store <4 x i32> %tmp9, <4 x i32> addrspace(1)* %tmp10, align 16
ret void
}
define void @load_global_d16_hi(i16 addrspace(1)* %in, i16 %reg, <2 x i16> addrspace(1)* %out) {
; GCN-LABEL: load_global_d16_hi:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_mov_b32_e32 v5, v2
; GCN-NEXT: global_load_short_d16_hi v5, v[0:1], off
; GCN-NEXT: s_nop 0
; GCN-NEXT: global_load_short_d16_hi v2, v[0:1], off offset:64
; GCN-NEXT: s_waitcnt vmcnt(1)
; GCN-NEXT: global_store_dword v[3:4], v5, off
; GCN-NEXT: s_waitcnt vmcnt(1)
; GCN-NEXT: global_store_dword v[3:4], v2, off offset:128
; GCN-NEXT: s_waitcnt vmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GCN-SCRATCH-LABEL: load_global_d16_hi:
; GCN-SCRATCH: ; %bb.0: ; %entry
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-SCRATCH-NEXT: s_waitcnt_vscnt null, 0x0
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v5, v2
; GCN-SCRATCH-NEXT: s_clause 0x1
; GCN-SCRATCH-NEXT: global_load_short_d16_hi v5, v[0:1], off
; GCN-SCRATCH-NEXT: global_load_short_d16_hi v2, v[0:1], off offset:64
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(1)
; GCN-SCRATCH-NEXT: global_store_dword v[3:4], v5, off
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0)
; GCN-SCRATCH-NEXT: global_store_dword v[3:4], v2, off offset:128
; GCN-SCRATCH-NEXT: s_waitcnt_vscnt null, 0x0
; GCN-SCRATCH-NEXT: s_setpc_b64 s[30:31]
entry:
%gep = getelementptr inbounds i16, i16 addrspace(1)* %in, i64 32
%load1 = load i16, i16 addrspace(1)* %in
%load2 = load i16, i16 addrspace(1)* %gep
%build0 = insertelement <2 x i16> undef, i16 %reg, i32 0
%build1 = insertelement <2 x i16> %build0, i16 %load1, i32 1
store <2 x i16> %build1, <2 x i16> addrspace(1)* %out
%build2 = insertelement <2 x i16> undef, i16 %reg, i32 0
%build3 = insertelement <2 x i16> %build2, i16 %load2, i32 1
%gep2 = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %out, i64 32
store <2 x i16> %build3, <2 x i16> addrspace(1)* %gep2
ret void
}
define void @load_global_d16_lo(i16 addrspace(1)* %in, i32 %reg, <2 x i16> addrspace(1)* %out) {
; GCN-LABEL: load_global_d16_lo:
; GCN: ; %bb.0: ; %entry
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_mov_b32_e32 v5, v2
; GCN-NEXT: global_load_short_d16 v5, v[0:1], off
; GCN-NEXT: s_nop 0
; GCN-NEXT: global_load_short_d16 v2, v[0:1], off offset:64
; GCN-NEXT: s_waitcnt vmcnt(1)
; GCN-NEXT: global_store_dword v[3:4], v5, off
; GCN-NEXT: s_waitcnt vmcnt(1)
; GCN-NEXT: global_store_dword v[3:4], v2, off offset:128
; GCN-NEXT: s_waitcnt vmcnt(0)
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GCN-SCRATCH-LABEL: load_global_d16_lo:
; GCN-SCRATCH: ; %bb.0: ; %entry
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-SCRATCH-NEXT: s_waitcnt_vscnt null, 0x0
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v5, v2
; GCN-SCRATCH-NEXT: s_clause 0x1
; GCN-SCRATCH-NEXT: global_load_short_d16 v5, v[0:1], off
; GCN-SCRATCH-NEXT: global_load_short_d16 v2, v[0:1], off offset:64
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(1)
; GCN-SCRATCH-NEXT: global_store_dword v[3:4], v5, off
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0)
; GCN-SCRATCH-NEXT: global_store_dword v[3:4], v2, off offset:128
; GCN-SCRATCH-NEXT: s_waitcnt_vscnt null, 0x0
; GCN-SCRATCH-NEXT: s_setpc_b64 s[30:31]
entry:
%gep = getelementptr inbounds i16, i16 addrspace(1)* %in, i64 32
%reg.bc1 = bitcast i32 %reg to <2 x i16>
%reg.bc2 = bitcast i32 %reg to <2 x i16>
%load1 = load i16, i16 addrspace(1)* %in
%load2 = load i16, i16 addrspace(1)* %gep
%build1 = insertelement <2 x i16> %reg.bc1, i16 %load1, i32 0
%build2 = insertelement <2 x i16> %reg.bc2, i16 %load2, i32 0
%gep2 = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %out, i64 32
store <2 x i16> %build1, <2 x i16> addrspace(1)* %out
store <2 x i16> %build2, <2 x i16> addrspace(1)* %gep2
ret void
}
define amdgpu_kernel void @flat_scratch_load(float %a, float %b, <8 x i32> %desc) {
; GCN-LABEL: flat_scratch_load:
; GCN: ; %bb.0: ; %.entry
; GCN-NEXT: s_mov_b32 s16, SCRATCH_RSRC_DWORD0
; GCN-NEXT: s_mov_b32 s17, SCRATCH_RSRC_DWORD1
; GCN-NEXT: s_mov_b32 s18, -1
; GCN-NEXT: s_load_dwordx2 s[12:13], s[0:1], 0x24
; GCN-NEXT: s_load_dwordx8 s[4:11], s[0:1], 0x44
; GCN-NEXT: s_mov_b32 s19, 0xe00000
; GCN-NEXT: s_add_u32 s16, s16, s3
; GCN-NEXT: s_addc_u32 s17, s17, 0
; GCN-NEXT: v_mov_b32_e32 v0, 0x40b00000
; GCN-NEXT: buffer_store_dword v0, off, s[16:19], 0 offset:4
; GCN-NEXT: s_waitcnt vmcnt(0)
; GCN-NEXT: s_brev_b32 s0, 1
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: v_mov_b32_e32 v0, s12
; GCN-NEXT: s_mov_b32 s3, 0
; GCN-NEXT: s_mov_b32 s1, s0
; GCN-NEXT: s_mov_b32 s2, s0
; GCN-NEXT: v_mov_b32_e32 v1, s13
; GCN-NEXT: ;;#ASMSTART
; GCN-NEXT: ;;#ASMEND
; GCN-NEXT: buffer_load_dword v2, off, s[16:19], 0 offset:4
; GCN-NEXT: s_nop 0
; GCN-NEXT: image_sample v0, v[0:1], s[4:11], s[0:3] dmask:0x1
; GCN-NEXT: s_waitcnt vmcnt(0)
; GCN-NEXT: v_add_f32_e32 v0, v2, v0
; GCN-NEXT: exp mrt0 v0, off, off, off done vm
; GCN-NEXT: s_endpgm
;
; GCN-SCRATCH-LABEL: flat_scratch_load:
; GCN-SCRATCH: ; %bb.0: ; %.entry
; GCN-SCRATCH-NEXT: s_add_u32 s2, s2, s5
; GCN-SCRATCH-NEXT: s_addc_u32 s3, s3, 0
; GCN-SCRATCH-NEXT: s_setreg_b32 hwreg(HW_REG_FLAT_SCR_LO), s2
; GCN-SCRATCH-NEXT: s_setreg_b32 hwreg(HW_REG_FLAT_SCR_HI), s3
; GCN-SCRATCH-NEXT: s_clause 0x1
; GCN-SCRATCH-NEXT: s_load_dwordx2 s[10:11], s[0:1], 0x24
; GCN-SCRATCH-NEXT: s_load_dwordx8 s[0:7], s[0:1], 0x44
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v0, 0x40b00000
; GCN-SCRATCH-NEXT: s_brev_b32 s8, 1
; GCN-SCRATCH-NEXT: s_mov_b32 s9, s8
; GCN-SCRATCH-NEXT: scratch_store_dword off, v0, off offset:4
; GCN-SCRATCH-NEXT: s_waitcnt_vscnt null, 0x0
; GCN-SCRATCH-NEXT: ;;#ASMSTART
; GCN-SCRATCH-NEXT: ;;#ASMEND
; GCN-SCRATCH-NEXT: scratch_load_dword v2, off, off offset:4
; GCN-SCRATCH-NEXT: s_waitcnt lgkmcnt(0)
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v0, s10
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v1, s11
; GCN-SCRATCH-NEXT: s_mov_b32 s11, 0
; GCN-SCRATCH-NEXT: s_mov_b32 s10, s8
; GCN-SCRATCH-NEXT: image_sample v0, v[0:1], s[0:7], s[8:11] dmask:0x1 dim:SQ_RSRC_IMG_2D
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0)
; GCN-SCRATCH-NEXT: v_add_f32_e32 v0, v2, v0
; GCN-SCRATCH-NEXT: exp mrt0 v0, off, off, off done vm
; GCN-SCRATCH-NEXT: s_endpgm
.entry:
%alloca = alloca float, align 4, addrspace(5)
store volatile float 5.5, float addrspace(5)* %alloca
call void asm sideeffect "", ""()
; There was a bug with flat scratch instructions that do not not use any address registers (ST mode).
; To trigger, the scratch_load has to be immediately before the image_sample in MIR.
%load = load float, float addrspace(5)* %alloca
%val = call <2 x float> @llvm.amdgcn.image.sample.2d.v2f32.f32(i32 9, float %a, float %b, <8 x i32> %desc, <4 x i32> <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 0>, i1 false, i32 0, i32 0)
%val0 = extractelement <2 x float> %val, i32 0
%valadd = fadd float %load, %val0
call void @llvm.amdgcn.exp.f32(i32 immarg 0, i32 immarg 1, float %valadd, float undef, float undef, float undef, i1 immarg true, i1 immarg true)
ret void
}
define amdgpu_kernel void @flat_scratch_load_clause(float %a, float %b, <8 x i32> %desc) {
; GCN-LABEL: flat_scratch_load_clause:
; GCN: ; %bb.0: ; %.entry
; GCN-NEXT: s_mov_b32 s4, SCRATCH_RSRC_DWORD0
; GCN-NEXT: s_mov_b32 s5, SCRATCH_RSRC_DWORD1
; GCN-NEXT: s_mov_b32 s6, -1
; GCN-NEXT: s_mov_b32 s7, 0xe00000
; GCN-NEXT: s_add_u32 s4, s4, s3
; GCN-NEXT: s_addc_u32 s5, s5, 0
; GCN-NEXT: v_mov_b32_e32 v0, 0x40b00000
; GCN-NEXT: buffer_store_dword v0, off, s[4:7], 0 offset:4
; GCN-NEXT: s_waitcnt vmcnt(0)
; GCN-NEXT: v_mov_b32_e32 v0, 0x40d00000
; GCN-NEXT: buffer_store_dword v0, off, s[4:7], 0 offset:8
; GCN-NEXT: s_waitcnt vmcnt(0)
; GCN-NEXT: ;;#ASMSTART
; GCN-NEXT: ;;#ASMEND
; GCN-NEXT: buffer_load_dword v0, off, s[4:7], 0 offset:4
; GCN-NEXT: buffer_load_dword v1, off, s[4:7], 0 offset:8
; GCN-NEXT: s_waitcnt vmcnt(0)
; GCN-NEXT: v_add_f32_e32 v0, v0, v1
; GCN-NEXT: exp mrt0 v0, off, off, off done vm
; GCN-NEXT: s_endpgm
;
; GCN-SCRATCH-LABEL: flat_scratch_load_clause:
; GCN-SCRATCH: ; %bb.0: ; %.entry
; GCN-SCRATCH-NEXT: s_add_u32 s2, s2, s5
; GCN-SCRATCH-NEXT: s_addc_u32 s3, s3, 0
; GCN-SCRATCH-NEXT: s_setreg_b32 hwreg(HW_REG_FLAT_SCR_LO), s2
; GCN-SCRATCH-NEXT: s_setreg_b32 hwreg(HW_REG_FLAT_SCR_HI), s3
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v0, 0x40b00000
; GCN-SCRATCH-NEXT: v_mov_b32_e32 v1, 0x40d00000
; GCN-SCRATCH-NEXT: scratch_store_dword off, v0, off offset:4
; GCN-SCRATCH-NEXT: s_waitcnt_vscnt null, 0x0
; GCN-SCRATCH-NEXT: scratch_store_dword off, v1, off offset:8
; GCN-SCRATCH-NEXT: s_waitcnt_vscnt null, 0x0
; GCN-SCRATCH-NEXT: ;;#ASMSTART
; GCN-SCRATCH-NEXT: ;;#ASMEND
; GCN-SCRATCH-NEXT: s_clause 0x1
; GCN-SCRATCH-NEXT: scratch_load_dword v0, off, off offset:4
; GCN-SCRATCH-NEXT: scratch_load_dword v1, off, off offset:8
; GCN-SCRATCH-NEXT: s_waitcnt vmcnt(0)
; GCN-SCRATCH-NEXT: v_add_f32_e32 v0, v0, v1
; GCN-SCRATCH-NEXT: exp mrt0 v0, off, off, off done vm
; GCN-SCRATCH-NEXT: s_endpgm
.entry:
%alloca = alloca float, align 4, addrspace(5)
%alloca2 = alloca float, align 4, addrspace(5)
store volatile float 5.5, float addrspace(5)* %alloca
store volatile float 6.5, float addrspace(5)* %alloca2
call void asm sideeffect "", ""()
%load0 = load float, float addrspace(5)* %alloca
%load1 = load float, float addrspace(5)* %alloca2
%valadd = fadd float %load0, %load1
call void @llvm.amdgcn.exp.f32(i32 immarg 0, i32 immarg 1, float %valadd, float undef, float undef, float undef, i1 immarg true, i1 immarg true)
ret void
}
declare i32 @llvm.amdgcn.workitem.id.x()
declare void @llvm.amdgcn.exp.f32(i32 immarg, i32 immarg, float, float, float, float, i1 immarg, i1 immarg)
declare <2 x float> @llvm.amdgcn.image.sample.2d.v2f32.f32(i32 immarg, float, float, <8 x i32>, <4 x i32>, i1 immarg, i32 immarg, i32 immarg)
|
