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
|
;=========================== begin_copyright_notice ============================
;
; Copyright (C) 2024 Intel Corporation
;
; SPDX-License-Identifier: MIT
;
;============================ end_copyright_notice =============================
; REQUIRES: llvm-14-plus
; RUN: igc_opt --opaque-pointers -igc-fix-alignment -S %s -o %t
; RUN: FileCheck %s --input-file=%t
target datalayout = "e-p:32:32-p1:64:64-p2:64:64-p3:32:32-p4:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v16:16:16-v24:32:32-v32:32:32-v48:64:64-v64:64:64-v96:128:128-v128:128:128-v192:256:256-v256:256:256"
target triple = "spir64-unknown-unknown"
; Function Attrs: alwaysinline nounwind
; CHECK-LABEL: @alloca
define void @alloca() #0 {
entry:
%alloc = alloca <4 x i32>
; CHECK: load
; CHECK: align 16
%val = load <4 x i32>, ptr %alloc
ret void
}
; CHECK-LABEL: @param
define void @param(ptr align 1 %byteptr, ptr align 4 %dwordptr) #0 {
; CHECK: load
; CHECK: align 1
%val1 = load i8, ptr %byteptr
; CHECK: load
; CHECK: align 4
%val4 = load <4 x i8>, ptr %dwordptr
ret void
}
; CHECK-LABEL: @ptrtointtoptr
define void @ptrtointtoptr() #0 {
entry:
%alloc = alloca <4 x i32>
%int = ptrtoint ptr %alloc to i32
%ptr = inttoptr i32 %int to ptr
; CHECK: load
; CHECK: align 16
%val = load i8, ptr %ptr
ret void
}
; CHECK-LABEL: @add
define void @add(i32 %unknown) #0 {
entry:
%alloc = alloca <4 x i32>
%int = ptrtoint ptr %alloc to i32
%add1 = add i32 %int, 1
%ptr1 = inttoptr i32 %add1 to ptr
; CHECK: load
; CHECK: align 1
%val1 = load i8, ptr %ptr1
%add4 = add i32 %int, 4
%ptr4 = inttoptr i32 %add4 to ptr
; CHECK: load
; CHECK: align 4
%val4 = load i8, ptr %ptr4
%add17 = add i32 %int, 17
%ptr17 = inttoptr i32 %add17 to ptr
; CHECK: load
; CHECK: align 1
%val17 = load i8, ptr %ptr17
%add256 = add i32 %int, 256
%ptr256 = inttoptr i32 %add256 to ptr
; CHECK: load
; CHECK: align 16
%val256 = load i8, ptr %ptr256
%addunk = add i32 %int, %unknown
%ptrunk = inttoptr i32 %addunk to ptr
; CHECK: load
; CHECK: align 1
%valunk = load i8, ptr %ptrunk
ret void
}
; CHECK-LABEL: @mul
define void @mul(i32 %unknown) #0 {
entry:
%alloc = alloca <4 x i8>
%int = ptrtoint ptr %alloc to i32
%mul1 = mul i32 %int, 1
%ptr1 = inttoptr i32 %mul1 to ptr
; CHECK: load
; CHECK: align {{1|4}}
%val1 = load i8, ptr %ptr1
%mul4 = mul i32 %int, 4
%ptr4 = inttoptr i32 %mul4 to ptr
; CHECK: load
; CHECK: align {{1|16}}
%val4 = load i8, ptr %ptr4
%mul17 = mul i32 %int, 17
%ptr17 = inttoptr i32 %mul17 to ptr
; CHECK: load
; CHECK: align {{1|4}}
%val17 = load i8, ptr %ptr17
%mul108 = mul i32 %int, 108
%ptr108 = inttoptr i32 %mul108 to ptr
; CHECK: load
; CHECK: align {{1|16}}
%val108 = load i8, ptr %ptr108
%mulunk = mul i32 %int, %unknown
%ptrunk = inttoptr i32 %mulunk to ptr
; CHECK: load
; CHECK: align {{1|4}}
%valunk = load i8, ptr %ptrunk
ret void
}
; CHECK-LABEL: @muladd
define void @muladd(ptr %src, i32 %offset)
{
%int = ptrtoint ptr %src to i32
%mul = mul i32 %offset, 16
%add = add i32 %int, %mul
%ptr = inttoptr i32 %add to ptr
; CHECK: load
; CHECK: align {{1|4}}
%val = load i8, ptr %ptr
ret void
}
; CHECK-LABEL: @shl
define void @shl(i32 %unknown) #0 {
entry:
%alloc = alloca <4 x i8>
%int = ptrtoint ptr %alloc to i32
%shl1 = shl i32 %int, 0
%ptr1 = inttoptr i32 %shl1 to ptr
; CHECK: load
; CHECK: align {{1|4}}
%val1 = load i8, ptr %ptr1
%shl2 = shl i32 %int, 2
%ptr4 = inttoptr i32 %shl2 to ptr
; CHECK: load
; CHECK: align {{1|16}}
%val4 = load i8, ptr %ptr4
%shlunk = shl i32 %int, %unknown
%ptrunk = inttoptr i32 %shlunk to ptr
; CHECK: load
; CHECK: align 4
%valunk = load i8, ptr %ptrunk
ret void
}
; CHECK-LABEL: @and
define void @and (ptr %ptr) #0 {
entry:
%int = ptrtoint ptr %ptr to i32
; ~0x03
%and3 = and i32 %int, 4294967292
%ptr3 = inttoptr i32 %and3 to ptr
; CHECK: load
; CHECK: align {{4|16}}
%val3 = load <16 x i8>, ptr %ptr3
%andsilly = and i32 %int, 255
%ptrsilly = inttoptr i32 %andsilly to ptr
; CHECK: load
; CHECK: align {{1|16}}
%valsilly = load <16 x i8>, ptr %ptrsilly
ret void
}
; CHECK-LABEL: @select
define void @select(ptr align 4 %char4ptr, ptr align 16 %char16ptr, i1 %flag) #0 {
%ptr = select i1 %flag, ptr %char16ptr, ptr %char4ptr
; CHECK: load
; CHECK: align 16
%val = load <16 x i8>, ptr %ptr
ret void
}
; CHECK-LABEL: @loadval
define void @loadval(<4 x i8>* %src, i32* %offset) #0 {
; CHECK: load
; CHECK: align 4
%off = load i32, i32* %offset
%int = ptrtoint <4 x i8>* %src to i32
%add = add i32 %int, %off
%ptr = inttoptr i32 %add to <4 x i8>*
; CHECK: load
; CHECK: align {{1|4}}
%val = load <4 x i8>, <4 x i8>* %ptr
ret void
}
; CHECK-LABEL: @simplephi
define void @simplephi(ptr align 16 %src, i1 %flag) #0 {
%int = ptrtoint ptr %src to i32
br i1 %flag, label %true, label %false
true:
%add4 = add i32 %int, 4
%ptr4 = inttoptr i32 %add4 to ptr
; CHECK: load
; CHECK: align 4
%val4 = load i8, ptr %ptr4
br label %end
false:
%add16 = add i32 %int, 16
%ptr16 = inttoptr i32 %add16 to ptr
; CHECK: load
; CHECK: align 16
%val16 = load i8, ptr %ptr16
br label %end
end:
%p = phi ptr [ %ptr4, %true], [ %ptr16, %false]
; CHECK: store
; CHECK: align 4
store i8 0, ptr %p
ret void
}
; CHECK-LABEL: @loop
define void @loop(<4 x i8>* %src, i1 %flag) #0 {
entry:
br label %body
body:
%reduced = phi <4 x i8>* [ %src, %entry ], [ %addptr, %body ]
%improved = phi <4 x i8>* [ %src, %entry ], [ %mulptr, %body ]
; CHECK: load
; CHECK: align {{1|4}}
%valreduced = load <4 x i8>, <4 x i8>* %reduced
; CHECK: load
; CHECK: align 4
%valimproved = load <4 x i8>, <4 x i8>* %improved
%int = ptrtoint <4 x i8>* %improved to i32
%add = add i32 %int, 1
%addptr = inttoptr i32 %add to <4 x i8>*
%mul = mul i32 %int, 32
%mulptr = inttoptr i32 %mul to <4 x i8>*
br i1 %flag, label %body, label %end
end:
ret void
}
; CHECK-LABEL: @gep1
define void @gep1(ptr %src, i32 %offset) {
%ptr = getelementptr <4 x i8>, ptr %src, i32 %offset
; CHECK: load
; CHECK: align 4
%val = load <4 x i8>, ptr %ptr
ret void
}
; CHECK-LABEL: @gep2
define void @gep2(ptr %src, i32 %offset) {
%ptr = getelementptr i8, ptr %src, i32 %offset
; CHECK: load
; CHECK: align 1
%val = load i8, ptr %ptr
ret void
}
; CHECK-LABEL: @gep3
define void @gep3(ptr align 4 %src) {
%ptr = getelementptr i8, ptr %src, i32 4
; CHECK: load
; CHECK: align 4
%val = load i8, ptr %ptr
ret void
}
; CHECK-LABEL: @globalgep
@globalarr = addrspace(3) global [10 x i32] [i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9], align 16
define void @globalgep() {
%start = getelementptr [10 x i32], ptr addrspace(3) @globalarr, i32 0, i32 0
; CHECK: load
; CHECK: align 16
%val = load i32, ptr addrspace(3) %start
ret void
}
; -----------------------
; Function Attrs: alwaysinline nounwind
; CHECK: @regular
define void @regular(ptr addrspace(1) align 1 %src, ptr addrspace(1) align 4 %dst, <8 x i32> %arg1, <8 x i32> %arg2, i16 %arg3, i16 %arg4, i16 %arg5) #0 {
entry:
%scalar = extractelement <8 x i32> %arg2, i32 0
%groupId = extractelement <8 x i32> %arg1, i32 1
%localSize = extractelement <8 x i32> %arg2, i32 3
%0 = mul i32 %localSize, %groupId
%1 = zext i16 %arg3 to i32
%2 = add i32 %0, %1
%3 = add i32 %2, %scalar
%arrayidx = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %3
; CHECK: load
; CHECK: align 1
%4 = load i8, ptr addrspace(1) %arrayidx
%add = add nsw i32 %3, 1
%arrayidx1 = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %add
; CHECK: load
; CHECK: align 1
%5 = load i8, ptr addrspace(1) %arrayidx1
%add3 = add nsw i32 %3, 2
%arrayidx4 = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %add3
; CHECK: load
; CHECK: align 1
%6 = load i8, ptr addrspace(1) %arrayidx4
%add6 = add nsw i32 %3, 3
%arrayidx7 = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %add6
; CHECK: load
; CHECK: align 1
%7 = load i8, ptr addrspace(1) %arrayidx7
%assembled.vect = insertelement <4 x i8> undef, i8 %4, i32 0
%assembled.vect8 = insertelement <4 x i8> %assembled.vect, i8 %5, i32 1
%assembled.vect9 = insertelement <4 x i8> %assembled.vect8, i8 %6, i32 2
%assembled.vect10 = insertelement <4 x i8> %assembled.vect9, i8 %7, i32 3
%arrayidx9 = getelementptr inbounds <4 x i8>, ptr addrspace(1) %dst, i32 %3
; CHECK: store
; CHECK: align 4
store <4 x i8> %assembled.vect10, ptr addrspace(1) %arrayidx9
ret void
}
; Function Attrs: alwaysinline nounwind
; CHECK: @virtload
define void @virtload(ptr addrspace(1) align 1 %src, ptr addrspace(1) align 4 %dst, <8 x i32> %arg1, <8 x i32> %arg2, i16 %arg3, i16 %arg4, i16 %arg5) #0 {
entry:
%scalar = extractelement <8 x i32> %arg2, i32 0
%groupId = extractelement <8 x i32> %arg1, i32 1
%localSize = extractelement <8 x i32> %arg2, i32 3
%0 = mul i32 %localSize, %groupId
%1 = zext i16 %arg3 to i32
%2 = add i32 %0, %1
%3 = add i32 %2, %scalar
%add.ptr = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %3
; CHECK: load
; CHECK: align {{1|4}}
%4 = load <4 x i8>, ptr addrspace(1) %add.ptr
%arrayidx = getelementptr inbounds <4 x i8>, ptr addrspace(1) %dst, i32 %3
; CHECK: store
; CHECK: align 4
store <4 x i8> %4, ptr addrspace(1) %arrayidx
ret void
}
; Function Attrs: alwaysinline nounwind
; CHECK: @illegal
define void @illegal(ptr addrspace(1) align 1 %src, ptr addrspace(1) align 4 %dst, <8 x i32> %arg1, <8 x i32> %arg2, i16 %arg3, i16 %arg4, i16 %arg5) #0 {
entry:
%scalar = extractelement <8 x i32> %arg2, i32 0
%groupId = extractelement <8 x i32> %arg1, i32 1
%localSize = extractelement <8 x i32> %arg2, i32 3
%0 = mul i32 %localSize, %groupId
%1 = zext i16 %arg3 to i32
%2 = add i32 %0, %1
%3 = add i32 %2, %scalar
%add.ptr = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %3
; CHECK: load
; CHECK: align {{1|4}}
%4 = load <4 x i8>, <4 x i8> addrspace(1)* %add.ptr
%arrayidx = getelementptr inbounds <4 x i8>, ptr addrspace(1) %dst, i32 %3
; CHECK: store
; CHECK: align 4
store <4 x i8> %4, ptr addrspace(1) %arrayidx
ret void
}
; Function Attrs: alwaysinline nounwind
; CHECK: @virtload_aligned
define void @virtload_aligned(ptr addrspace(1) align 4 %src, ptr addrspace(1) align 4 %dst, <8 x i32> %arg1, <8 x i32> %arg2, i16 %arg3, i16 %arg4, i16 %arg5) #0 {
entry:
%scalar = extractelement <8 x i32> %arg2, i32 0
%groupId = extractelement <8 x i32> %arg1, i32 1
%localSize = extractelement <8 x i32> %arg2, i32 3
%0 = mul i32 %localSize, %groupId
%1 = zext i16 %arg3 to i32
%2 = add i32 %0, %1
%3 = add i32 %2, %scalar
%4 = shl i32 %3, 2
%5 = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %4
; CHECK: load
; CHECK: align 4
%6 = load <4 x i8>, ptr addrspace(1) %5
%arrayidx = getelementptr inbounds <4 x i8>, ptr addrspace(1) %dst, i32 %3
; CHECK: store
; CHECK: align 4
store <4 x i8> %6, ptr addrspace(1) %arrayidx
ret void
}
; Function Attrs: alwaysinline nounwind
; CHECK: @aligned
define void @aligned(ptr addrspace(1) align 4 %src, ptr addrspace(1) align 4 %dst, <8 x i32> %arg1, <8 x i32> %arg2, i16 %arg3, i16 %arg4, i16 %arg5) #0 {
entry:
%scalar = extractelement <8 x i32> %arg2, i32 0
%groupId = extractelement <8 x i32> %arg1, i32 1
%localSize = extractelement <8 x i32> %arg2, i32 3
%0 = mul i32 %localSize, %groupId
%1 = zext i16 %arg3 to i32
%2 = add i32 %0, %1
%3 = add i32 %2, %scalar
%add.ptr = getelementptr inbounds <4 x i8>, ptr addrspace(1) %src, i32 %3
; CHECK: load
; CHECK: align 4
%4 = load <4 x i8>, ptr addrspace(1) %add.ptr
%arrayidx = getelementptr inbounds <4 x i8>, ptr addrspace(1) %dst, i32 %3
; CHECK: store
; CHECK: align 4
store <4 x i8> %4, ptr addrspace(1) %arrayidx
ret void
}
; CHECK: @regular_scalarized
define void @regular_scalarized(ptr addrspace(1) align 1 %src, ptr addrspace(1) align 4 %dst, <8 x i32> %arg1, <8 x i32> %arg2, i16 %arg3, i16 %arg4, i16 %arg5) #0 {
entry:
%scalar = extractelement <8 x i32> %arg2, i32 0
%groupId = extractelement <8 x i32> %arg1, i32 1
%localSize = extractelement <8 x i32> %arg2, i32 3
%0 = mul i32 %localSize, %groupId
%1 = zext i16 %arg3 to i32
%2 = add i32 %0, %1
%3 = add i32 %2, %scalar
%arrayidx = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %3
; CHECK: load
; CHECK: align 1
%4 = load i8, ptr addrspace(1) %arrayidx
%add = add nsw i32 %3, 1
%arrayidx1 = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %add
; CHECK: load
; CHECK: align 1
%5 = load i8, ptr addrspace(1) %arrayidx1
%add3 = add nsw i32 %3, 2
%arrayidx4 = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %add3
; CHECK: load
; CHECK: align 1
%6 = load i8, ptr addrspace(1) %arrayidx4
%add6 = add nsw i32 %3, 3
%arrayidx7 = getelementptr inbounds i8, ptr addrspace(1) %src, i32 %add6
; CHECK: load
; CHECK: align 1
%7 = load i8, ptr addrspace(1) %arrayidx7
%ptrVec2ptrScl = getelementptr inbounds <4 x i8>, ptr addrspace(1) %dst, i32 %3, i32 0
; CHECK: store
; CHECK: align 4
store i8 %4, ptr addrspace(1) %ptrVec2ptrScl
%GEP_lane8 = getelementptr <4 x i8>, ptr addrspace(1) %dst, i32 %3, i32 1
; CHECK: store
; CHECK: align 1
store i8 %5, ptr addrspace(1) %GEP_lane8
%GEP_lane9 = getelementptr <4 x i8>, ptr addrspace(1) %dst, i32 %3, i32 2
; CHECK: store
; CHECK: align 2
store i8 %6, ptr addrspace(1) %GEP_lane9
%GEP_lane10 = getelementptr <4 x i8>, ptr addrspace(1) %dst, i32 %3, i32 3
; CHECK: store
; CHECK: align 1
store i8 %7, ptr addrspace(1) %GEP_lane10
ret void
}
attributes #0 = { alwaysinline nounwind }
|
