MIMGInstructions.td
42.5 KB
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
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
//===-- MIMGInstructions.td - MIMG Instruction Definitions ----------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
// MIMG-specific encoding families to distinguish between semantically
// equivalent machine instructions with different encoding.
//
// - MIMGEncGfx6: encoding introduced with gfx6 (obsoleted for atomics in gfx8)
// - MIMGEncGfx8: encoding introduced with gfx8 for atomics
// - MIMGEncGfx10Default: gfx default (non-NSA) encoding
// - MIMGEncGfx10NSA: gfx10 NSA encoding
class MIMGEncoding;
def MIMGEncGfx6 : MIMGEncoding;
def MIMGEncGfx8 : MIMGEncoding;
def MIMGEncGfx10Default : MIMGEncoding;
def MIMGEncGfx10NSA : MIMGEncoding;
def MIMGEncoding : GenericEnum {
let FilterClass = "MIMGEncoding";
}
// Represent an ISA-level opcode, independent of the encoding and the
// vdata/vaddr size.
class MIMGBaseOpcode : PredicateControl {
MIMGBaseOpcode BaseOpcode = !cast<MIMGBaseOpcode>(NAME);
bit Store = 0;
bit Atomic = 0;
bit AtomicX2 = 0; // (f)cmpswap
bit Sampler = 0;
bit Gather4 = 0;
bits<8> NumExtraArgs = 0;
bit Gradients = 0;
bit G16 = 0;
bit Coordinates = 1;
bit LodOrClampOrMip = 0;
bit HasD16 = 0;
}
def MIMGBaseOpcode : GenericEnum {
let FilterClass = "MIMGBaseOpcode";
}
def MIMGBaseOpcodesTable : GenericTable {
let FilterClass = "MIMGBaseOpcode";
let CppTypeName = "MIMGBaseOpcodeInfo";
let Fields = ["BaseOpcode", "Store", "Atomic", "AtomicX2", "Sampler",
"Gather4", "NumExtraArgs", "Gradients", "G16", "Coordinates",
"LodOrClampOrMip", "HasD16"];
GenericEnum TypeOf_BaseOpcode = MIMGBaseOpcode;
let PrimaryKey = ["BaseOpcode"];
let PrimaryKeyName = "getMIMGBaseOpcodeInfo";
}
def MIMGDim : GenericEnum {
let FilterClass = "AMDGPUDimProps";
}
def MIMGDimInfoTable : GenericTable {
let FilterClass = "AMDGPUDimProps";
let CppTypeName = "MIMGDimInfo";
let Fields = ["Dim", "NumCoords", "NumGradients", "DA", "Encoding", "AsmSuffix"];
GenericEnum TypeOf_Dim = MIMGDim;
let PrimaryKey = ["Dim"];
let PrimaryKeyName = "getMIMGDimInfo";
}
def getMIMGDimInfoByEncoding : SearchIndex {
let Table = MIMGDimInfoTable;
let Key = ["Encoding"];
}
def getMIMGDimInfoByAsmSuffix : SearchIndex {
let Table = MIMGDimInfoTable;
let Key = ["AsmSuffix"];
}
class mimg <bits<8> si_gfx10, bits<8> vi = si_gfx10> {
field bits<8> SI_GFX10 = si_gfx10;
field bits<8> VI = vi;
}
class MIMGLZMapping<MIMGBaseOpcode l, MIMGBaseOpcode lz> {
MIMGBaseOpcode L = l;
MIMGBaseOpcode LZ = lz;
}
def MIMGLZMappingTable : GenericTable {
let FilterClass = "MIMGLZMapping";
let CppTypeName = "MIMGLZMappingInfo";
let Fields = ["L", "LZ"];
GenericEnum TypeOf_L = MIMGBaseOpcode;
GenericEnum TypeOf_LZ = MIMGBaseOpcode;
let PrimaryKey = ["L"];
let PrimaryKeyName = "getMIMGLZMappingInfo";
}
class MIMGMIPMapping<MIMGBaseOpcode mip, MIMGBaseOpcode nonmip> {
MIMGBaseOpcode MIP = mip;
MIMGBaseOpcode NONMIP = nonmip;
}
def MIMGMIPMappingTable : GenericTable {
let FilterClass = "MIMGMIPMapping";
let CppTypeName = "MIMGMIPMappingInfo";
let Fields = ["MIP", "NONMIP"];
GenericEnum TypeOf_MIP = MIMGBaseOpcode;
GenericEnum TypeOf_NONMIP = MIMGBaseOpcode;
let PrimaryKey = ["MIP"];
let PrimaryKeyName = "getMIMGMIPMappingInfo";
}
class MIMGG16Mapping<MIMGBaseOpcode g, MIMGBaseOpcode g16> {
MIMGBaseOpcode G = g;
MIMGBaseOpcode G16 = g16;
}
def MIMGG16MappingTable : GenericTable {
let FilterClass = "MIMGG16Mapping";
let CppTypeName = "MIMGG16MappingInfo";
let Fields = ["G", "G16"];
GenericEnum TypeOf_G = MIMGBaseOpcode;
GenericEnum TypeOf_G16 = MIMGBaseOpcode;
let PrimaryKey = ["G"];
let PrimaryKeyName = "getMIMGG16MappingInfo";
}
class MIMG_Base <dag outs, string dns = "">
: InstSI <outs, (ins), "", []> {
let VM_CNT = 1;
let EXP_CNT = 1;
let MIMG = 1;
let Uses = [EXEC];
let mayLoad = 1;
let mayStore = 0;
let SchedRW = [WriteVMEM];
let UseNamedOperandTable = 1;
let hasSideEffects = 0; // XXX ????
let DecoderNamespace = dns;
let isAsmParserOnly = !if(!eq(dns,""), 1, 0);
}
class MIMG <dag outs, string dns = "">
: MIMG_Base <outs, dns> {
let hasPostISelHook = 1;
let AsmMatchConverter = "cvtMIMG";
Instruction Opcode = !cast<Instruction>(NAME);
MIMGBaseOpcode BaseOpcode;
MIMGEncoding MIMGEncoding;
bits<8> VDataDwords;
bits<8> VAddrDwords;
}
def MIMGInfoTable : GenericTable {
let FilterClass = "MIMG";
let CppTypeName = "MIMGInfo";
let Fields = ["Opcode", "BaseOpcode", "MIMGEncoding", "VDataDwords", "VAddrDwords"];
GenericEnum TypeOf_BaseOpcode = MIMGBaseOpcode;
GenericEnum TypeOf_MIMGEncoding = MIMGEncoding;
let PrimaryKey = ["BaseOpcode", "MIMGEncoding", "VDataDwords", "VAddrDwords"];
let PrimaryKeyName = "getMIMGOpcodeHelper";
}
def getMIMGInfo : SearchIndex {
let Table = MIMGInfoTable;
let Key = ["Opcode"];
}
// This is a separate class so that TableGen memoizes the computations.
class MIMGNSAHelper<int num_addrs> {
list<string> AddrAsmNames =
!foldl([]<string>, [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11], lhs, i,
!if(!lt(i, num_addrs), !listconcat(lhs, ["vaddr"#!size(lhs)]), lhs));
dag AddrIns = !dag(ins, !foreach(arg, AddrAsmNames, VGPR_32), AddrAsmNames);
string AddrAsm = "[" # !foldl("$" # !head(AddrAsmNames), !tail(AddrAsmNames), lhs, rhs,
lhs # ", $" # rhs) # "]";
int NSA = !if(!le(num_addrs, 1), ?,
!if(!le(num_addrs, 5), 1,
!if(!le(num_addrs, 9), 2,
!if(!le(num_addrs, 13), 3, ?))));
}
// Base class of all pre-gfx10 MIMG instructions.
class MIMG_gfx6789<bits<8> op, dag outs, string dns = "">
: MIMG<outs, dns>, MIMGe_gfx6789<op> {
let SubtargetPredicate = isGFX6GFX7GFX8GFX9;
let AssemblerPredicate = isGFX6GFX7GFX8GFX9;
let MIMGEncoding = MIMGEncGfx6;
let d16 = !if(BaseOpcode.HasD16, ?, 0);
}
// Base class of all non-NSA gfx10 MIMG instructions.
class MIMG_gfx10<int op, dag outs, string dns = "">
: MIMG<outs, dns>, MIMGe_gfx10<op> {
let SubtargetPredicate = isGFX10Plus;
let AssemblerPredicate = isGFX10Plus;
let MIMGEncoding = MIMGEncGfx10Default;
let d16 = !if(BaseOpcode.HasD16, ?, 0);
let nsa = 0;
}
// Base class for all NSA MIMG instructions. Note that 1-dword addresses always
// use non-NSA variants.
class MIMG_nsa_gfx10<int op, dag outs, int num_addrs, string dns="">
: MIMG<outs, dns>, MIMGe_gfx10<op> {
let SubtargetPredicate = isGFX10Plus;
let AssemblerPredicate = isGFX10Plus;
let MIMGEncoding = MIMGEncGfx10NSA;
MIMGNSAHelper nsah = MIMGNSAHelper<num_addrs>;
dag AddrIns = nsah.AddrIns;
string AddrAsm = nsah.AddrAsm;
let d16 = !if(BaseOpcode.HasD16, ?, 0);
let nsa = nsah.NSA;
}
class MIMG_NoSampler_Helper <bits<8> op, string asm,
RegisterClass dst_rc,
RegisterClass addr_rc,
string dns="">
: MIMG_gfx6789 <op, (outs dst_rc:$vdata), dns> {
let InOperandList = !con((ins addr_rc:$vaddr, SReg_256:$srsrc,
DMask:$dmask, UNorm:$unorm, GLC:$glc, SLC:$slc,
R128A16:$r128, TFE:$tfe, LWE:$lwe, DA:$da),
!if(BaseOpcode.HasD16, (ins D16:$d16), (ins)));
let AsmString = asm#" $vdata, $vaddr, $srsrc$dmask$unorm$glc$slc$r128$tfe$lwe$da"
#!if(BaseOpcode.HasD16, "$d16", "");
}
class MIMG_NoSampler_gfx10<int op, string opcode,
RegisterClass DataRC, RegisterClass AddrRC,
string dns="">
: MIMG_gfx10<op, (outs DataRC:$vdata), dns> {
let InOperandList = !con((ins AddrRC:$vaddr0, SReg_256:$srsrc, DMask:$dmask,
Dim:$dim, UNorm:$unorm, DLC:$dlc, GLC:$glc,
SLC:$slc, R128A16:$r128, GFX10A16:$a16, TFE:$tfe, LWE:$lwe),
!if(BaseOpcode.HasD16, (ins D16:$d16), (ins)));
let AsmString = opcode#" $vdata, $vaddr0, $srsrc$dmask$dim$unorm$dlc$glc$slc$r128$a16$tfe$lwe"
#!if(BaseOpcode.HasD16, "$d16", "");
}
class MIMG_NoSampler_nsa_gfx10<int op, string opcode,
RegisterClass DataRC, int num_addrs,
string dns="">
: MIMG_nsa_gfx10<op, (outs DataRC:$vdata), num_addrs, dns> {
let InOperandList = !con(AddrIns,
(ins SReg_256:$srsrc, DMask:$dmask,
Dim:$dim, UNorm:$unorm, DLC:$dlc, GLC:$glc,
SLC:$slc, R128A16:$r128, GFX10A16:$a16, TFE:$tfe, LWE:$lwe),
!if(BaseOpcode.HasD16, (ins D16:$d16), (ins)));
let AsmString = opcode#" $vdata, "#AddrAsm#", $srsrc$dmask$dim$unorm$dlc$glc$slc$r128$a16$tfe$lwe"
#!if(BaseOpcode.HasD16, "$d16", "");
}
multiclass MIMG_NoSampler_Src_Helper <bits<8> op, string asm,
RegisterClass dst_rc,
bit enableDisasm> {
let ssamp = 0 in {
let VAddrDwords = 1 in {
def _V1 : MIMG_NoSampler_Helper <op, asm, dst_rc, VGPR_32,
!if(enableDisasm, "AMDGPU", "")>;
def _V1_gfx10 : MIMG_NoSampler_gfx10<op, asm, dst_rc, VGPR_32,
!if(enableDisasm, "AMDGPU", "")>;
}
let VAddrDwords = 2 in {
def _V2 : MIMG_NoSampler_Helper <op, asm, dst_rc, VReg_64>;
def _V2_gfx10 : MIMG_NoSampler_gfx10<op, asm, dst_rc, VReg_64>;
def _V2_nsa_gfx10 : MIMG_NoSampler_nsa_gfx10<op, asm, dst_rc, 2>;
}
let VAddrDwords = 3 in {
def _V3 : MIMG_NoSampler_Helper <op, asm, dst_rc, VReg_96>;
def _V3_gfx10 : MIMG_NoSampler_gfx10<op, asm, dst_rc, VReg_96>;
def _V3_nsa_gfx10 : MIMG_NoSampler_nsa_gfx10<op, asm, dst_rc, 3>;
}
let VAddrDwords = 4 in {
def _V4 : MIMG_NoSampler_Helper <op, asm, dst_rc, VReg_128>;
def _V4_gfx10 : MIMG_NoSampler_gfx10<op, asm, dst_rc, VReg_128>;
def _V4_nsa_gfx10 : MIMG_NoSampler_nsa_gfx10<op, asm, dst_rc, 4,
!if(enableDisasm, "AMDGPU", "")>;
}
}
}
multiclass MIMG_NoSampler <bits<8> op, string asm, bit has_d16, bit mip = 0,
bit isResInfo = 0> {
def "" : MIMGBaseOpcode {
let Coordinates = !if(isResInfo, 0, 1);
let LodOrClampOrMip = mip;
let HasD16 = has_d16;
}
let BaseOpcode = !cast<MIMGBaseOpcode>(NAME),
mayLoad = !if(isResInfo, 0, 1) in {
let VDataDwords = 1 in
defm _V1 : MIMG_NoSampler_Src_Helper <op, asm, VGPR_32, 1>;
let VDataDwords = 2 in
defm _V2 : MIMG_NoSampler_Src_Helper <op, asm, VReg_64, 0>;
let VDataDwords = 3 in
defm _V3 : MIMG_NoSampler_Src_Helper <op, asm, VReg_96, 0>;
let VDataDwords = 4 in
defm _V4 : MIMG_NoSampler_Src_Helper <op, asm, VReg_128, 0>;
let VDataDwords = 5 in
defm _V5 : MIMG_NoSampler_Src_Helper <op, asm, VReg_160, 0>;
}
}
class MIMG_Store_Helper <bits<8> op, string asm,
RegisterClass data_rc,
RegisterClass addr_rc,
string dns = "">
: MIMG_gfx6789<op, (outs), dns> {
let InOperandList = !con((ins data_rc:$vdata, addr_rc:$vaddr, SReg_256:$srsrc,
DMask:$dmask, UNorm:$unorm, GLC:$glc, SLC:$slc,
R128A16:$r128, TFE:$tfe, LWE:$lwe, DA:$da),
!if(BaseOpcode.HasD16, (ins D16:$d16), (ins)));
let AsmString = asm#" $vdata, $vaddr, $srsrc$dmask$unorm$glc$slc$r128$tfe$lwe$da"
#!if(BaseOpcode.HasD16, "$d16", "");
}
class MIMG_Store_gfx10<int op, string opcode,
RegisterClass DataRC, RegisterClass AddrRC,
string dns="">
: MIMG_gfx10<op, (outs), dns> {
let InOperandList = !con((ins DataRC:$vdata, AddrRC:$vaddr0, SReg_256:$srsrc,
DMask:$dmask, Dim:$dim, UNorm:$unorm, DLC:$dlc,
GLC:$glc, SLC:$slc, R128A16:$r128, GFX10A16:$a16, TFE:$tfe, LWE:$lwe),
!if(BaseOpcode.HasD16, (ins D16:$d16), (ins)));
let AsmString = opcode#" $vdata, $vaddr0, $srsrc$dmask$dim$unorm$dlc$glc$slc$r128$a16$tfe$lwe"
#!if(BaseOpcode.HasD16, "$d16", "");
}
class MIMG_Store_nsa_gfx10<int op, string opcode,
RegisterClass DataRC, int num_addrs,
string dns="">
: MIMG_nsa_gfx10<op, (outs), num_addrs, dns> {
let InOperandList = !con((ins DataRC:$vdata),
AddrIns,
(ins SReg_256:$srsrc, DMask:$dmask,
Dim:$dim, UNorm:$unorm, DLC:$dlc, GLC:$glc,
SLC:$slc, R128A16:$r128, GFX10A16:$a16, TFE:$tfe, LWE:$lwe),
!if(BaseOpcode.HasD16, (ins D16:$d16), (ins)));
let AsmString = opcode#" $vdata, "#AddrAsm#", $srsrc$dmask$dim$unorm$dlc$glc$slc$r128$a16$tfe$lwe"
#!if(BaseOpcode.HasD16, "$d16", "");
}
multiclass MIMG_Store_Addr_Helper <int op, string asm,
RegisterClass data_rc,
bit enableDisasm> {
let mayLoad = 0, mayStore = 1, hasSideEffects = 0, hasPostISelHook = 0,
DisableWQM = 1, ssamp = 0 in {
let VAddrDwords = 1 in {
def _V1 : MIMG_Store_Helper <op, asm, data_rc, VGPR_32,
!if(enableDisasm, "AMDGPU", "")>;
def _V1_gfx10 : MIMG_Store_gfx10 <op, asm, data_rc, VGPR_32,
!if(enableDisasm, "AMDGPU", "")>;
}
let VAddrDwords = 2 in {
def _V2 : MIMG_Store_Helper <op, asm, data_rc, VReg_64>;
def _V2_gfx10 : MIMG_Store_gfx10 <op, asm, data_rc, VReg_64>;
def _V2_nsa_gfx10 : MIMG_Store_nsa_gfx10 <op, asm, data_rc, 2>;
}
let VAddrDwords = 3 in {
def _V3 : MIMG_Store_Helper <op, asm, data_rc, VReg_96>;
def _V3_gfx10 : MIMG_Store_gfx10 <op, asm, data_rc, VReg_96>;
def _V3_nsa_gfx10 : MIMG_Store_nsa_gfx10 <op, asm, data_rc, 3>;
}
let VAddrDwords = 4 in {
def _V4 : MIMG_Store_Helper <op, asm, data_rc, VReg_128>;
def _V4_gfx10 : MIMG_Store_gfx10 <op, asm, data_rc, VReg_128>;
def _V4_nsa_gfx10 : MIMG_Store_nsa_gfx10 <op, asm, data_rc, 4,
!if(enableDisasm, "AMDGPU", "")>;
}
}
}
multiclass MIMG_Store <bits<8> op, string asm, bit has_d16, bit mip = 0> {
def "" : MIMGBaseOpcode {
let Store = 1;
let LodOrClampOrMip = mip;
let HasD16 = has_d16;
}
let BaseOpcode = !cast<MIMGBaseOpcode>(NAME) in {
let VDataDwords = 1 in
defm _V1 : MIMG_Store_Addr_Helper <op, asm, VGPR_32, 1>;
let VDataDwords = 2 in
defm _V2 : MIMG_Store_Addr_Helper <op, asm, VReg_64, 0>;
let VDataDwords = 3 in
defm _V3 : MIMG_Store_Addr_Helper <op, asm, VReg_96, 0>;
let VDataDwords = 4 in
defm _V4 : MIMG_Store_Addr_Helper <op, asm, VReg_128, 0>;
}
}
class MIMG_Atomic_gfx6789_base <bits<8> op, string asm, RegisterClass data_rc,
RegisterClass addr_rc, string dns="">
: MIMG_gfx6789 <op, (outs data_rc:$vdst), dns> {
let Constraints = "$vdst = $vdata";
let AsmMatchConverter = "cvtMIMGAtomic";
let InOperandList = (ins data_rc:$vdata, addr_rc:$vaddr, SReg_256:$srsrc,
DMask:$dmask, UNorm:$unorm, GLC:$glc, SLC:$slc,
R128A16:$r128, TFE:$tfe, LWE:$lwe, DA:$da);
let AsmString = asm#" $vdst, $vaddr, $srsrc$dmask$unorm$glc$slc$r128$tfe$lwe$da";
}
class MIMG_Atomic_si<mimg op, string asm, RegisterClass data_rc,
RegisterClass addr_rc, bit enableDasm = 0>
: MIMG_Atomic_gfx6789_base<op.SI_GFX10, asm, data_rc, addr_rc,
!if(enableDasm, "GFX6GFX7", "")> {
let AssemblerPredicate = isGFX6GFX7;
}
class MIMG_Atomic_vi<mimg op, string asm, RegisterClass data_rc,
RegisterClass addr_rc, bit enableDasm = 0>
: MIMG_Atomic_gfx6789_base<op.VI, asm, data_rc, addr_rc, !if(enableDasm, "GFX8", "")> {
let AssemblerPredicate = isGFX8GFX9;
let MIMGEncoding = MIMGEncGfx8;
}
class MIMG_Atomic_gfx10<mimg op, string opcode,
RegisterClass DataRC, RegisterClass AddrRC,
bit enableDisasm = 0>
: MIMG_gfx10<!cast<int>(op.SI_GFX10), (outs DataRC:$vdst),
!if(enableDisasm, "AMDGPU", "")> {
let Constraints = "$vdst = $vdata";
let AsmMatchConverter = "cvtMIMGAtomic";
let InOperandList = (ins DataRC:$vdata, AddrRC:$vaddr0, SReg_256:$srsrc,
DMask:$dmask, Dim:$dim, UNorm:$unorm, DLC:$dlc,
GLC:$glc, SLC:$slc, R128A16:$r128, GFX10A16:$a16, TFE:$tfe, LWE:$lwe);
let AsmString = opcode#" $vdst, $vaddr0, $srsrc$dmask$dim$unorm$dlc$glc$slc$r128$a16$tfe$lwe";
}
class MIMG_Atomic_nsa_gfx10<mimg op, string opcode,
RegisterClass DataRC, int num_addrs,
bit enableDisasm = 0>
: MIMG_nsa_gfx10<!cast<int>(op.SI_GFX10), (outs DataRC:$vdst), num_addrs,
!if(enableDisasm, "AMDGPU", "")> {
let Constraints = "$vdst = $vdata";
let AsmMatchConverter = "cvtMIMGAtomic";
let InOperandList = !con((ins DataRC:$vdata),
AddrIns,
(ins SReg_256:$srsrc, DMask:$dmask,
Dim:$dim, UNorm:$unorm, DLC:$dlc, GLC:$glc,
SLC:$slc, R128A16:$r128, GFX10A16:$a16, TFE:$tfe, LWE:$lwe));
let AsmString = opcode#" $vdata, "#AddrAsm#", $srsrc$dmask$dim$unorm$dlc$glc$slc$r128$a16$tfe$lwe";
}
multiclass MIMG_Atomic_Addr_Helper_m <mimg op, string asm,
RegisterClass data_rc,
bit enableDasm = 0> {
let hasSideEffects = 1, // FIXME: remove this
mayLoad = 1, mayStore = 1, hasPostISelHook = 0, DisableWQM = 1,
ssamp = 0 in {
let VAddrDwords = 1 in {
def _V1_si : MIMG_Atomic_si <op, asm, data_rc, VGPR_32, enableDasm>;
def _V1_vi : MIMG_Atomic_vi <op, asm, data_rc, VGPR_32, enableDasm>;
def _V1_gfx10 : MIMG_Atomic_gfx10 <op, asm, data_rc, VGPR_32, enableDasm>;
}
let VAddrDwords = 2 in {
def _V2_si : MIMG_Atomic_si <op, asm, data_rc, VReg_64, 0>;
def _V2_vi : MIMG_Atomic_vi <op, asm, data_rc, VReg_64, 0>;
def _V2_gfx10 : MIMG_Atomic_gfx10 <op, asm, data_rc, VReg_64, 0>;
def _V2_nsa_gfx10 : MIMG_Atomic_nsa_gfx10 <op, asm, data_rc, 2, 0>;
}
let VAddrDwords = 3 in {
def _V3_si : MIMG_Atomic_si <op, asm, data_rc, VReg_96, 0>;
def _V3_vi : MIMG_Atomic_vi <op, asm, data_rc, VReg_96, 0>;
def _V3_gfx10 : MIMG_Atomic_gfx10 <op, asm, data_rc, VReg_96, 0>;
def _V3_nsa_gfx10 : MIMG_Atomic_nsa_gfx10 <op, asm, data_rc, 3, 0>;
}
let VAddrDwords = 4 in {
def _V4_si : MIMG_Atomic_si <op, asm, data_rc, VReg_128, 0>;
def _V4_vi : MIMG_Atomic_vi <op, asm, data_rc, VReg_128, 0>;
def _V4_gfx10 : MIMG_Atomic_gfx10 <op, asm, data_rc, VReg_128, 0>;
def _V4_nsa_gfx10 : MIMG_Atomic_nsa_gfx10 <op, asm, data_rc, 4, enableDasm>;
}
}
}
multiclass MIMG_Atomic <mimg op, string asm, bit isCmpSwap = 0> { // 64-bit atomics
def "" : MIMGBaseOpcode {
let Atomic = 1;
let AtomicX2 = isCmpSwap;
}
let BaseOpcode = !cast<MIMGBaseOpcode>(NAME) in {
// _V* variants have different dst size, but the size is encoded implicitly,
// using dmask and tfe. Only 32-bit variant is registered with disassembler.
// Other variants are reconstructed by disassembler using dmask and tfe.
let VDataDwords = !if(isCmpSwap, 2, 1) in
defm _V1 : MIMG_Atomic_Addr_Helper_m <op, asm, !if(isCmpSwap, VReg_64, VGPR_32), 1>;
let VDataDwords = !if(isCmpSwap, 4, 2) in
defm _V2 : MIMG_Atomic_Addr_Helper_m <op, asm, !if(isCmpSwap, VReg_128, VReg_64)>;
}
}
class MIMG_Sampler_Helper <bits<8> op, string asm, RegisterClass dst_rc,
RegisterClass src_rc, string dns="">
: MIMG_gfx6789 <op, (outs dst_rc:$vdata), dns> {
let InOperandList = !con((ins src_rc:$vaddr, SReg_256:$srsrc, SReg_128:$ssamp,
DMask:$dmask, UNorm:$unorm, GLC:$glc, SLC:$slc,
R128A16:$r128, TFE:$tfe, LWE:$lwe, DA:$da),
!if(BaseOpcode.HasD16, (ins D16:$d16), (ins)));
let AsmString = asm#" $vdata, $vaddr, $srsrc, $ssamp$dmask$unorm$glc$slc$r128$tfe$lwe$da"
#!if(BaseOpcode.HasD16, "$d16", "");
}
class MIMG_Sampler_gfx10<int op, string opcode,
RegisterClass DataRC, RegisterClass AddrRC,
string dns="">
: MIMG_gfx10<op, (outs DataRC:$vdata), dns> {
let InOperandList = !con((ins AddrRC:$vaddr0, SReg_256:$srsrc, SReg_128:$ssamp,
DMask:$dmask, Dim:$dim, UNorm:$unorm, DLC:$dlc,
GLC:$glc, SLC:$slc, R128A16:$r128, GFX10A16:$a16, TFE:$tfe, LWE:$lwe),
!if(BaseOpcode.HasD16, (ins D16:$d16), (ins)));
let AsmString = opcode#" $vdata, $vaddr0, $srsrc, $ssamp$dmask$dim$unorm"
#"$dlc$glc$slc$r128$a16$tfe$lwe"
#!if(BaseOpcode.HasD16, "$d16", "");
}
class MIMG_Sampler_nsa_gfx10<int op, string opcode,
RegisterClass DataRC, int num_addrs,
string dns="">
: MIMG_nsa_gfx10<op, (outs DataRC:$vdata), num_addrs, dns> {
let InOperandList = !con(AddrIns,
(ins SReg_256:$srsrc, SReg_128:$ssamp, DMask:$dmask,
Dim:$dim, UNorm:$unorm, DLC:$dlc, GLC:$glc,
SLC:$slc, R128A16:$r128, GFX10A16:$a16, TFE:$tfe, LWE:$lwe),
!if(BaseOpcode.HasD16, (ins D16:$d16), (ins)));
let AsmString = opcode#" $vdata, "#AddrAsm#", $srsrc, $ssamp$dmask$dim$unorm"
#"$dlc$glc$slc$r128$a16$tfe$lwe"
#!if(BaseOpcode.HasD16, "$d16", "");
}
class MIMGAddrSize<int dw, bit enable_disasm> {
int NumWords = dw;
RegisterClass RegClass = !if(!le(NumWords, 0), ?,
!if(!eq(NumWords, 1), VGPR_32,
!if(!eq(NumWords, 2), VReg_64,
!if(!eq(NumWords, 3), VReg_96,
!if(!eq(NumWords, 4), VReg_128,
!if(!le(NumWords, 8), VReg_256,
!if(!le(NumWords, 16), VReg_512, ?)))))));
// Whether the instruction variant with this vaddr size should be enabled for
// the auto-generated disassembler.
bit Disassemble = enable_disasm;
}
// Return whether x is in lst.
class isIntInList<int x, list<int> lst> {
bit ret = !foldl(0, lst, lhs, y, !or(lhs, !eq(x, y)));
}
// Return whether a value inside the range [min, max] (endpoints inclusive)
// is in the given list.
class isRangeInList<int min, int max, list<int> lst> {
bit ret = !foldl(0, lst, lhs, y, !or(lhs, !and(!le(min, y), !le(y, max))));
}
class MIMGAddrSizes_tmp<list<MIMGAddrSize> lst, int min> {
list<MIMGAddrSize> List = lst;
int Min = min;
}
class MIMG_Sampler_AddrSizes<AMDGPUSampleVariant sample> {
// List of all possible numbers of address words, taking all combinations of
// A16 and image dimension into account (note: no MSAA, since this is for
// sample/gather ops).
list<int> AllNumAddrWords =
!foreach(dw, !if(sample.Gradients,
!if(!eq(sample.LodOrClamp, ""),
[2, 3, 4, 5, 6, 7, 9],
[2, 3, 4, 5, 7, 8, 10]),
!if(!eq(sample.LodOrClamp, ""),
[1, 2, 3],
[1, 2, 3, 4])),
!add(dw, !size(sample.ExtraAddrArgs)));
// Generate machine instructions based on possible register classes for the
// required numbers of address words. The disassembler defaults to the
// smallest register class.
list<MIMGAddrSize> MachineInstrs =
!foldl(MIMGAddrSizes_tmp<[], 0>, [1, 2, 3, 4, 8, 16], lhs, dw,
!if(isRangeInList<lhs.Min, dw, AllNumAddrWords>.ret,
MIMGAddrSizes_tmp<
!listconcat(lhs.List, [MIMGAddrSize<dw, !empty(lhs.List)>]),
!if(!eq(dw, 3), 3, !add(dw, 1))>, // we still need _V4 for codegen w/ 3 dwords
lhs)).List;
// For NSA, generate machine instructions for all possible numbers of words
// except 1 (which is already covered by the non-NSA case).
// The disassembler defaults to the largest number of arguments among the
// variants with the same number of NSA words, and custom code then derives
// the exact variant based on the sample variant and the image dimension.
list<MIMGAddrSize> NSAInstrs =
!foldl([]<MIMGAddrSize>, [[12, 11, 10], [9, 8, 7, 6], [5, 4, 3, 2]], prev, nsa_group,
!listconcat(prev,
!foldl([]<MIMGAddrSize>, nsa_group, lhs, dw,
!if(isIntInList<dw, AllNumAddrWords>.ret,
!listconcat(lhs, [MIMGAddrSize<dw, !empty(lhs)>]),
lhs))));
}
multiclass MIMG_Sampler_Src_Helper <bits<8> op, string asm,
AMDGPUSampleVariant sample, RegisterClass dst_rc,
bit enableDisasm = 0> {
foreach addr = MIMG_Sampler_AddrSizes<sample>.MachineInstrs in {
let VAddrDwords = addr.NumWords in {
def _V # addr.NumWords
: MIMG_Sampler_Helper <op, asm, dst_rc, addr.RegClass,
!if(!and(enableDisasm, addr.Disassemble), "AMDGPU", "")>;
def _V # addr.NumWords # _gfx10
: MIMG_Sampler_gfx10 <op, asm, dst_rc, addr.RegClass,
!if(!and(enableDisasm, addr.Disassemble), "AMDGPU", "")>;
}
}
foreach addr = MIMG_Sampler_AddrSizes<sample>.NSAInstrs in {
let VAddrDwords = addr.NumWords in {
def _V # addr.NumWords # _nsa_gfx10
: MIMG_Sampler_nsa_gfx10<op, asm, dst_rc, addr.NumWords,
!if(!and(enableDisasm, addr.Disassemble), "AMDGPU", "")>;
}
}
}
class MIMG_Sampler_BaseOpcode<AMDGPUSampleVariant sample>
: MIMGBaseOpcode {
let Sampler = 1;
let NumExtraArgs = !size(sample.ExtraAddrArgs);
let Gradients = sample.Gradients;
let LodOrClampOrMip = !ne(sample.LodOrClamp, "");
}
multiclass MIMG_Sampler <bits<8> op, AMDGPUSampleVariant sample, bit wqm = 0,
bit isG16 = 0, bit isGetLod = 0,
string asm = "image_sample"#sample.LowerCaseMod#!if(isG16, "_g16", "")> {
def "" : MIMG_Sampler_BaseOpcode<sample> {
let HasD16 = !if(isGetLod, 0, 1);
let G16 = isG16;
}
let BaseOpcode = !cast<MIMGBaseOpcode>(NAME), WQM = wqm,
mayLoad = !if(isGetLod, 0, 1) in {
let VDataDwords = 1 in
defm _V1 : MIMG_Sampler_Src_Helper<op, asm, sample, VGPR_32, 1>;
let VDataDwords = 2 in
defm _V2 : MIMG_Sampler_Src_Helper<op, asm, sample, VReg_64>;
let VDataDwords = 3 in
defm _V3 : MIMG_Sampler_Src_Helper<op, asm, sample, VReg_96>;
let VDataDwords = 4 in
defm _V4 : MIMG_Sampler_Src_Helper<op, asm, sample, VReg_128>;
let VDataDwords = 5 in
defm _V5 : MIMG_Sampler_Src_Helper<op, asm, sample, VReg_160>;
}
}
multiclass MIMG_Sampler_WQM <bits<8> op, AMDGPUSampleVariant sample>
: MIMG_Sampler<op, sample, 1>;
multiclass MIMG_Gather <bits<8> op, AMDGPUSampleVariant sample, bit wqm = 0,
string asm = "image_gather4"#sample.LowerCaseMod> {
def "" : MIMG_Sampler_BaseOpcode<sample> {
let HasD16 = 1;
let Gather4 = 1;
}
let BaseOpcode = !cast<MIMGBaseOpcode>(NAME), WQM = wqm,
Gather4 = 1, hasPostISelHook = 0 in {
let VDataDwords = 2 in
defm _V2 : MIMG_Sampler_Src_Helper<op, asm, sample, VReg_64>; /* for packed D16 only */
let VDataDwords = 4 in
defm _V4 : MIMG_Sampler_Src_Helper<op, asm, sample, VReg_128, 1>;
let VDataDwords = 5 in
defm _V5 : MIMG_Sampler_Src_Helper<op, asm, sample, VReg_160>;
}
}
multiclass MIMG_Gather_WQM <bits<8> op, AMDGPUSampleVariant sample>
: MIMG_Gather<op, sample, 1>;
class MIMG_IntersectRay_gfx10<int op, string opcode, RegisterClass AddrRC, bit A16>
: MIMG_gfx10<op, (outs VReg_128:$vdata), "AMDGPU"> {
let InOperandList = !con((ins AddrRC:$vaddr0, SReg_128:$srsrc),
!if(!eq(A16,1), (ins GFX10A16:$a16), (ins)));
let AsmString = opcode#" $vdata, $vaddr0, $srsrc"#!if(!eq(A16,1), "$a16", "");
let nsa = 0;
}
class MIMG_IntersectRay_nsa_gfx10<int op, string opcode, int num_addrs, bit A16>
: MIMG_nsa_gfx10<op, (outs VReg_128:$vdata), num_addrs, "AMDGPU"> {
let InOperandList = !con(nsah.AddrIns,
(ins SReg_128:$srsrc),
!if(!eq(A16,1), (ins GFX10A16:$a16), (ins)));
let AsmString = opcode#" $vdata, "#nsah.AddrAsm#", $srsrc"#!if(!eq(A16,1), "$a16", "");
}
multiclass MIMG_IntersectRay<int op, string opcode, int num_addrs, bit A16> {
def "" : MIMGBaseOpcode;
let SubtargetPredicate = HasGFX10_BEncoding,
AssemblerPredicate = HasGFX10_BEncoding,
AsmMatchConverter = !if(!eq(A16,1), "cvtIntersectRay", ""),
dmask = 0xf,
unorm = 1,
d16 = 0,
glc = 0,
slc = 0,
dlc = 0,
tfe = 0,
lwe = 0,
r128 = 1,
ssamp = 0,
dim = {0, 0, 0},
a16 = A16,
d16 = 0,
BaseOpcode = !cast<MIMGBaseOpcode>(NAME),
VDataDwords = 4 in {
// TODO: MIMGAddrSize will choose VReg_512 which is a 16 register tuple,
// when we only need 9, 11 or 12 depending on A16 field and ptr size.
def "_sa" : MIMG_IntersectRay_gfx10<op, opcode, MIMGAddrSize<num_addrs, 0>.RegClass, A16> {
let VAddrDwords = !srl(MIMGAddrSize<num_addrs, 0>.RegClass.Size, 5);
}
def _nsa : MIMG_IntersectRay_nsa_gfx10<op, opcode, num_addrs, A16> {
let VAddrDwords = num_addrs;
}
}
}
//===----------------------------------------------------------------------===//
// MIMG Instructions
//===----------------------------------------------------------------------===//
defm IMAGE_LOAD : MIMG_NoSampler <0x00000000, "image_load", 1>;
defm IMAGE_LOAD_MIP : MIMG_NoSampler <0x00000001, "image_load_mip", 1, 1>;
defm IMAGE_LOAD_PCK : MIMG_NoSampler <0x00000002, "image_load_pck", 0>;
defm IMAGE_LOAD_PCK_SGN : MIMG_NoSampler <0x00000003, "image_load_pck_sgn", 0>;
defm IMAGE_LOAD_MIP_PCK : MIMG_NoSampler <0x00000004, "image_load_mip_pck", 0, 1>;
defm IMAGE_LOAD_MIP_PCK_SGN : MIMG_NoSampler <0x00000005, "image_load_mip_pck_sgn", 0, 1>;
defm IMAGE_STORE : MIMG_Store <0x00000008, "image_store", 1>;
defm IMAGE_STORE_MIP : MIMG_Store <0x00000009, "image_store_mip", 1, 1>;
defm IMAGE_STORE_PCK : MIMG_Store <0x0000000a, "image_store_pck", 0>;
defm IMAGE_STORE_MIP_PCK : MIMG_Store <0x0000000b, "image_store_mip_pck", 0, 1>;
defm IMAGE_GET_RESINFO : MIMG_NoSampler <0x0000000e, "image_get_resinfo", 0, 1, 1>;
defm IMAGE_ATOMIC_SWAP : MIMG_Atomic <mimg<0x0f, 0x10>, "image_atomic_swap">;
defm IMAGE_ATOMIC_CMPSWAP : MIMG_Atomic <mimg<0x10, 0x11>, "image_atomic_cmpswap", 1>;
defm IMAGE_ATOMIC_ADD : MIMG_Atomic <mimg<0x11, 0x12>, "image_atomic_add">;
defm IMAGE_ATOMIC_SUB : MIMG_Atomic <mimg<0x12, 0x13>, "image_atomic_sub">;
//def IMAGE_ATOMIC_RSUB : MIMG_NoPattern_ <"image_atomic_rsub", 0x00000013>; -- not on VI
defm IMAGE_ATOMIC_SMIN : MIMG_Atomic <mimg<0x14>, "image_atomic_smin">;
defm IMAGE_ATOMIC_UMIN : MIMG_Atomic <mimg<0x15>, "image_atomic_umin">;
defm IMAGE_ATOMIC_SMAX : MIMG_Atomic <mimg<0x16>, "image_atomic_smax">;
defm IMAGE_ATOMIC_UMAX : MIMG_Atomic <mimg<0x17>, "image_atomic_umax">;
defm IMAGE_ATOMIC_AND : MIMG_Atomic <mimg<0x18>, "image_atomic_and">;
defm IMAGE_ATOMIC_OR : MIMG_Atomic <mimg<0x19>, "image_atomic_or">;
defm IMAGE_ATOMIC_XOR : MIMG_Atomic <mimg<0x1a>, "image_atomic_xor">;
defm IMAGE_ATOMIC_INC : MIMG_Atomic <mimg<0x1b>, "image_atomic_inc">;
defm IMAGE_ATOMIC_DEC : MIMG_Atomic <mimg<0x1c>, "image_atomic_dec">;
//let FPAtomic = 1 in {
//def IMAGE_ATOMIC_FCMPSWAP : MIMG_NoPattern_ <"image_atomic_fcmpswap", 0x0000001d, 1>; -- not on VI
//def IMAGE_ATOMIC_FMIN : MIMG_NoPattern_ <"image_atomic_fmin", 0x0000001e>; -- not on VI
//def IMAGE_ATOMIC_FMAX : MIMG_NoPattern_ <"image_atomic_fmax", 0x0000001f>; -- not on VI
//} // End let FPAtomic = 1
defm IMAGE_SAMPLE : MIMG_Sampler_WQM <0x00000020, AMDGPUSample>;
defm IMAGE_SAMPLE_CL : MIMG_Sampler_WQM <0x00000021, AMDGPUSample_cl>;
defm IMAGE_SAMPLE_D : MIMG_Sampler <0x00000022, AMDGPUSample_d>;
defm IMAGE_SAMPLE_D_CL : MIMG_Sampler <0x00000023, AMDGPUSample_d_cl>;
defm IMAGE_SAMPLE_D_G16 : MIMG_Sampler <0x000000a2, AMDGPUSample_d, 0, 1>;
defm IMAGE_SAMPLE_D_CL_G16 : MIMG_Sampler <0x000000a3, AMDGPUSample_d_cl, 0, 1>;
defm IMAGE_SAMPLE_L : MIMG_Sampler <0x00000024, AMDGPUSample_l>;
defm IMAGE_SAMPLE_B : MIMG_Sampler_WQM <0x00000025, AMDGPUSample_b>;
defm IMAGE_SAMPLE_B_CL : MIMG_Sampler_WQM <0x00000026, AMDGPUSample_b_cl>;
defm IMAGE_SAMPLE_LZ : MIMG_Sampler <0x00000027, AMDGPUSample_lz>;
defm IMAGE_SAMPLE_C : MIMG_Sampler_WQM <0x00000028, AMDGPUSample_c>;
defm IMAGE_SAMPLE_C_CL : MIMG_Sampler_WQM <0x00000029, AMDGPUSample_c_cl>;
defm IMAGE_SAMPLE_C_D : MIMG_Sampler <0x0000002a, AMDGPUSample_c_d>;
defm IMAGE_SAMPLE_C_D_CL : MIMG_Sampler <0x0000002b, AMDGPUSample_c_d_cl>;
defm IMAGE_SAMPLE_C_D_G16 : MIMG_Sampler <0x000000aa, AMDGPUSample_c_d, 0, 1>;
defm IMAGE_SAMPLE_C_D_CL_G16 : MIMG_Sampler <0x000000ab, AMDGPUSample_c_d_cl, 0, 1>;
defm IMAGE_SAMPLE_C_L : MIMG_Sampler <0x0000002c, AMDGPUSample_c_l>;
defm IMAGE_SAMPLE_C_B : MIMG_Sampler_WQM <0x0000002d, AMDGPUSample_c_b>;
defm IMAGE_SAMPLE_C_B_CL : MIMG_Sampler_WQM <0x0000002e, AMDGPUSample_c_b_cl>;
defm IMAGE_SAMPLE_C_LZ : MIMG_Sampler <0x0000002f, AMDGPUSample_c_lz>;
defm IMAGE_SAMPLE_O : MIMG_Sampler_WQM <0x00000030, AMDGPUSample_o>;
defm IMAGE_SAMPLE_CL_O : MIMG_Sampler_WQM <0x00000031, AMDGPUSample_cl_o>;
defm IMAGE_SAMPLE_D_O : MIMG_Sampler <0x00000032, AMDGPUSample_d_o>;
defm IMAGE_SAMPLE_D_CL_O : MIMG_Sampler <0x00000033, AMDGPUSample_d_cl_o>;
defm IMAGE_SAMPLE_D_O_G16 : MIMG_Sampler <0x000000b2, AMDGPUSample_d_o, 0, 1>;
defm IMAGE_SAMPLE_D_CL_O_G16 : MIMG_Sampler <0x000000b3, AMDGPUSample_d_cl_o, 0, 1>;
defm IMAGE_SAMPLE_L_O : MIMG_Sampler <0x00000034, AMDGPUSample_l_o>;
defm IMAGE_SAMPLE_B_O : MIMG_Sampler_WQM <0x00000035, AMDGPUSample_b_o>;
defm IMAGE_SAMPLE_B_CL_O : MIMG_Sampler_WQM <0x00000036, AMDGPUSample_b_cl_o>;
defm IMAGE_SAMPLE_LZ_O : MIMG_Sampler <0x00000037, AMDGPUSample_lz_o>;
defm IMAGE_SAMPLE_C_O : MIMG_Sampler_WQM <0x00000038, AMDGPUSample_c_o>;
defm IMAGE_SAMPLE_C_CL_O : MIMG_Sampler_WQM <0x00000039, AMDGPUSample_c_cl_o>;
defm IMAGE_SAMPLE_C_D_O : MIMG_Sampler <0x0000003a, AMDGPUSample_c_d_o>;
defm IMAGE_SAMPLE_C_D_CL_O : MIMG_Sampler <0x0000003b, AMDGPUSample_c_d_cl_o>;
defm IMAGE_SAMPLE_C_D_O_G16 : MIMG_Sampler <0x000000ba, AMDGPUSample_c_d_o, 0, 1>;
defm IMAGE_SAMPLE_C_D_CL_O_G16 : MIMG_Sampler <0x000000bb, AMDGPUSample_c_d_cl_o, 0, 1>;
defm IMAGE_SAMPLE_C_L_O : MIMG_Sampler <0x0000003c, AMDGPUSample_c_l_o>;
defm IMAGE_SAMPLE_C_B_CL_O : MIMG_Sampler_WQM <0x0000003e, AMDGPUSample_c_b_cl_o>;
defm IMAGE_SAMPLE_C_B_O : MIMG_Sampler_WQM <0x0000003d, AMDGPUSample_c_b_o>;
defm IMAGE_SAMPLE_C_LZ_O : MIMG_Sampler <0x0000003f, AMDGPUSample_c_lz_o>;
defm IMAGE_GATHER4 : MIMG_Gather_WQM <0x00000040, AMDGPUSample>;
defm IMAGE_GATHER4_CL : MIMG_Gather_WQM <0x00000041, AMDGPUSample_cl>;
defm IMAGE_GATHER4_L : MIMG_Gather <0x00000044, AMDGPUSample_l>;
defm IMAGE_GATHER4_B : MIMG_Gather_WQM <0x00000045, AMDGPUSample_b>;
defm IMAGE_GATHER4_B_CL : MIMG_Gather_WQM <0x00000046, AMDGPUSample_b_cl>;
defm IMAGE_GATHER4_LZ : MIMG_Gather <0x00000047, AMDGPUSample_lz>;
defm IMAGE_GATHER4_C : MIMG_Gather_WQM <0x00000048, AMDGPUSample_c>;
defm IMAGE_GATHER4_C_CL : MIMG_Gather_WQM <0x00000049, AMDGPUSample_c_cl>;
defm IMAGE_GATHER4_C_L : MIMG_Gather <0x0000004c, AMDGPUSample_c_l>;
defm IMAGE_GATHER4_C_B : MIMG_Gather_WQM <0x0000004d, AMDGPUSample_c_b>;
defm IMAGE_GATHER4_C_B_CL : MIMG_Gather_WQM <0x0000004e, AMDGPUSample_c_b_cl>;
defm IMAGE_GATHER4_C_LZ : MIMG_Gather <0x0000004f, AMDGPUSample_c_lz>;
defm IMAGE_GATHER4_O : MIMG_Gather_WQM <0x00000050, AMDGPUSample_o>;
defm IMAGE_GATHER4_CL_O : MIMG_Gather_WQM <0x00000051, AMDGPUSample_cl_o>;
defm IMAGE_GATHER4_L_O : MIMG_Gather <0x00000054, AMDGPUSample_l_o>;
defm IMAGE_GATHER4_B_O : MIMG_Gather_WQM <0x00000055, AMDGPUSample_b_o>;
defm IMAGE_GATHER4_B_CL_O : MIMG_Gather <0x00000056, AMDGPUSample_b_cl_o>;
defm IMAGE_GATHER4_LZ_O : MIMG_Gather <0x00000057, AMDGPUSample_lz_o>;
defm IMAGE_GATHER4_C_O : MIMG_Gather_WQM <0x00000058, AMDGPUSample_c_o>;
defm IMAGE_GATHER4_C_CL_O : MIMG_Gather_WQM <0x00000059, AMDGPUSample_c_cl_o>;
defm IMAGE_GATHER4_C_L_O : MIMG_Gather <0x0000005c, AMDGPUSample_c_l_o>;
defm IMAGE_GATHER4_C_B_O : MIMG_Gather_WQM <0x0000005d, AMDGPUSample_c_b_o>;
defm IMAGE_GATHER4_C_B_CL_O : MIMG_Gather_WQM <0x0000005e, AMDGPUSample_c_b_cl_o>;
defm IMAGE_GATHER4_C_LZ_O : MIMG_Gather <0x0000005f, AMDGPUSample_c_lz_o>;
defm IMAGE_GET_LOD : MIMG_Sampler <0x00000060, AMDGPUSample, 1, 0, 1, "image_get_lod">;
defm IMAGE_SAMPLE_CD : MIMG_Sampler <0x00000068, AMDGPUSample_cd>;
defm IMAGE_SAMPLE_CD_CL : MIMG_Sampler <0x00000069, AMDGPUSample_cd_cl>;
defm IMAGE_SAMPLE_C_CD : MIMG_Sampler <0x0000006a, AMDGPUSample_c_cd>;
defm IMAGE_SAMPLE_C_CD_CL : MIMG_Sampler <0x0000006b, AMDGPUSample_c_cd_cl>;
defm IMAGE_SAMPLE_CD_O : MIMG_Sampler <0x0000006c, AMDGPUSample_cd_o>;
defm IMAGE_SAMPLE_CD_CL_O : MIMG_Sampler <0x0000006d, AMDGPUSample_cd_cl_o>;
defm IMAGE_SAMPLE_C_CD_O : MIMG_Sampler <0x0000006e, AMDGPUSample_c_cd_o>;
defm IMAGE_SAMPLE_C_CD_CL_O : MIMG_Sampler <0x0000006f, AMDGPUSample_c_cd_cl_o>;
defm IMAGE_SAMPLE_CD_G16 : MIMG_Sampler <0x000000e8, AMDGPUSample_cd, 0, 1>;
defm IMAGE_SAMPLE_CD_CL_G16 : MIMG_Sampler <0x000000e9, AMDGPUSample_cd_cl, 0, 1>;
defm IMAGE_SAMPLE_C_CD_G16 : MIMG_Sampler <0x000000ea, AMDGPUSample_c_cd, 0, 1>;
defm IMAGE_SAMPLE_C_CD_CL_G16 : MIMG_Sampler <0x000000eb, AMDGPUSample_c_cd_cl, 0, 1>;
defm IMAGE_SAMPLE_CD_O_G16 : MIMG_Sampler <0x000000ec, AMDGPUSample_cd_o, 0, 1>;
defm IMAGE_SAMPLE_CD_CL_O_G16 : MIMG_Sampler <0x000000ed, AMDGPUSample_cd_cl_o, 0, 1>;
defm IMAGE_SAMPLE_C_CD_O_G16 : MIMG_Sampler <0x000000ee, AMDGPUSample_c_cd_o, 0, 1>;
defm IMAGE_SAMPLE_C_CD_CL_O_G16 : MIMG_Sampler <0x000000ef, AMDGPUSample_c_cd_cl_o, 0, 1>;
//def IMAGE_RSRC256 : MIMG_NoPattern_RSRC256 <"image_rsrc256", 0x0000007e>;
//def IMAGE_SAMPLER : MIMG_NoPattern_ <"image_sampler", 0x0000007f>;
let SubtargetPredicate = HasGFX10_BEncoding in
defm IMAGE_MSAA_LOAD : MIMG_NoSampler <0x00000080, "image_msaa_load", 1>;
defm IMAGE_BVH_INTERSECT_RAY : MIMG_IntersectRay<0xe6, "image_bvh_intersect_ray", 11, 0>;
defm IMAGE_BVH_INTERSECT_RAY_a16 : MIMG_IntersectRay<0xe6, "image_bvh_intersect_ray", 8, 1>;
defm IMAGE_BVH64_INTERSECT_RAY : MIMG_IntersectRay<0xe7, "image_bvh64_intersect_ray", 12, 0>;
defm IMAGE_BVH64_INTERSECT_RAY_a16 : MIMG_IntersectRay<0xe7, "image_bvh64_intersect_ray", 9, 1>;
/********** ========================================= **********/
/********** Table of dimension-aware image intrinsics **********/
/********** ========================================= **********/
class ImageDimIntrinsicInfo<AMDGPUImageDimIntrinsic I> {
Intrinsic Intr = I;
MIMGBaseOpcode BaseOpcode = !cast<MIMGBaseOpcode>(!strconcat("IMAGE_", I.P.OpMod));
AMDGPUDimProps Dim = I.P.Dim;
AMDGPUImageDimIntrinsicEval DimEval = AMDGPUImageDimIntrinsicEval<I.P>;
bits<8> GradientStart = DimEval.GradientArgIndex;
bits<8> CoordStart = DimEval.CoordArgIndex;
bits<8> VAddrEnd = !add(DimEval.VAddrArgIndex, DimEval.NumVAddrArgs);
bits<8> GradientTyArg = !add(I.P.NumRetAndDataAnyTypes,
!foldl(0, I.P.ExtraAddrArgs, cnt, arg, !add(cnt, arg.Type.isAny)));
bits<8> CoordTyArg = !add(GradientTyArg, !if(I.P.Gradients, 1, 0));
}
def ImageDimIntrinsicTable : GenericTable {
let FilterClass = "ImageDimIntrinsicInfo";
let Fields = ["Intr", "BaseOpcode", "Dim", "GradientStart", "CoordStart", "VAddrEnd", "GradientTyArg", "CoordTyArg"];
GenericEnum TypeOf_BaseOpcode = MIMGBaseOpcode;
GenericEnum TypeOf_Dim = MIMGDim;
let PrimaryKey = ["Intr"];
let PrimaryKeyName = "getImageDimIntrinsicInfo";
let PrimaryKeyEarlyOut = 1;
}
def getImageDimInstrinsicByBaseOpcode : SearchIndex {
let Table = ImageDimIntrinsicTable;
let Key = ["BaseOpcode", "Dim"];
}
foreach intr = !listconcat(AMDGPUImageDimIntrinsics,
AMDGPUImageDimAtomicIntrinsics) in {
def : ImageDimIntrinsicInfo<intr>;
}
// L to LZ Optimization Mapping
def : MIMGLZMapping<IMAGE_SAMPLE_L, IMAGE_SAMPLE_LZ>;
def : MIMGLZMapping<IMAGE_SAMPLE_C_L, IMAGE_SAMPLE_C_LZ>;
def : MIMGLZMapping<IMAGE_SAMPLE_L_O, IMAGE_SAMPLE_LZ_O>;
def : MIMGLZMapping<IMAGE_SAMPLE_C_L_O, IMAGE_SAMPLE_C_LZ_O>;
def : MIMGLZMapping<IMAGE_GATHER4_L, IMAGE_GATHER4_LZ>;
def : MIMGLZMapping<IMAGE_GATHER4_C_L, IMAGE_GATHER4_C_LZ>;
def : MIMGLZMapping<IMAGE_GATHER4_L_O, IMAGE_GATHER4_LZ_O>;
def : MIMGLZMapping<IMAGE_GATHER4_C_L_O, IMAGE_GATHER4_C_LZ_O>;
// MIP to NONMIP Optimization Mapping
def : MIMGMIPMapping<IMAGE_LOAD_MIP, IMAGE_LOAD>;
def : MIMGMIPMapping<IMAGE_STORE_MIP, IMAGE_STORE>;
// G to G16 Optimization Mapping
def : MIMGG16Mapping<IMAGE_SAMPLE_D, IMAGE_SAMPLE_D_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_D_CL, IMAGE_SAMPLE_D_CL_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_C_D, IMAGE_SAMPLE_C_D_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_C_D_CL, IMAGE_SAMPLE_C_D_CL_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_D_O, IMAGE_SAMPLE_D_O_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_D_CL_O, IMAGE_SAMPLE_D_CL_O_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_C_D_O, IMAGE_SAMPLE_C_D_O_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_C_D_CL_O, IMAGE_SAMPLE_C_D_CL_O_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_CD, IMAGE_SAMPLE_CD_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_CD_CL, IMAGE_SAMPLE_CD_CL_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_C_CD, IMAGE_SAMPLE_C_CD_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_C_CD_CL, IMAGE_SAMPLE_C_CD_CL_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_CD_O, IMAGE_SAMPLE_CD_O_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_CD_CL_O, IMAGE_SAMPLE_CD_CL_O_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_C_CD_O, IMAGE_SAMPLE_C_CD_O_G16>;
def : MIMGG16Mapping<IMAGE_SAMPLE_C_CD_CL_O, IMAGE_SAMPLE_C_CD_CL_O_G16>;