llvm.org GIT mirror llvm / 28fe7a0 lib / Target / AMDGPU / R600Instructions.td
28fe7a0

Tree @28fe7a0 (Download .tar.gz)

R600Instructions.td @28fe7a0raw · history · blame

   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
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
//===-- R600Instructions.td - R600 Instruction defs  -------*- tablegen -*-===//
//
//                     The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// TableGen definitions for instructions which are available on R600 family
// GPUs.
//
//===----------------------------------------------------------------------===//

include "R600Intrinsics.td"
include "R600InstrFormats.td"

class InstR600ISA <dag outs, dag ins, string asm, list<dag> pattern> :
    InstR600 <outs, ins, asm, pattern, NullALU> {

  let Namespace = "AMDGPU";
}

def MEMxi : Operand<iPTR> {
  let MIOperandInfo = (ops R600_TReg32_X:$ptr, i32imm:$index);
  let PrintMethod = "printMemOperand";
}

def MEMrr : Operand<iPTR> {
  let MIOperandInfo = (ops R600_Reg32:$ptr, R600_Reg32:$index);
}

// Operands for non-registers

class InstFlag<string PM = "printOperand", int Default = 0>
    : OperandWithDefaultOps <i32, (ops (i32 Default))> {
  let PrintMethod = PM;
}

// src_sel for ALU src operands, see also ALU_CONST, ALU_PARAM registers
def SEL : OperandWithDefaultOps <i32, (ops (i32 -1))> {
  let PrintMethod = "printSel";
}
def BANK_SWIZZLE : OperandWithDefaultOps <i32, (ops (i32 0))> {
  let PrintMethod = "printBankSwizzle";
}

def LITERAL : InstFlag<"printLiteral">;

def WRITE : InstFlag <"printWrite", 1>;
def OMOD : InstFlag <"printOMOD">;
def REL : InstFlag <"printRel">;
def CLAMP : InstFlag <"printClamp">;
def NEG : InstFlag <"printNeg">;
def ABS : InstFlag <"printAbs">;
def UEM : InstFlag <"printUpdateExecMask">;
def UP : InstFlag <"printUpdatePred">;

// XXX: The r600g finalizer in Mesa expects last to be one in most cases.
// Once we start using the packetizer in this backend we should have this
// default to 0.
def LAST : InstFlag<"printLast", 1>;
def RSel : Operand<i32> {
  let PrintMethod = "printRSel";
}
def CT: Operand<i32> {
  let PrintMethod = "printCT";
}

def FRAMEri : Operand<iPTR> {
  let MIOperandInfo = (ops R600_Reg32:$ptr, i32imm:$index);
}

def ADDRParam : ComplexPattern<i32, 2, "SelectADDRParam", [], []>;
def ADDRDWord : ComplexPattern<i32, 1, "SelectADDRDWord", [], []>;
def ADDRVTX_READ : ComplexPattern<i32, 2, "SelectADDRVTX_READ", [], []>;
def ADDRGA_CONST_OFFSET : ComplexPattern<i32, 1, "SelectGlobalValueConstantOffset", [], []>;
def ADDRGA_VAR_OFFSET : ComplexPattern<i32, 2, "SelectGlobalValueVariableOffset", [], []>;


def R600_Pred : PredicateOperand<i32, (ops R600_Predicate),
                                     (ops PRED_SEL_OFF)>;


let mayLoad = 0, mayStore = 0, hasSideEffects = 0 in {

// Class for instructions with only one source register.
// If you add new ins to this instruction, make sure they are listed before
// $literal, because the backend currently assumes that the last operand is
// a literal.  Also be sure to update the enum R600Op1OperandIndex::ROI in
// R600Defines.h, R600InstrInfo::buildDefaultInstruction(),
// and R600InstrInfo::getOperandIdx().
class R600_1OP <bits<11> inst, string opName, list<dag> pattern,
                InstrItinClass itin = AnyALU> :
    InstR600 <(outs R600_Reg32:$dst),
              (ins WRITE:$write, OMOD:$omod, REL:$dst_rel, CLAMP:$clamp,
                   R600_Reg32:$src0, NEG:$src0_neg, REL:$src0_rel, ABS:$src0_abs, SEL:$src0_sel,
                   LAST:$last, R600_Pred:$pred_sel, LITERAL:$literal,
                   BANK_SWIZZLE:$bank_swizzle),
              !strconcat("  ", opName,
                   "$clamp $last $dst$write$dst_rel$omod, "
                   "$src0_neg$src0_abs$src0$src0_abs$src0_rel, "
                   "$pred_sel $bank_swizzle"),
              pattern,
              itin>,
    R600ALU_Word0,
    R600ALU_Word1_OP2 <inst> {

  let src1 = 0;
  let src1_rel = 0;
  let src1_neg = 0;
  let src1_abs = 0;
  let update_exec_mask = 0;
  let update_pred = 0;
  let HasNativeOperands = 1;
  let Op1 = 1;
  let ALUInst = 1;
  let DisableEncoding = "$literal";
  let UseNamedOperandTable = 1;

  let Inst{31-0}  = Word0;
  let Inst{63-32} = Word1;
}

class R600_1OP_Helper <bits<11> inst, string opName, SDPatternOperator node,
                    InstrItinClass itin = AnyALU> :
    R600_1OP <inst, opName,
              [(set R600_Reg32:$dst, (node R600_Reg32:$src0))], itin
>;

// If you add or change the operands for R600_2OP instructions, you must
// also update the R600Op2OperandIndex::ROI enum in R600Defines.h,
// R600InstrInfo::buildDefaultInstruction(), and R600InstrInfo::getOperandIdx().
class R600_2OP <bits<11> inst, string opName, list<dag> pattern,
                InstrItinClass itin = AnyALU> :
  InstR600 <(outs R600_Reg32:$dst),
          (ins UEM:$update_exec_mask, UP:$update_pred, WRITE:$write,
               OMOD:$omod, REL:$dst_rel, CLAMP:$clamp,
               R600_Reg32:$src0, NEG:$src0_neg, REL:$src0_rel, ABS:$src0_abs, SEL:$src0_sel,
               R600_Reg32:$src1, NEG:$src1_neg, REL:$src1_rel, ABS:$src1_abs, SEL:$src1_sel,
               LAST:$last, R600_Pred:$pred_sel, LITERAL:$literal,
               BANK_SWIZZLE:$bank_swizzle),
          !strconcat("  ", opName,
                "$clamp $last $update_exec_mask$update_pred$dst$write$dst_rel$omod, "
                "$src0_neg$src0_abs$src0$src0_abs$src0_rel, "
                "$src1_neg$src1_abs$src1$src1_abs$src1_rel, "
                "$pred_sel $bank_swizzle"),
          pattern,
          itin>,
    R600ALU_Word0,
    R600ALU_Word1_OP2 <inst> {

  let HasNativeOperands = 1;
  let Op2 = 1;
  let ALUInst = 1;
  let DisableEncoding = "$literal";
  let UseNamedOperandTable = 1;

  let Inst{31-0}  = Word0;
  let Inst{63-32} = Word1;
}

class R600_2OP_Helper <bits<11> inst, string opName, SDPatternOperator node,
                       InstrItinClass itin = AnyALU> :
    R600_2OP <inst, opName,
              [(set R600_Reg32:$dst, (node R600_Reg32:$src0,
                                           R600_Reg32:$src1))], itin
>;

// If you add our change the operands for R600_3OP instructions, you must
// also update the R600Op3OperandIndex::ROI enum in R600Defines.h,
// R600InstrInfo::buildDefaultInstruction(), and
// R600InstrInfo::getOperandIdx().
class R600_3OP <bits<5> inst, string opName, list<dag> pattern,
                InstrItinClass itin = AnyALU> :
  InstR600 <(outs R600_Reg32:$dst),
          (ins REL:$dst_rel, CLAMP:$clamp,
               R600_Reg32:$src0, NEG:$src0_neg, REL:$src0_rel, SEL:$src0_sel,
               R600_Reg32:$src1, NEG:$src1_neg, REL:$src1_rel, SEL:$src1_sel,
               R600_Reg32:$src2, NEG:$src2_neg, REL:$src2_rel, SEL:$src2_sel,
               LAST:$last, R600_Pred:$pred_sel, LITERAL:$literal,
               BANK_SWIZZLE:$bank_swizzle),
          !strconcat("  ", opName, "$clamp $last $dst$dst_rel, "
                             "$src0_neg$src0$src0_rel, "
                             "$src1_neg$src1$src1_rel, "
                             "$src2_neg$src2$src2_rel, "
                             "$pred_sel"
                             "$bank_swizzle"),
          pattern,
          itin>,
    R600ALU_Word0,
    R600ALU_Word1_OP3<inst>{

  let HasNativeOperands = 1;
  let DisableEncoding = "$literal";
  let Op3 = 1;
  let UseNamedOperandTable = 1;
  let ALUInst = 1;

  let Inst{31-0}  = Word0;
  let Inst{63-32} = Word1;
}

class R600_REDUCTION <bits<11> inst, dag ins, string asm, list<dag> pattern,
                      InstrItinClass itin = VecALU> :
  InstR600 <(outs R600_Reg32:$dst),
          ins,
          asm,
          pattern,
          itin>;



} // End mayLoad = 1, mayStore = 0, hasSideEffects = 0

def TEX_SHADOW : PatLeaf<
  (imm),
  [{uint32_t TType = (uint32_t)N->getZExtValue();
    return (TType >= 6 && TType <= 8) || TType == 13;
  }]
>;

def TEX_RECT : PatLeaf<
  (imm),
  [{uint32_t TType = (uint32_t)N->getZExtValue();
    return TType == 5;
  }]
>;

def TEX_ARRAY : PatLeaf<
  (imm),
  [{uint32_t TType = (uint32_t)N->getZExtValue();
    return TType == 9 || TType == 10 || TType == 16;
  }]
>;

def TEX_SHADOW_ARRAY : PatLeaf<
  (imm),
  [{uint32_t TType = (uint32_t)N->getZExtValue();
    return TType == 11 || TType == 12 || TType == 17;
  }]
>;

def TEX_MSAA : PatLeaf<
  (imm),
  [{uint32_t TType = (uint32_t)N->getZExtValue();
    return TType == 14;
  }]
>;

def TEX_ARRAY_MSAA : PatLeaf<
  (imm),
  [{uint32_t TType = (uint32_t)N->getZExtValue();
    return TType == 15;
  }]
>;

class EG_CF_RAT <bits <8> cfinst, bits <6> ratinst, bits<4> ratid, bits<4> mask,
                 dag outs, dag ins, string asm, list<dag> pattern> :
    InstR600ISA <outs, ins, asm, pattern>,
    CF_ALLOC_EXPORT_WORD0_RAT, CF_ALLOC_EXPORT_WORD1_BUF  {

  let rat_id = ratid;
  let rat_inst = ratinst;
  let rim         = 0;
  // XXX: Have a separate instruction for non-indexed writes.
  let type        = 1;
  let rw_rel      = 0;
  let elem_size   = 0;

  let array_size  = 0;
  let comp_mask   = mask;
  let burst_count = 0;
  let vpm         = 0;
  let cf_inst = cfinst;
  let mark        = 0;
  let barrier     = 1;

  let Inst{31-0} = Word0;
  let Inst{63-32} = Word1;
  let IsExport = 1;

}

class VTX_READ <string name, bits<8> buffer_id, dag outs, list<dag> pattern>
    : InstR600ISA <outs, (ins MEMxi:$src_gpr), name, pattern>,
      VTX_WORD1_GPR {

  // Static fields
  let DST_REL = 0;
  // The docs say that if this bit is set, then DATA_FORMAT, NUM_FORMAT_ALL,
  // FORMAT_COMP_ALL, SRF_MODE_ALL, and ENDIAN_SWAP fields will be ignored,
  // however, based on my testing if USE_CONST_FIELDS is set, then all
  // these fields need to be set to 0.
  let USE_CONST_FIELDS = 0;
  let NUM_FORMAT_ALL = 1;
  let FORMAT_COMP_ALL = 0;
  let SRF_MODE_ALL = 0;

  let Inst{63-32} = Word1;
  // LLVM can only encode 64-bit instructions, so these fields are manually
  // encoded in R600CodeEmitter
  //
  // bits<16> OFFSET;
  // bits<2>  ENDIAN_SWAP = 0;
  // bits<1>  CONST_BUF_NO_STRIDE = 0;
  // bits<1>  MEGA_FETCH = 0;
  // bits<1>  ALT_CONST = 0;
  // bits<2>  BUFFER_INDEX_MODE = 0;

  // VTX_WORD2 (LLVM can only encode 64-bit instructions, so WORD2 encoding
  // is done in R600CodeEmitter
  //
  // Inst{79-64} = OFFSET;
  // Inst{81-80} = ENDIAN_SWAP;
  // Inst{82}    = CONST_BUF_NO_STRIDE;
  // Inst{83}    = MEGA_FETCH;
  // Inst{84}    = ALT_CONST;
  // Inst{86-85} = BUFFER_INDEX_MODE;
  // Inst{95-86} = 0; Reserved

  // VTX_WORD3 (Padding)
  //
  // Inst{127-96} = 0;

  let VTXInst = 1;
}

class LoadParamFrag <PatFrag load_type> : PatFrag <
  (ops node:$ptr), (load_type node:$ptr),
  [{ return isConstantLoad(dyn_cast<LoadSDNode>(N), 0); }]
>;

def load_param : LoadParamFrag<load>;
def load_param_exti8 : LoadParamFrag<az_extloadi8>;
def load_param_exti16 : LoadParamFrag<az_extloadi16>;

def isR600 : Predicate<"Subtarget->getGeneration() <= AMDGPUSubtarget::R700">;

def isR600toCayman
    : Predicate<
          "Subtarget->getGeneration() <= AMDGPUSubtarget::NORTHERN_ISLANDS">;

//===----------------------------------------------------------------------===//
// R600 SDNodes
//===----------------------------------------------------------------------===//

def INTERP_PAIR_XY :  AMDGPUShaderInst <
  (outs R600_TReg32_X:$dst0, R600_TReg32_Y:$dst1),
  (ins i32imm:$src0, R600_TReg32_Y:$src1, R600_TReg32_X:$src2),
  "INTERP_PAIR_XY $src0 $src1 $src2 : $dst0 dst1",
  []>;

def INTERP_PAIR_ZW :  AMDGPUShaderInst <
  (outs R600_TReg32_Z:$dst0, R600_TReg32_W:$dst1),
  (ins i32imm:$src0, R600_TReg32_Y:$src1, R600_TReg32_X:$src2),
  "INTERP_PAIR_ZW $src0 $src1 $src2 : $dst0 dst1",
  []>;

def CONST_ADDRESS: SDNode<"AMDGPUISD::CONST_ADDRESS",
  SDTypeProfile<1, -1, [SDTCisInt<0>, SDTCisPtrTy<1>]>,
  [SDNPVariadic]
>;

def DOT4 : SDNode<"AMDGPUISD::DOT4",
  SDTypeProfile<1, 8, [SDTCisFP<0>, SDTCisVT<1, f32>, SDTCisVT<2, f32>,
      SDTCisVT<3, f32>, SDTCisVT<4, f32>, SDTCisVT<5, f32>,
      SDTCisVT<6, f32>, SDTCisVT<7, f32>, SDTCisVT<8, f32>]>,
  []
>;

def COS_HW : SDNode<"AMDGPUISD::COS_HW",
  SDTypeProfile<1, 1, [SDTCisFP<0>, SDTCisFP<1>]>
>;

def SIN_HW : SDNode<"AMDGPUISD::SIN_HW",
  SDTypeProfile<1, 1, [SDTCisFP<0>, SDTCisFP<1>]>
>;

def TEXTURE_FETCH_Type : SDTypeProfile<1, 19, [SDTCisFP<0>]>;

def TEXTURE_FETCH: SDNode<"AMDGPUISD::TEXTURE_FETCH", TEXTURE_FETCH_Type, []>;

multiclass TexPattern<bits<32> TextureOp, Instruction inst, ValueType vt = v4f32> {
def : Pat<(TEXTURE_FETCH (i32 TextureOp), vt:$SRC_GPR,
          (i32 imm:$srcx), (i32 imm:$srcy), (i32 imm:$srcz), (i32 imm:$srcw),
          (i32 imm:$offsetx), (i32 imm:$offsety), (i32 imm:$offsetz),
          (i32 imm:$DST_SEL_X), (i32 imm:$DST_SEL_Y), (i32 imm:$DST_SEL_Z),
          (i32 imm:$DST_SEL_W),
          (i32 imm:$RESOURCE_ID), (i32 imm:$SAMPLER_ID),
          (i32 imm:$COORD_TYPE_X), (i32 imm:$COORD_TYPE_Y), (i32 imm:$COORD_TYPE_Z),
          (i32 imm:$COORD_TYPE_W)),
          (inst R600_Reg128:$SRC_GPR,
          imm:$srcx, imm:$srcy, imm:$srcz, imm:$srcw,
          imm:$offsetx, imm:$offsety, imm:$offsetz,
          imm:$DST_SEL_X, imm:$DST_SEL_Y, imm:$DST_SEL_Z,
          imm:$DST_SEL_W,
          imm:$RESOURCE_ID, imm:$SAMPLER_ID,
          imm:$COORD_TYPE_X, imm:$COORD_TYPE_Y, imm:$COORD_TYPE_Z,
          imm:$COORD_TYPE_W)>;
}

//===----------------------------------------------------------------------===//
// Interpolation Instructions
//===----------------------------------------------------------------------===//

def INTERP_VEC_LOAD :  AMDGPUShaderInst <
  (outs R600_Reg128:$dst),
  (ins i32imm:$src0),
  "INTERP_LOAD $src0 : $dst",
  [(set R600_Reg128:$dst, (int_R600_interp_const imm:$src0))]>;

def INTERP_XY : R600_2OP <0xD6, "INTERP_XY", []> {
  let bank_swizzle = 5;
}

def INTERP_ZW : R600_2OP <0xD7, "INTERP_ZW", []> {
  let bank_swizzle = 5;
}

def INTERP_LOAD_P0 : R600_1OP <0xE0, "INTERP_LOAD_P0", []>;

//===----------------------------------------------------------------------===//
// Export Instructions
//===----------------------------------------------------------------------===//

def ExportType : SDTypeProfile<0, 7, [SDTCisFP<0>, SDTCisInt<1>]>;

def EXPORT: SDNode<"AMDGPUISD::EXPORT", ExportType,
  [SDNPHasChain, SDNPSideEffect]>;

class ExportWord0 {
  field bits<32> Word0;

  bits<13> arraybase;
  bits<2> type;
  bits<7> gpr;
  bits<2> elem_size;

  let Word0{12-0} = arraybase;
  let Word0{14-13} = type;
  let Word0{21-15} = gpr;
  let Word0{22} = 0; // RW_REL
  let Word0{29-23} = 0; // INDEX_GPR
  let Word0{31-30} = elem_size;
}

class ExportSwzWord1 {
  field bits<32> Word1;

  bits<3> sw_x;
  bits<3> sw_y;
  bits<3> sw_z;
  bits<3> sw_w;
  bits<1> eop;
  bits<8> inst;

  let Word1{2-0} = sw_x;
  let Word1{5-3} = sw_y;
  let Word1{8-6} = sw_z;
  let Word1{11-9} = sw_w;
}

class ExportBufWord1 {
  field bits<32> Word1;

  bits<12> arraySize;
  bits<4> compMask;
  bits<1> eop;
  bits<8> inst;

  let Word1{11-0} = arraySize;
  let Word1{15-12} = compMask;
}

multiclass ExportPattern<Instruction ExportInst, bits<8> cf_inst> {
  def : Pat<(int_R600_store_pixel_depth R600_Reg32:$reg),
    (ExportInst
        (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), $reg, sub0),
        0, 61, 0, 7, 7, 7, cf_inst, 0)
  >;

  def : Pat<(int_R600_store_pixel_stencil R600_Reg32:$reg),
    (ExportInst
        (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), $reg, sub0),
        0, 61, 7, 0, 7, 7, cf_inst, 0)
  >;

  def : Pat<(int_R600_store_dummy (i32 imm:$type)),
    (ExportInst
        (v4f32 (IMPLICIT_DEF)), imm:$type, 0, 7, 7, 7, 7, cf_inst, 0)
  >;

  def : Pat<(int_R600_store_dummy 1),
    (ExportInst
        (v4f32 (IMPLICIT_DEF)), 1, 60, 7, 7, 7, 7, cf_inst, 0)
  >;

  def : Pat<(EXPORT (v4f32 R600_Reg128:$src), (i32 imm:$base), (i32 imm:$type),
    (i32 imm:$swz_x), (i32 imm:$swz_y), (i32 imm:$swz_z), (i32 imm:$swz_w)),
        (ExportInst R600_Reg128:$src, imm:$type, imm:$base,
        imm:$swz_x, imm:$swz_y, imm:$swz_z, imm:$swz_w, cf_inst, 0)
  >;

}

multiclass SteamOutputExportPattern<Instruction ExportInst,
    bits<8> buf0inst, bits<8> buf1inst, bits<8> buf2inst, bits<8> buf3inst> {
// Stream0
  def : Pat<(int_R600_store_stream_output (v4f32 R600_Reg128:$src),
      (i32 imm:$arraybase), (i32 0), (i32 imm:$mask)),
      (ExportInst R600_Reg128:$src, 0, imm:$arraybase,
      4095, imm:$mask, buf0inst, 0)>;
// Stream1
  def : Pat<(int_R600_store_stream_output (v4f32 R600_Reg128:$src),
      (i32 imm:$arraybase), (i32 1), (i32 imm:$mask)),
      (ExportInst $src, 0, imm:$arraybase,
      4095, imm:$mask, buf1inst, 0)>;
// Stream2
  def : Pat<(int_R600_store_stream_output (v4f32 R600_Reg128:$src),
      (i32 imm:$arraybase), (i32 2), (i32 imm:$mask)),
      (ExportInst $src, 0, imm:$arraybase,
      4095, imm:$mask, buf2inst, 0)>;
// Stream3
  def : Pat<(int_R600_store_stream_output (v4f32 R600_Reg128:$src),
      (i32 imm:$arraybase), (i32 3), (i32 imm:$mask)),
      (ExportInst $src, 0, imm:$arraybase,
      4095, imm:$mask, buf3inst, 0)>;
}

// Export Instructions should not be duplicated by TailDuplication pass
// (which assumes that duplicable instruction are affected by exec mask)
let usesCustomInserter = 1, isNotDuplicable = 1 in {

class ExportSwzInst : InstR600ISA<(
    outs),
    (ins R600_Reg128:$gpr, i32imm:$type, i32imm:$arraybase,
    RSel:$sw_x, RSel:$sw_y, RSel:$sw_z, RSel:$sw_w, i32imm:$inst,
    i32imm:$eop),
    !strconcat("EXPORT", " $gpr.$sw_x$sw_y$sw_z$sw_w"),
    []>, ExportWord0, ExportSwzWord1 {
  let elem_size = 3;
  let Inst{31-0} = Word0;
  let Inst{63-32} = Word1;
  let IsExport = 1;
}

} // End usesCustomInserter = 1

class ExportBufInst : InstR600ISA<(
    outs),
    (ins R600_Reg128:$gpr, i32imm:$type, i32imm:$arraybase,
    i32imm:$arraySize, i32imm:$compMask, i32imm:$inst, i32imm:$eop),
    !strconcat("EXPORT", " $gpr"),
    []>, ExportWord0, ExportBufWord1 {
  let elem_size = 0;
  let Inst{31-0} = Word0;
  let Inst{63-32} = Word1;
  let IsExport = 1;
}

//===----------------------------------------------------------------------===//
// Control Flow Instructions
//===----------------------------------------------------------------------===//


def KCACHE : InstFlag<"printKCache">;

class ALU_CLAUSE<bits<4> inst, string OpName> : AMDGPUInst <(outs),
(ins i32imm:$ADDR, i32imm:$KCACHE_BANK0, i32imm:$KCACHE_BANK1,
KCACHE:$KCACHE_MODE0, KCACHE:$KCACHE_MODE1,
i32imm:$KCACHE_ADDR0, i32imm:$KCACHE_ADDR1,
i32imm:$COUNT, i32imm:$Enabled),
!strconcat(OpName, " $COUNT, @$ADDR, "
"KC0[$KCACHE_MODE0], KC1[$KCACHE_MODE1]"),
[] >, CF_ALU_WORD0, CF_ALU_WORD1 {
  field bits<64> Inst;

  let CF_INST = inst;
  let ALT_CONST = 0;
  let WHOLE_QUAD_MODE = 0;
  let BARRIER = 1;
  let isCodeGenOnly = 1;
  let UseNamedOperandTable = 1;

  let Inst{31-0} = Word0;
  let Inst{63-32} = Word1;
}

class CF_WORD0_R600 {
  field bits<32> Word0;

  bits<32> ADDR;

  let Word0 = ADDR;
}

class CF_CLAUSE_R600 <bits<7> inst, dag ins, string AsmPrint> : AMDGPUInst <(outs),
ins, AsmPrint, [] >, CF_WORD0_R600, CF_WORD1_R600 {
  field bits<64> Inst;
  bits<4> CNT;

  let CF_INST = inst;
  let BARRIER = 1;
  let CF_CONST = 0;
  let VALID_PIXEL_MODE = 0;
  let COND = 0;
  let COUNT = CNT{2-0};
  let CALL_COUNT = 0;
  let COUNT_3 = CNT{3};
  let END_OF_PROGRAM = 0;
  let WHOLE_QUAD_MODE = 0;

  let Inst{31-0} = Word0;
  let Inst{63-32} = Word1;
}

class CF_CLAUSE_EG <bits<8> inst, dag ins, string AsmPrint> : AMDGPUInst <(outs),
ins, AsmPrint, [] >, CF_WORD0_EG, CF_WORD1_EG {
  field bits<64> Inst;

  let CF_INST = inst;
  let BARRIER = 1;
  let JUMPTABLE_SEL = 0;
  let CF_CONST = 0;
  let VALID_PIXEL_MODE = 0;
  let COND = 0;
  let END_OF_PROGRAM = 0;

  let Inst{31-0} = Word0;
  let Inst{63-32} = Word1;
}

def CF_ALU : ALU_CLAUSE<8, "ALU">;
def CF_ALU_PUSH_BEFORE : ALU_CLAUSE<9, "ALU_PUSH_BEFORE">;
def CF_ALU_POP_AFTER : ALU_CLAUSE<10, "ALU_POP_AFTER">;
def CF_ALU_CONTINUE : ALU_CLAUSE<13, "ALU_CONTINUE">;
def CF_ALU_BREAK : ALU_CLAUSE<14, "ALU_BREAK">;
def CF_ALU_ELSE_AFTER : ALU_CLAUSE<15, "ALU_ELSE_AFTER">;

def FETCH_CLAUSE : AMDGPUInst <(outs),
(ins i32imm:$addr), "Fetch clause starting at $addr:", [] > {
  field bits<8> Inst;
  bits<8> num;
  let Inst = num;
  let isCodeGenOnly = 1;
}

def ALU_CLAUSE : AMDGPUInst <(outs),
(ins i32imm:$addr), "ALU clause starting at $addr:", [] > {
  field bits<8> Inst;
  bits<8> num;
  let Inst = num;
  let isCodeGenOnly = 1;
}

def LITERALS : AMDGPUInst <(outs),
(ins LITERAL:$literal1, LITERAL:$literal2), "$literal1, $literal2", [] > {
  let isCodeGenOnly = 1;

  field bits<64> Inst;
  bits<32> literal1;
  bits<32> literal2;

  let Inst{31-0} = literal1;
  let Inst{63-32} = literal2;
}

def PAD : AMDGPUInst <(outs), (ins), "PAD", [] > {
  field bits<64> Inst;
}

let Predicates = [isR600toCayman] in {

//===----------------------------------------------------------------------===//
// Common Instructions R600, R700, Evergreen, Cayman
//===----------------------------------------------------------------------===//

def ADD : R600_2OP_Helper <0x0, "ADD", fadd>;
// Non-IEEE MUL: 0 * anything = 0
def MUL : R600_2OP_Helper <0x1, "MUL NON-IEEE", int_AMDGPU_mul>;
def MUL_IEEE : R600_2OP_Helper <0x2, "MUL_IEEE", fmul>;
// TODO: Do these actually match the regular fmin/fmax behavior?
def MAX : R600_2OP_Helper <0x3, "MAX", AMDGPUfmax_legacy>;
def MIN : R600_2OP_Helper <0x4, "MIN", AMDGPUfmin_legacy>;
// According to https://msdn.microsoft.com/en-us/library/windows/desktop/cc308050%28v=vs.85%29.aspx
// DX10 min/max returns the other operand if one is NaN,
// this matches http://llvm.org/docs/LangRef.html#llvm-minnum-intrinsic
def MAX_DX10 : R600_2OP_Helper <0x5, "MAX_DX10", fmaxnum>;
def MIN_DX10 : R600_2OP_Helper <0x6, "MIN_DX10", fminnum>;

// For the SET* instructions there is a naming conflict in TargetSelectionDAG.td,
// so some of the instruction names don't match the asm string.
// XXX: Use the defs in TargetSelectionDAG.td instead of intrinsics.
def SETE : R600_2OP <
  0x08, "SETE",
  [(set f32:$dst, (selectcc f32:$src0, f32:$src1, FP_ONE, FP_ZERO, COND_OEQ))]
>;

def SGT : R600_2OP <
  0x09, "SETGT",
  [(set f32:$dst, (selectcc f32:$src0, f32:$src1, FP_ONE, FP_ZERO, COND_OGT))]
>;

def SGE : R600_2OP <
  0xA, "SETGE",
  [(set f32:$dst, (selectcc f32:$src0, f32:$src1, FP_ONE, FP_ZERO, COND_OGE))]
>;

def SNE : R600_2OP <
  0xB, "SETNE",
  [(set f32:$dst, (selectcc f32:$src0, f32:$src1, FP_ONE, FP_ZERO, COND_UNE_NE))]
>;

def SETE_DX10 : R600_2OP <
  0xC, "SETE_DX10",
  [(set i32:$dst, (selectcc f32:$src0, f32:$src1, -1, 0, COND_OEQ))]
>;

def SETGT_DX10 : R600_2OP <
  0xD, "SETGT_DX10",
  [(set i32:$dst, (selectcc f32:$src0, f32:$src1, -1, 0, COND_OGT))]
>;

def SETGE_DX10 : R600_2OP <
  0xE, "SETGE_DX10",
  [(set i32:$dst, (selectcc f32:$src0, f32:$src1, -1, 0, COND_OGE))]
>;

// FIXME: This should probably be COND_ONE
def SETNE_DX10 : R600_2OP <
  0xF, "SETNE_DX10",
  [(set i32:$dst, (selectcc f32:$src0, f32:$src1, -1, 0, COND_UNE_NE))]
>;

def FRACT : R600_1OP_Helper <0x10, "FRACT", AMDGPUfract>;
def TRUNC : R600_1OP_Helper <0x11, "TRUNC", ftrunc>;
def CEIL : R600_1OP_Helper <0x12, "CEIL", fceil>;
def RNDNE : R600_1OP_Helper <0x13, "RNDNE", frint>;
def FLOOR : R600_1OP_Helper <0x14, "FLOOR", ffloor>;

def MOV : R600_1OP <0x19, "MOV", []>;

let isPseudo = 1, isCodeGenOnly = 1, usesCustomInserter = 1 in {

class MOV_IMM <ValueType vt, Operand immType> : AMDGPUInst <
  (outs R600_Reg32:$dst),
  (ins immType:$imm),
  "",
  []
>;

} // end let isPseudo = 1, isCodeGenOnly = 1, usesCustomInserter = 1

def MOV_IMM_I32 : MOV_IMM<i32, i32imm>;
def : Pat <
  (imm:$val),
  (MOV_IMM_I32 imm:$val)
>;

def MOV_IMM_F32 : MOV_IMM<f32, f32imm>;
def : Pat <
  (fpimm:$val),
  (MOV_IMM_F32  fpimm:$val)
>;

def PRED_SETE : R600_2OP <0x20, "PRED_SETE", []>;
def PRED_SETGT : R600_2OP <0x21, "PRED_SETGT", []>;
def PRED_SETGE : R600_2OP <0x22, "PRED_SETGE", []>;
def PRED_SETNE : R600_2OP <0x23, "PRED_SETNE", []>;

let hasSideEffects = 1 in {

def KILLGT : R600_2OP <0x2D, "KILLGT", []>;

} // end hasSideEffects

def AND_INT : R600_2OP_Helper <0x30, "AND_INT", and>;
def OR_INT : R600_2OP_Helper <0x31, "OR_INT", or>;
def XOR_INT : R600_2OP_Helper <0x32, "XOR_INT", xor>;
def NOT_INT : R600_1OP_Helper <0x33, "NOT_INT", not>;
def ADD_INT : R600_2OP_Helper <0x34, "ADD_INT", add>;
def SUB_INT : R600_2OP_Helper <0x35, "SUB_INT", sub>;
def MAX_INT : R600_2OP_Helper <0x36, "MAX_INT", smax>;
def MIN_INT : R600_2OP_Helper <0x37, "MIN_INT", smin>;
def MAX_UINT : R600_2OP_Helper <0x38, "MAX_UINT", umax>;
def MIN_UINT : R600_2OP_Helper <0x39, "MIN_UINT", umin>;

def SETE_INT : R600_2OP <
  0x3A, "SETE_INT",
  [(set i32:$dst, (selectcc i32:$src0, i32:$src1, -1, 0, SETEQ))]
>;

def SETGT_INT : R600_2OP <
  0x3B, "SETGT_INT",
  [(set i32:$dst, (selectcc i32:$src0, i32:$src1, -1, 0, SETGT))]
>;

def SETGE_INT : R600_2OP <
  0x3C, "SETGE_INT",
  [(set i32:$dst, (selectcc i32:$src0, i32:$src1, -1, 0, SETGE))]
>;

def SETNE_INT : R600_2OP <
  0x3D, "SETNE_INT",
  [(set i32:$dst, (selectcc i32:$src0, i32:$src1, -1, 0, SETNE))]
>;

def SETGT_UINT : R600_2OP <
  0x3E, "SETGT_UINT",
  [(set i32:$dst, (selectcc i32:$src0, i32:$src1, -1, 0, SETUGT))]
>;

def SETGE_UINT : R600_2OP <
  0x3F, "SETGE_UINT",
  [(set i32:$dst, (selectcc i32:$src0, i32:$src1, -1, 0, SETUGE))]
>;

def PRED_SETE_INT : R600_2OP <0x42, "PRED_SETE_INT", []>;
def PRED_SETGT_INT : R600_2OP <0x43, "PRED_SETGE_INT", []>;
def PRED_SETGE_INT : R600_2OP <0x44, "PRED_SETGE_INT", []>;
def PRED_SETNE_INT : R600_2OP <0x45, "PRED_SETNE_INT", []>;

def CNDE_INT : R600_3OP <
  0x1C, "CNDE_INT",
  [(set i32:$dst, (selectcc i32:$src0, 0, i32:$src1, i32:$src2, COND_EQ))]
>;

def CNDGE_INT : R600_3OP <
  0x1E, "CNDGE_INT",
  [(set i32:$dst, (selectcc i32:$src0, 0, i32:$src1, i32:$src2, COND_SGE))]
>;

def CNDGT_INT : R600_3OP <
  0x1D, "CNDGT_INT",
  [(set i32:$dst, (selectcc i32:$src0, 0, i32:$src1, i32:$src2, COND_SGT))]
>;

//===----------------------------------------------------------------------===//
// Texture instructions
//===----------------------------------------------------------------------===//

let mayLoad = 0, mayStore = 0, hasSideEffects = 0 in {

class R600_TEX <bits<11> inst, string opName> :
  InstR600 <(outs R600_Reg128:$DST_GPR),
          (ins R600_Reg128:$SRC_GPR,
          RSel:$srcx, RSel:$srcy, RSel:$srcz, RSel:$srcw,
          i32imm:$offsetx, i32imm:$offsety, i32imm:$offsetz,
          RSel:$DST_SEL_X, RSel:$DST_SEL_Y, RSel:$DST_SEL_Z, RSel:$DST_SEL_W,
          i32imm:$RESOURCE_ID, i32imm:$SAMPLER_ID,
          CT:$COORD_TYPE_X, CT:$COORD_TYPE_Y, CT:$COORD_TYPE_Z,
          CT:$COORD_TYPE_W),
          !strconcat(opName,
          " $DST_GPR.$DST_SEL_X$DST_SEL_Y$DST_SEL_Z$DST_SEL_W, "
          "$SRC_GPR.$srcx$srcy$srcz$srcw "
          "RID:$RESOURCE_ID SID:$SAMPLER_ID "
          "CT:$COORD_TYPE_X$COORD_TYPE_Y$COORD_TYPE_Z$COORD_TYPE_W"),
          [],
          NullALU>, TEX_WORD0, TEX_WORD1, TEX_WORD2 {
  let Inst{31-0} = Word0;
  let Inst{63-32} = Word1;

  let TEX_INST = inst{4-0};
  let SRC_REL = 0;
  let DST_REL = 0;
  let LOD_BIAS = 0;

  let INST_MOD = 0;
  let FETCH_WHOLE_QUAD = 0;
  let ALT_CONST = 0;
  let SAMPLER_INDEX_MODE = 0;
  let RESOURCE_INDEX_MODE = 0;

  let TEXInst = 1;
}

} // End mayLoad = 0, mayStore = 0, hasSideEffects = 0



def TEX_SAMPLE : R600_TEX <0x10, "TEX_SAMPLE">;
def TEX_SAMPLE_C : R600_TEX <0x18, "TEX_SAMPLE_C">;
def TEX_SAMPLE_L : R600_TEX <0x11, "TEX_SAMPLE_L">;
def TEX_SAMPLE_C_L : R600_TEX <0x19, "TEX_SAMPLE_C_L">;
def TEX_SAMPLE_LB : R600_TEX <0x12, "TEX_SAMPLE_LB">;
def TEX_SAMPLE_C_LB : R600_TEX <0x1A, "TEX_SAMPLE_C_LB">;
def TEX_LD : R600_TEX <0x03, "TEX_LD">;
def TEX_LDPTR : R600_TEX <0x03, "TEX_LDPTR"> {
  let INST_MOD = 1;
}
def TEX_GET_TEXTURE_RESINFO : R600_TEX <0x04, "TEX_GET_TEXTURE_RESINFO">;
def TEX_GET_GRADIENTS_H : R600_TEX <0x07, "TEX_GET_GRADIENTS_H">;
def TEX_GET_GRADIENTS_V : R600_TEX <0x08, "TEX_GET_GRADIENTS_V">;
def TEX_SET_GRADIENTS_H : R600_TEX <0x0B, "TEX_SET_GRADIENTS_H">;
def TEX_SET_GRADIENTS_V : R600_TEX <0x0C, "TEX_SET_GRADIENTS_V">;
def TEX_SAMPLE_G : R600_TEX <0x14, "TEX_SAMPLE_G">;
def TEX_SAMPLE_C_G : R600_TEX <0x1C, "TEX_SAMPLE_C_G">;

defm : TexPattern<0, TEX_SAMPLE>;
defm : TexPattern<1, TEX_SAMPLE_C>;
defm : TexPattern<2, TEX_SAMPLE_L>;
defm : TexPattern<3, TEX_SAMPLE_C_L>;
defm : TexPattern<4, TEX_SAMPLE_LB>;
defm : TexPattern<5, TEX_SAMPLE_C_LB>;
defm : TexPattern<6, TEX_LD, v4i32>;
defm : TexPattern<7, TEX_GET_TEXTURE_RESINFO, v4i32>;
defm : TexPattern<8, TEX_GET_GRADIENTS_H>;
defm : TexPattern<9, TEX_GET_GRADIENTS_V>;
defm : TexPattern<10, TEX_LDPTR, v4i32>;

//===----------------------------------------------------------------------===//
// Helper classes for common instructions
//===----------------------------------------------------------------------===//

class MUL_LIT_Common <bits<5> inst> : R600_3OP <
  inst, "MUL_LIT",
  []
>;

class MULADD_Common <bits<5> inst> : R600_3OP <
  inst, "MULADD",
  []
>;

class MULADD_IEEE_Common <bits<5> inst> : R600_3OP <
  inst, "MULADD_IEEE",
  [(set f32:$dst, (fmad f32:$src0, f32:$src1, f32:$src2))]
>;

class FMA_Common <bits<5> inst> : R600_3OP <
  inst, "FMA",
  [(set f32:$dst, (fma f32:$src0, f32:$src1, f32:$src2))], VecALU
>;

class CNDE_Common <bits<5> inst> : R600_3OP <
  inst, "CNDE",
  [(set f32:$dst, (selectcc f32:$src0, FP_ZERO, f32:$src1, f32:$src2, COND_OEQ))]
>;

class CNDGT_Common <bits<5> inst> : R600_3OP <
  inst, "CNDGT",
  [(set f32:$dst, (selectcc f32:$src0, FP_ZERO, f32:$src1, f32:$src2, COND_OGT))]
> {
  let Itinerary = VecALU;
}

class CNDGE_Common <bits<5> inst> : R600_3OP <
  inst, "CNDGE",
  [(set f32:$dst, (selectcc f32:$src0, FP_ZERO, f32:$src1, f32:$src2, COND_OGE))]
> {
  let Itinerary = VecALU;
}


let isCodeGenOnly = 1, isPseudo = 1, Namespace = "AMDGPU"  in {
class R600_VEC2OP<list<dag> pattern> : InstR600 <(outs R600_Reg32:$dst), (ins
// Slot X
   UEM:$update_exec_mask_X, UP:$update_pred_X, WRITE:$write_X,
   OMOD:$omod_X, REL:$dst_rel_X, CLAMP:$clamp_X,
   R600_TReg32_X:$src0_X, NEG:$src0_neg_X, REL:$src0_rel_X, ABS:$src0_abs_X, SEL:$src0_sel_X,
   R600_TReg32_X:$src1_X, NEG:$src1_neg_X, REL:$src1_rel_X, ABS:$src1_abs_X, SEL:$src1_sel_X,
   R600_Pred:$pred_sel_X,
// Slot Y
   UEM:$update_exec_mask_Y, UP:$update_pred_Y, WRITE:$write_Y,
   OMOD:$omod_Y, REL:$dst_rel_Y, CLAMP:$clamp_Y,
   R600_TReg32_Y:$src0_Y, NEG:$src0_neg_Y, REL:$src0_rel_Y, ABS:$src0_abs_Y, SEL:$src0_sel_Y,
   R600_TReg32_Y:$src1_Y, NEG:$src1_neg_Y, REL:$src1_rel_Y, ABS:$src1_abs_Y, SEL:$src1_sel_Y,
   R600_Pred:$pred_sel_Y,
// Slot Z
   UEM:$update_exec_mask_Z, UP:$update_pred_Z, WRITE:$write_Z,
   OMOD:$omod_Z, REL:$dst_rel_Z, CLAMP:$clamp_Z,
   R600_TReg32_Z:$src0_Z, NEG:$src0_neg_Z, REL:$src0_rel_Z, ABS:$src0_abs_Z, SEL:$src0_sel_Z,
   R600_TReg32_Z:$src1_Z, NEG:$src1_neg_Z, REL:$src1_rel_Z, ABS:$src1_abs_Z, SEL:$src1_sel_Z,
   R600_Pred:$pred_sel_Z,
// Slot W
   UEM:$update_exec_mask_W, UP:$update_pred_W, WRITE:$write_W,
   OMOD:$omod_W, REL:$dst_rel_W, CLAMP:$clamp_W,
   R600_TReg32_W:$src0_W, NEG:$src0_neg_W, REL:$src0_rel_W, ABS:$src0_abs_W, SEL:$src0_sel_W,
   R600_TReg32_W:$src1_W, NEG:$src1_neg_W, REL:$src1_rel_W, ABS:$src1_abs_W, SEL:$src1_sel_W,
   R600_Pred:$pred_sel_W,
   LITERAL:$literal0, LITERAL:$literal1),
  "",
  pattern,
  AnyALU> {

  let UseNamedOperandTable = 1;

}
}

def DOT_4 : R600_VEC2OP<[(set R600_Reg32:$dst, (DOT4
  R600_TReg32_X:$src0_X, R600_TReg32_X:$src1_X,
  R600_TReg32_Y:$src0_Y, R600_TReg32_Y:$src1_Y,
  R600_TReg32_Z:$src0_Z, R600_TReg32_Z:$src1_Z,
  R600_TReg32_W:$src0_W, R600_TReg32_W:$src1_W))]>;


class DOT4_Common <bits<11> inst> : R600_2OP <inst, "DOT4", []>;


let mayLoad = 0, mayStore = 0, hasSideEffects = 0 in {
multiclass CUBE_Common <bits<11> inst> {

  def _pseudo : InstR600 <
    (outs R600_Reg128:$dst),
    (ins R600_Reg128:$src0),
    "CUBE $dst $src0",
    [(set v4f32:$dst, (int_AMDGPU_cube v4f32:$src0))],
    VecALU
  > {
    let isPseudo = 1;
    let UseNamedOperandTable = 1;
  }

  def _real : R600_2OP <inst, "CUBE", []>;
}
} // End mayLoad = 0, mayStore = 0, hasSideEffects = 0

class EXP_IEEE_Common <bits<11> inst> : R600_1OP_Helper <
  inst, "EXP_IEEE", fexp2
> {
  let Itinerary = TransALU;
}

class FLT_TO_INT_Common <bits<11> inst> : R600_1OP_Helper <
  inst, "FLT_TO_INT", fp_to_sint
> {
  let Itinerary = TransALU;
}

class INT_TO_FLT_Common <bits<11> inst> : R600_1OP_Helper <
  inst, "INT_TO_FLT", sint_to_fp
> {
  let Itinerary = TransALU;
}

class FLT_TO_UINT_Common <bits<11> inst> : R600_1OP_Helper <
  inst, "FLT_TO_UINT", fp_to_uint
> {
  let Itinerary = TransALU;
}

class UINT_TO_FLT_Common <bits<11> inst> : R600_1OP_Helper <
  inst, "UINT_TO_FLT", uint_to_fp
> {
  let Itinerary = TransALU;
}

class LOG_CLAMPED_Common <bits<11> inst> : R600_1OP <
  inst, "LOG_CLAMPED", []
>;

class LOG_IEEE_Common <bits<11> inst> : R600_1OP_Helper <
  inst, "LOG_IEEE", flog2
> {
  let Itinerary = TransALU;
}

class LSHL_Common <bits<11> inst> : R600_2OP_Helper <inst, "LSHL", shl>;
class LSHR_Common <bits<11> inst> : R600_2OP_Helper <inst, "LSHR", srl>;
class ASHR_Common <bits<11> inst> : R600_2OP_Helper <inst, "ASHR", sra>;
class MULHI_INT_Common <bits<11> inst> : R600_2OP_Helper <
  inst, "MULHI_INT", mulhs
> {
  let Itinerary = TransALU;
}
class MULHI_UINT_Common <bits<11> inst> : R600_2OP_Helper <
  inst, "MULHI", mulhu
> {
  let Itinerary = TransALU;
}
class MULLO_INT_Common <bits<11> inst> : R600_2OP_Helper <
  inst, "MULLO_INT", mul
> {
  let Itinerary = TransALU;
}
class MULLO_UINT_Common <bits<11> inst> : R600_2OP <inst, "MULLO_UINT", []> {
  let Itinerary = TransALU;
}

class RECIP_CLAMPED_Common <bits<11> inst> : R600_1OP <
  inst, "RECIP_CLAMPED", []
> {
  let Itinerary = TransALU;
}

class RECIP_IEEE_Common <bits<11> inst> : R600_1OP <
  inst, "RECIP_IEEE", [(set f32:$dst, (AMDGPUrcp f32:$src0))]
> {
  let Itinerary = TransALU;
}

class RECIP_UINT_Common <bits<11> inst> : R600_1OP_Helper <
  inst, "RECIP_UINT", AMDGPUurecip
> {
  let Itinerary = TransALU;
}

// Clamped to maximum.
class RECIPSQRT_CLAMPED_Common <bits<11> inst> : R600_1OP_Helper <
  inst, "RECIPSQRT_CLAMPED", AMDGPUrsq_clamped
> {
  let Itinerary = TransALU;
}

class RECIPSQRT_IEEE_Common <bits<11> inst> : R600_1OP_Helper <
  inst, "RECIPSQRT_IEEE", AMDGPUrsq_legacy
> {
  let Itinerary = TransALU;
}

// TODO: There is also RECIPSQRT_FF which clamps to zero.

class SIN_Common <bits<11> inst> : R600_1OP <
  inst, "SIN", [(set f32:$dst, (SIN_HW f32:$src0))]>{
  let Trig = 1;
  let Itinerary = TransALU;
}

class COS_Common <bits<11> inst> : R600_1OP <
  inst, "COS", [(set f32:$dst, (COS_HW f32:$src0))]> {
  let Trig = 1;
  let Itinerary = TransALU;
}

def CLAMP_R600 :  CLAMP <R600_Reg32>;
def FABS_R600 : FABS<R600_Reg32>;
def FNEG_R600 : FNEG<R600_Reg32>;

//===----------------------------------------------------------------------===//
// Helper patterns for complex intrinsics
//===----------------------------------------------------------------------===//

// FIXME: Should be predicated on unsafe fp math.
multiclass DIV_Common <InstR600 recip_ieee> {
def : Pat<
  (int_AMDGPU_div f32:$src0, f32:$src1),
  (MUL_IEEE $src0, (recip_ieee $src1))
>;

def : Pat<
  (fdiv f32:$src0, f32:$src1),
  (MUL_IEEE $src0, (recip_ieee $src1))
>;

def : RcpPat<recip_ieee, f32>;
}

class TGSI_LIT_Z_Common <InstR600 mul_lit, InstR600 log_clamped, InstR600 exp_ieee>
  : Pat <
  (int_TGSI_lit_z f32:$src_x, f32:$src_y, f32:$src_w),
  (exp_ieee (mul_lit (log_clamped (MAX $src_y, (f32 ZERO))), $src_w, $src_x))
>;

//===----------------------------------------------------------------------===//
// R600 / R700 Instructions
//===----------------------------------------------------------------------===//

let Predicates = [isR600] in {

  def MUL_LIT_r600 : MUL_LIT_Common<0x0C>;
  def MULADD_r600 : MULADD_Common<0x10>;
  def MULADD_IEEE_r600 : MULADD_IEEE_Common<0x14>;
  def CNDE_r600 : CNDE_Common<0x18>;
  def CNDGT_r600 : CNDGT_Common<0x19>;
  def CNDGE_r600 : CNDGE_Common<0x1A>;
  def DOT4_r600 : DOT4_Common<0x50>;
  defm CUBE_r600 : CUBE_Common<0x52>;
  def EXP_IEEE_r600 : EXP_IEEE_Common<0x61>;
  def LOG_CLAMPED_r600 : LOG_CLAMPED_Common<0x62>;
  def LOG_IEEE_r600 : LOG_IEEE_Common<0x63>;
  def RECIP_CLAMPED_r600 : RECIP_CLAMPED_Common<0x64>;
  def RECIP_IEEE_r600 : RECIP_IEEE_Common<0x66>;
  def RECIPSQRT_CLAMPED_r600 : RECIPSQRT_CLAMPED_Common<0x67>;
  def RECIPSQRT_IEEE_r600 : RECIPSQRT_IEEE_Common<0x69>;
  def FLT_TO_INT_r600 : FLT_TO_INT_Common<0x6b>;
  def INT_TO_FLT_r600 : INT_TO_FLT_Common<0x6c>;
  def FLT_TO_UINT_r600 : FLT_TO_UINT_Common<0x79>;
  def UINT_TO_FLT_r600 : UINT_TO_FLT_Common<0x6d>;
  def SIN_r600 : SIN_Common<0x6E>;
  def COS_r600 : COS_Common<0x6F>;
  def ASHR_r600 : ASHR_Common<0x70>;
  def LSHR_r600 : LSHR_Common<0x71>;
  def LSHL_r600 : LSHL_Common<0x72>;
  def MULLO_INT_r600 : MULLO_INT_Common<0x73>;
  def MULHI_INT_r600 : MULHI_INT_Common<0x74>;
  def MULLO_UINT_r600 : MULLO_UINT_Common<0x75>;
  def MULHI_UINT_r600 : MULHI_UINT_Common<0x76>;
  def RECIP_UINT_r600 : RECIP_UINT_Common <0x78>;

  defm DIV_r600 : DIV_Common<RECIP_IEEE_r600>;
  def : POW_Common <LOG_IEEE_r600, EXP_IEEE_r600, MUL>;
  def TGSI_LIT_Z_r600 : TGSI_LIT_Z_Common<MUL_LIT_r600, LOG_CLAMPED_r600, EXP_IEEE_r600>;

  def : Pat<(fsqrt f32:$src), (MUL $src, (RECIPSQRT_CLAMPED_r600 $src))>;
  def : RsqPat<RECIPSQRT_IEEE_r600, f32>;

  def R600_ExportSwz : ExportSwzInst {
    let Word1{20-17} = 0; // BURST_COUNT
    let Word1{21} = eop;
    let Word1{22} = 0; // VALID_PIXEL_MODE
    let Word1{30-23} = inst;
    let Word1{31} = 1; // BARRIER
  }
  defm : ExportPattern<R600_ExportSwz, 39>;

  def R600_ExportBuf : ExportBufInst {
    let Word1{20-17} = 0; // BURST_COUNT
    let Word1{21} = eop;
    let Word1{22} = 0; // VALID_PIXEL_MODE
    let Word1{30-23} = inst;
    let Word1{31} = 1; // BARRIER
  }
  defm : SteamOutputExportPattern<R600_ExportBuf, 0x20, 0x21, 0x22, 0x23>;

  def CF_TC_R600 : CF_CLAUSE_R600<1, (ins i32imm:$ADDR, i32imm:$CNT),
  "TEX $CNT @$ADDR"> {
    let POP_COUNT = 0;
  }
  def CF_VC_R600 : CF_CLAUSE_R600<2, (ins i32imm:$ADDR, i32imm:$CNT),
  "VTX $CNT @$ADDR"> {
    let POP_COUNT = 0;
  }
  def WHILE_LOOP_R600 : CF_CLAUSE_R600<6, (ins i32imm:$ADDR),
  "LOOP_START_DX10 @$ADDR"> {
    let POP_COUNT = 0;
    let CNT = 0;
  }
  def END_LOOP_R600 : CF_CLAUSE_R600<5, (ins i32imm:$ADDR), "END_LOOP @$ADDR"> {
    let POP_COUNT = 0;
    let CNT = 0;
  }
  def LOOP_BREAK_R600 : CF_CLAUSE_R600<9, (ins i32imm:$ADDR),
  "LOOP_BREAK @$ADDR"> {
    let POP_COUNT = 0;
    let CNT = 0;
  }
  def CF_CONTINUE_R600 : CF_CLAUSE_R600<8, (ins i32imm:$ADDR),
  "CONTINUE @$ADDR"> {
    let POP_COUNT = 0;
    let CNT = 0;
  }
  def CF_JUMP_R600 : CF_CLAUSE_R600<10, (ins i32imm:$ADDR, i32imm:$POP_COUNT),
  "JUMP @$ADDR POP:$POP_COUNT"> {
    let CNT = 0;
  }
  def CF_PUSH_ELSE_R600 : CF_CLAUSE_R600<12, (ins i32imm:$ADDR),
  "PUSH_ELSE @$ADDR"> {
    let CNT = 0;
    let POP_COUNT = 0; // FIXME?
  }
  def CF_ELSE_R600 : CF_CLAUSE_R600<13, (ins i32imm:$ADDR, i32imm:$POP_COUNT),
  "ELSE @$ADDR POP:$POP_COUNT"> {
    let CNT = 0;
  }
  def CF_CALL_FS_R600 : CF_CLAUSE_R600<19, (ins), "CALL_FS"> {
    let ADDR = 0;
    let CNT = 0;
    let POP_COUNT = 0;
  }
  def POP_R600 : CF_CLAUSE_R600<14, (ins i32imm:$ADDR, i32imm:$POP_COUNT),
  "POP @$ADDR POP:$POP_COUNT"> {
    let CNT = 0;
  }
  def CF_END_R600 : CF_CLAUSE_R600<0, (ins), "CF_END"> {
    let CNT = 0;
    let POP_COUNT = 0;
    let ADDR = 0;
    let END_OF_PROGRAM = 1;
  }

}


//===----------------------------------------------------------------------===//
// Regist loads and stores - for indirect addressing
//===----------------------------------------------------------------------===//

defm R600_ : RegisterLoadStore <R600_Reg32, FRAMEri, ADDRIndirect>;


//===----------------------------------------------------------------------===//
// Pseudo instructions
//===----------------------------------------------------------------------===//

let isPseudo = 1 in {

def PRED_X : InstR600 <
  (outs R600_Predicate_Bit:$dst),
  (ins R600_Reg32:$src0, i32imm:$src1, i32imm:$flags),
  "", [], NullALU> {
  let FlagOperandIdx = 3;
}

let isTerminator = 1, isBranch = 1 in {
def JUMP_COND : InstR600 <
          (outs),
          (ins brtarget:$target, R600_Predicate_Bit:$p),
          "JUMP $target ($p)",
          [], AnyALU
  >;

def JUMP : InstR600 <
          (outs),
          (ins brtarget:$target),
          "JUMP $target",
          [], AnyALU
  >
{
  let isPredicable = 1;
  let isBarrier = 1;
}

}  // End isTerminator = 1, isBranch = 1

let usesCustomInserter = 1 in {

let mayLoad = 0, mayStore = 0, hasSideEffects = 1 in {

def MASK_WRITE : AMDGPUShaderInst <
    (outs),
    (ins R600_Reg32:$src),
    "MASK_WRITE $src",
    []
>;

} // End mayLoad = 0, mayStore = 0, hasSideEffects = 1


def TXD: InstR600 <
  (outs R600_Reg128:$dst),
  (ins R600_Reg128:$src0, R600_Reg128:$src1, R600_Reg128:$src2,
       i32imm:$resourceId, i32imm:$samplerId, i32imm:$textureTarget),
  "TXD $dst, $src0, $src1, $src2, $resourceId, $samplerId, $textureTarget",
  [(set v4f32:$dst, (int_AMDGPU_txd v4f32:$src0, v4f32:$src1, v4f32:$src2,
                     imm:$resourceId, imm:$samplerId, imm:$textureTarget))],
  NullALU > {
  let TEXInst = 1;
}

def TXD_SHADOW: InstR600 <
  (outs R600_Reg128:$dst),
  (ins R600_Reg128:$src0, R600_Reg128:$src1, R600_Reg128:$src2,
       i32imm:$resourceId, i32imm:$samplerId, i32imm:$textureTarget),
  "TXD_SHADOW $dst, $src0, $src1, $src2, $resourceId, $samplerId, $textureTarget",
  [(set v4f32:$dst, (int_AMDGPU_txd v4f32:$src0, v4f32:$src1, v4f32:$src2,
        imm:$resourceId, imm:$samplerId, TEX_SHADOW:$textureTarget))],
   NullALU
> {
  let TEXInst = 1;
}
} // End isPseudo = 1
} // End usesCustomInserter = 1


//===----------------------------------------------------------------------===//
// Constant Buffer Addressing Support
//===----------------------------------------------------------------------===//

let usesCustomInserter = 1, isCodeGenOnly = 1, isPseudo = 1, Namespace = "AMDGPU"  in {
def CONST_COPY : Instruction {
  let OutOperandList = (outs R600_Reg32:$dst);
  let InOperandList = (ins i32imm:$src);
  let Pattern =
      [(set R600_Reg32:$dst, (CONST_ADDRESS ADDRGA_CONST_OFFSET:$src))];
  let AsmString = "CONST_COPY";
  let hasSideEffects = 0;
  let isAsCheapAsAMove = 1;
  let Itinerary = NullALU;
}
} // end usesCustomInserter = 1, isCodeGenOnly = 1, isPseudo = 1, Namespace = "AMDGPU"

def TEX_VTX_CONSTBUF :
  InstR600ISA <(outs R600_Reg128:$dst), (ins MEMxi:$ptr, i32imm:$BUFFER_ID), "VTX_READ_eg $dst, $ptr",
      [(set v4i32:$dst, (CONST_ADDRESS ADDRGA_VAR_OFFSET:$ptr, (i32 imm:$BUFFER_ID)))]>,
  VTX_WORD1_GPR, VTX_WORD0_eg {

  let VC_INST = 0;
  let FETCH_TYPE = 2;
  let FETCH_WHOLE_QUAD = 0;
  let SRC_REL = 0;
  let SRC_SEL_X = 0;
  let DST_REL = 0;
  let USE_CONST_FIELDS = 0;
  let NUM_FORMAT_ALL = 2;
  let FORMAT_COMP_ALL = 1;
  let SRF_MODE_ALL = 1;
  let MEGA_FETCH_COUNT = 16;
  let DST_SEL_X        = 0;
  let DST_SEL_Y        = 1;
  let DST_SEL_Z        = 2;
  let DST_SEL_W        = 3;
  let DATA_FORMAT      = 35;

  let Inst{31-0} = Word0;
  let Inst{63-32} = Word1;

// LLVM can only encode 64-bit instructions, so these fields are manually
// encoded in R600CodeEmitter
//
// bits<16> OFFSET;
// bits<2>  ENDIAN_SWAP = 0;
// bits<1>  CONST_BUF_NO_STRIDE = 0;
// bits<1>  MEGA_FETCH = 0;
// bits<1>  ALT_CONST = 0;
// bits<2>  BUFFER_INDEX_MODE = 0;



// VTX_WORD2 (LLVM can only encode 64-bit instructions, so WORD2 encoding
// is done in R600CodeEmitter
//
// Inst{79-64} = OFFSET;
// Inst{81-80} = ENDIAN_SWAP;
// Inst{82}    = CONST_BUF_NO_STRIDE;
// Inst{83}    = MEGA_FETCH;
// Inst{84}    = ALT_CONST;
// Inst{86-85} = BUFFER_INDEX_MODE;
// Inst{95-86} = 0; Reserved

// VTX_WORD3 (Padding)
//
// Inst{127-96} = 0;
  let VTXInst = 1;
}

def TEX_VTX_TEXBUF:
  InstR600ISA <(outs R600_Reg128:$dst), (ins MEMxi:$ptr, i32imm:$BUFFER_ID), "TEX_VTX_EXPLICIT_READ $dst, $ptr",
      [(set v4f32:$dst, (int_R600_load_texbuf ADDRGA_VAR_OFFSET:$ptr, imm:$BUFFER_ID))]>,
VTX_WORD1_GPR, VTX_WORD0_eg {

let VC_INST = 0;
let FETCH_TYPE = 2;
let FETCH_WHOLE_QUAD = 0;
let SRC_REL = 0;
let SRC_SEL_X = 0;
let DST_REL = 0;
let USE_CONST_FIELDS = 1;
let NUM_FORMAT_ALL = 0;
let FORMAT_COMP_ALL = 0;
let SRF_MODE_ALL = 1;
let MEGA_FETCH_COUNT = 16;
let DST_SEL_X        = 0;
let DST_SEL_Y        = 1;
let DST_SEL_Z        = 2;
let DST_SEL_W        = 3;
let DATA_FORMAT      = 0;

let Inst{31-0} = Word0;
let Inst{63-32} = Word1;

// LLVM can only encode 64-bit instructions, so these fields are manually
// encoded in R600CodeEmitter
//
// bits<16> OFFSET;
// bits<2>  ENDIAN_SWAP = 0;
// bits<1>  CONST_BUF_NO_STRIDE = 0;
// bits<1>  MEGA_FETCH = 0;
// bits<1>  ALT_CONST = 0;
// bits<2>  BUFFER_INDEX_MODE = 0;



// VTX_WORD2 (LLVM can only encode 64-bit instructions, so WORD2 encoding
// is done in R600CodeEmitter
//
// Inst{79-64} = OFFSET;
// Inst{81-80} = ENDIAN_SWAP;
// Inst{82}    = CONST_BUF_NO_STRIDE;
// Inst{83}    = MEGA_FETCH;
// Inst{84}    = ALT_CONST;
// Inst{86-85} = BUFFER_INDEX_MODE;
// Inst{95-86} = 0; Reserved

// VTX_WORD3 (Padding)
//
// Inst{127-96} = 0;
  let VTXInst = 1;
}

//===---------------------------------------------------------------------===//
// Flow and Program control Instructions
//===---------------------------------------------------------------------===//
class ILFormat<dag outs, dag ins, string asmstr, list<dag> pattern>
: Instruction {

     let Namespace = "AMDGPU";
     dag OutOperandList = outs;
     dag InOperandList = ins;
     let Pattern = pattern;
     let AsmString = !strconcat(asmstr, "\n");
     let isPseudo = 1;
     let Itinerary = NullALU;
     bit hasIEEEFlag = 0;
     bit hasZeroOpFlag = 0;
     let mayLoad = 0;
     let mayStore = 0;
     let hasSideEffects = 0;
     let isCodeGenOnly = 1;
}

multiclass BranchConditional<SDNode Op, RegisterClass rci, RegisterClass rcf> {
    def _i32 : ILFormat<(outs),
  (ins brtarget:$target, rci:$src0),
        "; i32 Pseudo branch instruction",
  [(Op bb:$target, (i32 rci:$src0))]>;
    def _f32 : ILFormat<(outs),
  (ins brtarget:$target, rcf:$src0),
        "; f32 Pseudo branch instruction",
  [(Op bb:$target, (f32 rcf:$src0))]>;
}

// Only scalar types should generate flow control
multiclass BranchInstr<string name> {
  def _i32 : ILFormat<(outs), (ins R600_Reg32:$src),
      !strconcat(name, " $src"), []>;
  def _f32 : ILFormat<(outs), (ins R600_Reg32:$src),
      !strconcat(name, " $src"), []>;
}
// Only scalar types should generate flow control
multiclass BranchInstr2<string name> {
  def _i32 : ILFormat<(outs), (ins R600_Reg32:$src0, R600_Reg32:$src1),
      !strconcat(name, " $src0, $src1"), []>;
  def _f32 : ILFormat<(outs), (ins R600_Reg32:$src0, R600_Reg32:$src1),
      !strconcat(name, " $src0, $src1"), []>;
}

//===---------------------------------------------------------------------===//
// Custom Inserter for Branches and returns, this eventually will be a
// separate pass
//===---------------------------------------------------------------------===//
let isTerminator = 1, usesCustomInserter = 1, isBranch = 1, isBarrier = 1 in {
  def BRANCH : ILFormat<(outs), (ins brtarget:$target),
      "; Pseudo unconditional branch instruction",
      [(br bb:$target)]>;
  defm BRANCH_COND : BranchConditional<IL_brcond, R600_Reg32, R600_Reg32>;
}

//===---------------------------------------------------------------------===//
// Return instruction
//===---------------------------------------------------------------------===//
let isTerminator = 1, isReturn = 1, hasCtrlDep = 1,
    usesCustomInserter = 1 in {
  def RETURN          : ILFormat<(outs), (ins variable_ops),
      "RETURN", [(IL_retflag)]>;
}

//===----------------------------------------------------------------------===//
// Branch Instructions
//===----------------------------------------------------------------------===//

def IF_PREDICATE_SET  : ILFormat<(outs), (ins R600_Reg32:$src),
  "IF_PREDICATE_SET $src", []>;

let isTerminator=1 in {
  def BREAK       : ILFormat< (outs), (ins),
      "BREAK", []>;
  def CONTINUE    : ILFormat< (outs), (ins),
      "CONTINUE", []>;
  def DEFAULT     : ILFormat< (outs), (ins),
      "DEFAULT", []>;
  def ELSE        : ILFormat< (outs), (ins),
      "ELSE", []>;
  def ENDSWITCH   : ILFormat< (outs), (ins),
      "ENDSWITCH", []>;
  def ENDMAIN     : ILFormat< (outs), (ins),
      "ENDMAIN", []>;
  def END         : ILFormat< (outs), (ins),
      "END", []>;
  def ENDFUNC     : ILFormat< (outs), (ins),
      "ENDFUNC", []>;
  def ENDIF       : ILFormat< (outs), (ins),
      "ENDIF", []>;
  def WHILELOOP   : ILFormat< (outs), (ins),
      "WHILE", []>;
  def ENDLOOP     : ILFormat< (outs), (ins),
      "ENDLOOP", []>;
  def FUNC        : ILFormat< (outs), (ins),
      "FUNC", []>;
  def RETDYN      : ILFormat< (outs), (ins),
      "RET_DYN", []>;
  // This opcode has custom swizzle pattern encoded in Swizzle Encoder
  defm IF_LOGICALNZ  : BranchInstr<"IF_LOGICALNZ">;
  // This opcode has custom swizzle pattern encoded in Swizzle Encoder
  defm IF_LOGICALZ   : BranchInstr<"IF_LOGICALZ">;
  // This opcode has custom swizzle pattern encoded in Swizzle Encoder
  defm BREAK_LOGICALNZ : BranchInstr<"BREAK_LOGICALNZ">;
  // This opcode has custom swizzle pattern encoded in Swizzle Encoder
  defm BREAK_LOGICALZ : BranchInstr<"BREAK_LOGICALZ">;
  // This opcode has custom swizzle pattern encoded in Swizzle Encoder
  defm CONTINUE_LOGICALNZ : BranchInstr<"CONTINUE_LOGICALNZ">;
  // This opcode has custom swizzle pattern encoded in Swizzle Encoder
  defm CONTINUE_LOGICALZ : BranchInstr<"CONTINUE_LOGICALZ">;
  defm IFC         : BranchInstr2<"IFC">;
  defm BREAKC      : BranchInstr2<"BREAKC">;
  defm CONTINUEC   : BranchInstr2<"CONTINUEC">;
}

//===----------------------------------------------------------------------===//
// Indirect addressing pseudo instructions
//===----------------------------------------------------------------------===//

let isPseudo = 1 in {

class ExtractVertical <RegisterClass vec_rc> : InstR600 <
  (outs R600_Reg32:$dst),
  (ins vec_rc:$vec, R600_Reg32:$index), "",
  [],
  AnyALU
>;

let Constraints = "$dst = $vec" in {

class InsertVertical <RegisterClass vec_rc> : InstR600 <
  (outs vec_rc:$dst),
  (ins vec_rc:$vec, R600_Reg32:$value, R600_Reg32:$index), "",
  [],
  AnyALU
>;

} // End Constraints = "$dst = $vec"

} // End isPseudo = 1

def R600_EXTRACT_ELT_V2 : ExtractVertical <R600_Reg64Vertical>;
def R600_EXTRACT_ELT_V4 : ExtractVertical <R600_Reg128Vertical>;

def R600_INSERT_ELT_V2 : InsertVertical <R600_Reg64Vertical>;
def R600_INSERT_ELT_V4 : InsertVertical <R600_Reg128Vertical>;

class ExtractVerticalPat <Instruction inst, ValueType vec_ty,
                          ValueType scalar_ty> : Pat <
  (scalar_ty (extractelt vec_ty:$vec, i32:$index)),
  (inst $vec, $index)
>;

def : ExtractVerticalPat <R600_EXTRACT_ELT_V2, v2i32, i32>;
def : ExtractVerticalPat <R600_EXTRACT_ELT_V2, v2f32, f32>;
def : ExtractVerticalPat <R600_EXTRACT_ELT_V4, v4i32, i32>;
def : ExtractVerticalPat <R600_EXTRACT_ELT_V4, v4f32, f32>;

class InsertVerticalPat <Instruction inst, ValueType vec_ty,
                         ValueType scalar_ty> : Pat <
  (vec_ty (insertelt vec_ty:$vec, scalar_ty:$value, i32:$index)),
  (inst $vec, $value, $index)
>;

def : InsertVerticalPat <R600_INSERT_ELT_V2, v2i32, i32>;
def : InsertVerticalPat <R600_INSERT_ELT_V2, v2f32, f32>;
def : InsertVerticalPat <R600_INSERT_ELT_V4, v4i32, i32>;
def : InsertVerticalPat <R600_INSERT_ELT_V4, v4f32, f32>;

//===----------------------------------------------------------------------===//
// ISel Patterns
//===----------------------------------------------------------------------===//

// CND*_INT Patterns for f32 True / False values

class CND_INT_f32 <InstR600 cnd, CondCode cc> : Pat <
  (selectcc i32:$src0, 0, f32:$src1, f32:$src2, cc),
  (cnd $src0, $src1, $src2)
>;

def : CND_INT_f32 <CNDE_INT,  SETEQ>;
def : CND_INT_f32 <CNDGT_INT, SETGT>;
def : CND_INT_f32 <CNDGE_INT, SETGE>;

//CNDGE_INT extra pattern
def : Pat <
  (selectcc i32:$src0, -1, i32:$src1, i32:$src2, COND_SGT),
  (CNDGE_INT $src0, $src1, $src2)
>;

// KIL Patterns
def KILP : Pat <
  (int_AMDGPU_kilp),
  (MASK_WRITE (KILLGT (f32 ONE), (f32 ZERO)))
>;

def KIL : Pat <
  (int_AMDGPU_kill f32:$src0),
  (MASK_WRITE (KILLGT (f32 ZERO), $src0))
>;

def : Extract_Element <f32, v4f32, 0, sub0>;
def : Extract_Element <f32, v4f32, 1, sub1>;
def : Extract_Element <f32, v4f32, 2, sub2>;
def : Extract_Element <f32, v4f32, 3, sub3>;

def : Insert_Element <f32, v4f32, 0, sub0>;
def : Insert_Element <f32, v4f32, 1, sub1>;
def : Insert_Element <f32, v4f32, 2, sub2>;
def : Insert_Element <f32, v4f32, 3, sub3>;

def : Extract_Element <i32, v4i32, 0, sub0>;
def : Extract_Element <i32, v4i32, 1, sub1>;
def : Extract_Element <i32, v4i32, 2, sub2>;
def : Extract_Element <i32, v4i32, 3, sub3>;

def : Insert_Element <i32, v4i32, 0, sub0>;
def : Insert_Element <i32, v4i32, 1, sub1>;
def : Insert_Element <i32, v4i32, 2, sub2>;
def : Insert_Element <i32, v4i32, 3, sub3>;

def : Extract_Element <f32, v2f32, 0, sub0>;
def : Extract_Element <f32, v2f32, 1, sub1>;

def : Insert_Element <f32, v2f32, 0, sub0>;
def : Insert_Element <f32, v2f32, 1, sub1>;

def : Extract_Element <i32, v2i32, 0, sub0>;
def : Extract_Element <i32, v2i32, 1, sub1>;

def : Insert_Element <i32, v2i32, 0, sub0>;
def : Insert_Element <i32, v2i32, 1, sub1>;

// bitconvert patterns

def : BitConvert <i32, f32, R600_Reg32>;
def : BitConvert <f32, i32, R600_Reg32>;
def : BitConvert <v2f32, v2i32, R600_Reg64>;
def : BitConvert <v2i32, v2f32, R600_Reg64>;
def : BitConvert <v4f32, v4i32, R600_Reg128>;
def : BitConvert <v4i32, v4f32, R600_Reg128>;

// DWORDADDR pattern
def : DwordAddrPat  <i32, R600_Reg32>;

} // End isR600toCayman Predicate

let Predicates = [isR600] in {
// Intrinsic patterns
defm : Expand24IBitOps<MULLO_INT_r600, ADD_INT>;
defm : Expand24UBitOps<MULLO_UINT_r600, ADD_INT>;
} // End isR600

def getLDSNoRetOp : InstrMapping {
  let FilterClass = "R600_LDS_1A1D";
  let RowFields = ["BaseOp"];
  let ColFields = ["DisableEncoding"];
  let KeyCol = ["$dst"];
  let ValueCols = [[""""]];
}