aboutsummaryrefslogtreecommitdiffstats
path: root/lib/Target/X86/X86TargetTransformInfo.cpp
blob: 2b70fd0ecf8cd22416722b48aad9d405163c7582 (plain)
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
//===-- X86TargetTransformInfo.cpp - X86 specific TTI pass ----------------===//
//
//                     The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
/// \file
/// This file implements a TargetTransformInfo analysis pass specific to the
/// X86 target machine. It uses the target's detailed information to provide
/// more precise answers to certain TTI queries, while letting the target
/// independent and default TTI implementations handle the rest.
///
//===----------------------------------------------------------------------===//

#include "X86.h"
#include "X86TargetMachine.h"
#include "llvm/Analysis/TargetTransformInfo.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/Support/Debug.h"
#include "llvm/Target/CostTable.h"
#include "llvm/Target/TargetLowering.h"
using namespace llvm;

#define DEBUG_TYPE "x86tti"

// Declare the pass initialization routine locally as target-specific passes
// don't have a target-wide initialization entry point, and so we rely on the
// pass constructor initialization.
namespace llvm {
void initializeX86TTIPass(PassRegistry &);
}

namespace {

class X86TTI final : public ImmutablePass, public TargetTransformInfo {
  const X86Subtarget *ST;
  const X86TargetLowering *TLI;

  /// Estimate the overhead of scalarizing an instruction. Insert and Extract
  /// are set if the result needs to be inserted and/or extracted from vectors.
  unsigned getScalarizationOverhead(Type *Ty, bool Insert, bool Extract) const;

public:
  X86TTI() : ImmutablePass(ID), ST(nullptr), TLI(nullptr) {
    llvm_unreachable("This pass cannot be directly constructed");
  }

  X86TTI(const X86TargetMachine *TM)
      : ImmutablePass(ID), ST(TM->getSubtargetImpl()),
        TLI(TM->getSubtargetImpl()->getTargetLowering()) {
    initializeX86TTIPass(*PassRegistry::getPassRegistry());
  }

  void initializePass() override {
    pushTTIStack(this);
  }

  void getAnalysisUsage(AnalysisUsage &AU) const override {
    TargetTransformInfo::getAnalysisUsage(AU);
  }

  /// Pass identification.
  static char ID;

  /// Provide necessary pointer adjustments for the two base classes.
  void *getAdjustedAnalysisPointer(const void *ID) override {
    if (ID == &TargetTransformInfo::ID)
      return (TargetTransformInfo*)this;
    return this;
  }

  /// \name Scalar TTI Implementations
  /// @{
  PopcntSupportKind getPopcntSupport(unsigned TyWidth) const override;

  /// @}

  /// \name Vector TTI Implementations
  /// @{

  unsigned getNumberOfRegisters(bool Vector) const override;
  unsigned getRegisterBitWidth(bool Vector) const override;
  unsigned getMaxInterleaveFactor() const override;
  unsigned getArithmeticInstrCost(unsigned Opcode, Type *Ty, OperandValueKind,
                                  OperandValueKind, OperandValueProperties,
                                  OperandValueProperties) const override;
  unsigned getShuffleCost(ShuffleKind Kind, Type *Tp,
                          int Index, Type *SubTp) const override;
  unsigned getCastInstrCost(unsigned Opcode, Type *Dst,
                            Type *Src) const override;
  unsigned getCmpSelInstrCost(unsigned Opcode, Type *ValTy,
                              Type *CondTy) const override;
  unsigned getVectorInstrCost(unsigned Opcode, Type *Val,
                              unsigned Index) const override;
  unsigned getMemoryOpCost(unsigned Opcode, Type *Src, unsigned Alignment,
                           unsigned AddressSpace) const override;

  unsigned getAddressComputationCost(Type *PtrTy,
                                     bool IsComplex) const override;

  unsigned getReductionCost(unsigned Opcode, Type *Ty,
                            bool IsPairwiseForm) const override;

  unsigned getIntImmCost(int64_t) const;

  unsigned getIntImmCost(const APInt &Imm, Type *Ty) const override;

  unsigned getIntImmCost(unsigned Opcode, unsigned Idx, const APInt &Imm,
                         Type *Ty) const override;
  unsigned getIntImmCost(Intrinsic::ID IID, unsigned Idx, const APInt &Imm,
                         Type *Ty) const override;

  /// @}
};

} // end anonymous namespace

INITIALIZE_AG_PASS(X86TTI, TargetTransformInfo, "x86tti",
                   "X86 Target Transform Info", true, true, false)
char X86TTI::ID = 0;

ImmutablePass *
llvm::createX86TargetTransformInfoPass(const X86TargetMachine *TM) {
  return new X86TTI(TM);
}


//===----------------------------------------------------------------------===//
//
// X86 cost model.
//
//===----------------------------------------------------------------------===//

X86TTI::PopcntSupportKind X86TTI::getPopcntSupport(unsigned TyWidth) const {
  assert(isPowerOf2_32(TyWidth) && "Ty width must be power of 2");
  // TODO: Currently the __builtin_popcount() implementation using SSE3
  //   instructions is inefficient. Once the problem is fixed, we should
  //   call ST->hasSSE3() instead of ST->hasPOPCNT().
  return ST->hasPOPCNT() ? PSK_FastHardware : PSK_Software;
}

unsigned X86TTI::getNumberOfRegisters(bool Vector) const {
  if (Vector && !ST->hasSSE1())
    return 0;

  if (ST->is64Bit()) {
    if (Vector && ST->hasAVX512())
      return 32;
    return 16;
  }
  return 8;
}

unsigned X86TTI::getRegisterBitWidth(bool Vector) const {
  if (Vector) {
    if (ST->hasAVX512()) return 512;
    if (ST->hasAVX()) return 256;
    if (ST->hasSSE1()) return 128;
    return 0;
  }

  if (ST->is64Bit())
    return 64;
  return 32;

}

unsigned X86TTI::getMaxInterleaveFactor() const {
  if (ST->isAtom())
    return 1;

  // Sandybridge and Haswell have multiple execution ports and pipelined
  // vector units.
  if (ST->hasAVX())
    return 4;

  return 2;
}

unsigned X86TTI::getArithmeticInstrCost(
    unsigned Opcode, Type *Ty, OperandValueKind Op1Info,
    OperandValueKind Op2Info, OperandValueProperties Opd1PropInfo,
    OperandValueProperties Opd2PropInfo) const {
  // Legalize the type.
  std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Ty);

  int ISD = TLI->InstructionOpcodeToISD(Opcode);
  assert(ISD && "Invalid opcode");

  if (ISD == ISD::SDIV &&
      Op2Info == TargetTransformInfo::OK_UniformConstantValue &&
      Opd2PropInfo == TargetTransformInfo::OP_PowerOf2) {
    // On X86, vector signed division by constants power-of-two are
    // normally expanded to the sequence SRA + SRL + ADD + SRA.
    // The OperandValue properties many not be same as that of previous
    // operation;conservatively assume OP_None.
    unsigned Cost =
        2 * getArithmeticInstrCost(Instruction::AShr, Ty, Op1Info, Op2Info,
                                   TargetTransformInfo::OP_None,
                                   TargetTransformInfo::OP_None);
    Cost += getArithmeticInstrCost(Instruction::LShr, Ty, Op1Info, Op2Info,
                                   TargetTransformInfo::OP_None,
                                   TargetTransformInfo::OP_None);
    Cost += getArithmeticInstrCost(Instruction::Add, Ty, Op1Info, Op2Info,
                                   TargetTransformInfo::OP_None,
                                   TargetTransformInfo::OP_None);

    return Cost;
  }

  static const CostTblEntry<MVT::SimpleValueType>
  AVX2UniformConstCostTable[] = {
    { ISD::SDIV, MVT::v16i16,  6 }, // vpmulhw sequence
    { ISD::UDIV, MVT::v16i16,  6 }, // vpmulhuw sequence
    { ISD::SDIV, MVT::v8i32,  15 }, // vpmuldq sequence
    { ISD::UDIV, MVT::v8i32,  15 }, // vpmuludq sequence
  };

  if (Op2Info == TargetTransformInfo::OK_UniformConstantValue &&
      ST->hasAVX2()) {
    int Idx = CostTableLookup(AVX2UniformConstCostTable, ISD, LT.second);
    if (Idx != -1)
      return LT.first * AVX2UniformConstCostTable[Idx].Cost;
  }

  static const CostTblEntry<MVT::SimpleValueType> AVX512CostTable[] = {
    { ISD::SHL,     MVT::v16i32,    1 },
    { ISD::SRL,     MVT::v16i32,    1 },
    { ISD::SRA,     MVT::v16i32,    1 },
    { ISD::SHL,     MVT::v8i64,    1 },
    { ISD::SRL,     MVT::v8i64,    1 },
    { ISD::SRA,     MVT::v8i64,    1 },
  };

  static const CostTblEntry<MVT::SimpleValueType> AVX2CostTable[] = {
    // Shifts on v4i64/v8i32 on AVX2 is legal even though we declare to
    // customize them to detect the cases where shift amount is a scalar one.
    { ISD::SHL,     MVT::v4i32,    1 },
    { ISD::SRL,     MVT::v4i32,    1 },
    { ISD::SRA,     MVT::v4i32,    1 },
    { ISD::SHL,     MVT::v8i32,    1 },
    { ISD::SRL,     MVT::v8i32,    1 },
    { ISD::SRA,     MVT::v8i32,    1 },
    { ISD::SHL,     MVT::v2i64,    1 },
    { ISD::SRL,     MVT::v2i64,    1 },
    { ISD::SHL,     MVT::v4i64,    1 },
    { ISD::SRL,     MVT::v4i64,    1 },

    { ISD::SHL,  MVT::v32i8,  42 }, // cmpeqb sequence.
    { ISD::SHL,  MVT::v16i16,  16*10 }, // Scalarized.

    { ISD::SRL,  MVT::v32i8,  32*10 }, // Scalarized.
    { ISD::SRL,  MVT::v16i16,  8*10 }, // Scalarized.

    { ISD::SRA,  MVT::v32i8,  32*10 }, // Scalarized.
    { ISD::SRA,  MVT::v16i16,  16*10 }, // Scalarized.
    { ISD::SRA,  MVT::v4i64,  4*10 }, // Scalarized.

    // Vectorizing division is a bad idea. See the SSE2 table for more comments.
    { ISD::SDIV,  MVT::v32i8,  32*20 },
    { ISD::SDIV,  MVT::v16i16, 16*20 },
    { ISD::SDIV,  MVT::v8i32,  8*20 },
    { ISD::SDIV,  MVT::v4i64,  4*20 },
    { ISD::UDIV,  MVT::v32i8,  32*20 },
    { ISD::UDIV,  MVT::v16i16, 16*20 },
    { ISD::UDIV,  MVT::v8i32,  8*20 },
    { ISD::UDIV,  MVT::v4i64,  4*20 },
  };

  if (ST->hasAVX512()) {
    int Idx = CostTableLookup(AVX512CostTable, ISD, LT.second);
    if (Idx != -1)
      return LT.first * AVX512CostTable[Idx].Cost;
  }
  // Look for AVX2 lowering tricks.
  if (ST->hasAVX2()) {
    if (ISD == ISD::SHL && LT.second == MVT::v16i16 &&
        (Op2Info == TargetTransformInfo::OK_UniformConstantValue ||
         Op2Info == TargetTransformInfo::OK_NonUniformConstantValue))
      // On AVX2, a packed v16i16 shift left by a constant build_vector
      // is lowered into a vector multiply (vpmullw).
      return LT.first;

    int Idx = CostTableLookup(AVX2CostTable, ISD, LT.second);
    if (Idx != -1)
      return LT.first * AVX2CostTable[Idx].Cost;
  }

  static const CostTblEntry<MVT::SimpleValueType>
  SSE2UniformConstCostTable[] = {
    // We don't correctly identify costs of casts because they are marked as
    // custom.
    // Constant splats are cheaper for the following instructions.
    { ISD::SHL,  MVT::v16i8,  1 }, // psllw.
    { ISD::SHL,  MVT::v8i16,  1 }, // psllw.
    { ISD::SHL,  MVT::v4i32,  1 }, // pslld
    { ISD::SHL,  MVT::v2i64,  1 }, // psllq.

    { ISD::SRL,  MVT::v16i8,  1 }, // psrlw.
    { ISD::SRL,  MVT::v8i16,  1 }, // psrlw.
    { ISD::SRL,  MVT::v4i32,  1 }, // psrld.
    { ISD::SRL,  MVT::v2i64,  1 }, // psrlq.

    { ISD::SRA,  MVT::v16i8,  4 }, // psrlw, pand, pxor, psubb.
    { ISD::SRA,  MVT::v8i16,  1 }, // psraw.
    { ISD::SRA,  MVT::v4i32,  1 }, // psrad.

    { ISD::SDIV, MVT::v8i16,  6 }, // pmulhw sequence
    { ISD::UDIV, MVT::v8i16,  6 }, // pmulhuw sequence
    { ISD::SDIV, MVT::v4i32, 19 }, // pmuludq sequence
    { ISD::UDIV, MVT::v4i32, 15 }, // pmuludq sequence
  };

  if (Op2Info == TargetTransformInfo::OK_UniformConstantValue &&
      ST->hasSSE2()) {
    // pmuldq sequence.
    if (ISD == ISD::SDIV && LT.second == MVT::v4i32 && ST->hasSSE41())
      return LT.first * 15;

    int Idx = CostTableLookup(SSE2UniformConstCostTable, ISD, LT.second);
    if (Idx != -1)
      return LT.first * SSE2UniformConstCostTable[Idx].Cost;
  }

  if (ISD == ISD::SHL &&
      Op2Info == TargetTransformInfo::OK_NonUniformConstantValue) {
    EVT VT = LT.second;
    if ((VT == MVT::v8i16 && ST->hasSSE2()) ||
        (VT == MVT::v4i32 && ST->hasSSE41()))
      // Vector shift left by non uniform constant can be lowered
      // into vector multiply (pmullw/pmulld).
      return LT.first;
    if (VT == MVT::v4i32 && ST->hasSSE2())
      // A vector shift left by non uniform constant is converted
      // into a vector multiply; the new multiply is eventually
      // lowered into a sequence of shuffles and 2 x pmuludq.
      ISD = ISD::MUL;
  }

  static const CostTblEntry<MVT::SimpleValueType> SSE2CostTable[] = {
    // We don't correctly identify costs of casts because they are marked as
    // custom.
    // For some cases, where the shift amount is a scalar we would be able
    // to generate better code. Unfortunately, when this is the case the value
    // (the splat) will get hoisted out of the loop, thereby making it invisible
    // to ISel. The cost model must return worst case assumptions because it is
    // used for vectorization and we don't want to make vectorized code worse
    // than scalar code.
    { ISD::SHL,  MVT::v16i8,  30 }, // cmpeqb sequence.
    { ISD::SHL,  MVT::v8i16,  8*10 }, // Scalarized.
    { ISD::SHL,  MVT::v4i32,  2*5 }, // We optimized this using mul.
    { ISD::SHL,  MVT::v2i64,  2*10 }, // Scalarized.
    { ISD::SHL,  MVT::v4i64,  4*10 }, // Scalarized. 

    { ISD::SRL,  MVT::v16i8,  16*10 }, // Scalarized.
    { ISD::SRL,  MVT::v8i16,  8*10 }, // Scalarized.
    { ISD::SRL,  MVT::v4i32,  4*10 }, // Scalarized.
    { ISD::SRL,  MVT::v2i64,  2*10 }, // Scalarized.

    { ISD::SRA,  MVT::v16i8,  16*10 }, // Scalarized.
    { ISD::SRA,  MVT::v8i16,  8*10 }, // Scalarized.
    { ISD::SRA,  MVT::v4i32,  4*10 }, // Scalarized.
    { ISD::SRA,  MVT::v2i64,  2*10 }, // Scalarized.

    // It is not a good idea to vectorize division. We have to scalarize it and
    // in the process we will often end up having to spilling regular
    // registers. The overhead of division is going to dominate most kernels
    // anyways so try hard to prevent vectorization of division - it is
    // generally a bad idea. Assume somewhat arbitrarily that we have to be able
    // to hide "20 cycles" for each lane.
    { ISD::SDIV,  MVT::v16i8,  16*20 },
    { ISD::SDIV,  MVT::v8i16,  8*20 },
    { ISD::SDIV,  MVT::v4i32,  4*20 },
    { ISD::SDIV,  MVT::v2i64,  2*20 },
    { ISD::UDIV,  MVT::v16i8,  16*20 },
    { ISD::UDIV,  MVT::v8i16,  8*20 },
    { ISD::UDIV,  MVT::v4i32,  4*20 },
    { ISD::UDIV,  MVT::v2i64,  2*20 },
  };

  if (ST->hasSSE2()) {
    int Idx = CostTableLookup(SSE2CostTable, ISD, LT.second);
    if (Idx != -1)
      return LT.first * SSE2CostTable[Idx].Cost;
  }

  static const CostTblEntry<MVT::SimpleValueType> AVX1CostTable[] = {
    // We don't have to scalarize unsupported ops. We can issue two half-sized
    // operations and we only need to extract the upper YMM half.
    // Two ops + 1 extract + 1 insert = 4.
    { ISD::MUL,     MVT::v16i16,   4 },
    { ISD::MUL,     MVT::v8i32,    4 },
    { ISD::SUB,     MVT::v8i32,    4 },
    { ISD::ADD,     MVT::v8i32,    4 },
    { ISD::SUB,     MVT::v4i64,    4 },
    { ISD::ADD,     MVT::v4i64,    4 },
    // A v4i64 multiply is custom lowered as two split v2i64 vectors that then
    // are lowered as a series of long multiplies(3), shifts(4) and adds(2)
    // Because we believe v4i64 to be a legal type, we must also include the
    // split factor of two in the cost table. Therefore, the cost here is 18
    // instead of 9.
    { ISD::MUL,     MVT::v4i64,    18 },
  };

  // Look for AVX1 lowering tricks.
  if (ST->hasAVX() && !ST->hasAVX2()) {
    EVT VT = LT.second;

    // v16i16 and v8i32 shifts by non-uniform constants are lowered into a
    // sequence of extract + two vector multiply + insert.
    if (ISD == ISD::SHL && (VT == MVT::v8i32 || VT == MVT::v16i16) &&
        Op2Info == TargetTransformInfo::OK_NonUniformConstantValue)
      ISD = ISD::MUL;

    int Idx = CostTableLookup(AVX1CostTable, ISD, VT);
    if (Idx != -1)
      return LT.first * AVX1CostTable[Idx].Cost;
  }

  // Custom lowering of vectors.
  static const CostTblEntry<MVT::SimpleValueType> CustomLowered[] = {
    // A v2i64/v4i64 and multiply is custom lowered as a series of long
    // multiplies(3), shifts(4) and adds(2).
    { ISD::MUL,     MVT::v2i64,    9 },
    { ISD::MUL,     MVT::v4i64,    9 },
  };
  int Idx = CostTableLookup(CustomLowered, ISD, LT.second);
  if (Idx != -1)
    return LT.first * CustomLowered[Idx].Cost;

  // Special lowering of v4i32 mul on sse2, sse3: Lower v4i32 mul as 2x shuffle,
  // 2x pmuludq, 2x shuffle.
  if (ISD == ISD::MUL && LT.second == MVT::v4i32 && ST->hasSSE2() &&
      !ST->hasSSE41())
    return LT.first * 6;

  // Fallback to the default implementation.
  return TargetTransformInfo::getArithmeticInstrCost(Opcode, Ty, Op1Info,
                                                     Op2Info);
}

unsigned X86TTI::getShuffleCost(ShuffleKind Kind, Type *Tp, int Index,
                                Type *SubTp) const {
  // We only estimate the cost of reverse and alternate shuffles.
  if (Kind != SK_Reverse && Kind != SK_Alternate)
    return TargetTransformInfo::getShuffleCost(Kind, Tp, Index, SubTp);

  if (Kind == SK_Reverse) {
    std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Tp);
    unsigned Cost = 1;
    if (LT.second.getSizeInBits() > 128)
      Cost = 3; // Extract + insert + copy.

    // Multiple by the number of parts.
    return Cost * LT.first;
  }

  if (Kind == SK_Alternate) {
    // 64-bit packed float vectors (v2f32) are widened to type v4f32.
    // 64-bit packed integer vectors (v2i32) are promoted to type v2i64.
    std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Tp);

    // The backend knows how to generate a single VEX.256 version of
    // instruction VPBLENDW if the target supports AVX2.
    if (ST->hasAVX2() && LT.second == MVT::v16i16)
      return LT.first;

    static const CostTblEntry<MVT::SimpleValueType> AVXAltShuffleTbl[] = {
      {ISD::VECTOR_SHUFFLE, MVT::v4i64, 1},  // vblendpd
      {ISD::VECTOR_SHUFFLE, MVT::v4f64, 1},  // vblendpd

      {ISD::VECTOR_SHUFFLE, MVT::v8i32, 1},  // vblendps
      {ISD::VECTOR_SHUFFLE, MVT::v8f32, 1},  // vblendps

      // This shuffle is custom lowered into a sequence of:
      //  2x  vextractf128 , 2x vpblendw , 1x vinsertf128
      {ISD::VECTOR_SHUFFLE, MVT::v16i16, 5},

      // This shuffle is custom lowered into a long sequence of:
      //  2x vextractf128 , 4x vpshufb , 2x vpor ,  1x vinsertf128
      {ISD::VECTOR_SHUFFLE, MVT::v32i8, 9}
    };

    if (ST->hasAVX()) {
      int Idx = CostTableLookup(AVXAltShuffleTbl, ISD::VECTOR_SHUFFLE, LT.second);
      if (Idx != -1)
        return LT.first * AVXAltShuffleTbl[Idx].Cost;
    }

    static const CostTblEntry<MVT::SimpleValueType> SSE41AltShuffleTbl[] = {
      // These are lowered into movsd.
      {ISD::VECTOR_SHUFFLE, MVT::v2i64, 1},
      {ISD::VECTOR_SHUFFLE, MVT::v2f64, 1},

      // packed float vectors with four elements are lowered into BLENDI dag
      // nodes. A v4i32/v4f32 BLENDI generates a single 'blendps'/'blendpd'.
      {ISD::VECTOR_SHUFFLE, MVT::v4i32, 1},
      {ISD::VECTOR_SHUFFLE, MVT::v4f32, 1},

      // This shuffle generates a single pshufw.
      {ISD::VECTOR_SHUFFLE, MVT::v8i16, 1},

      // There is no instruction that matches a v16i8 alternate shuffle.
      // The backend will expand it into the sequence 'pshufb + pshufb + or'.
      {ISD::VECTOR_SHUFFLE, MVT::v16i8, 3}
    };

    if (ST->hasSSE41()) {
      int Idx = CostTableLookup(SSE41AltShuffleTbl, ISD::VECTOR_SHUFFLE, LT.second);
      if (Idx != -1)
        return LT.first * SSE41AltShuffleTbl[Idx].Cost;
    }

    static const CostTblEntry<MVT::SimpleValueType> SSSE3AltShuffleTbl[] = {
      {ISD::VECTOR_SHUFFLE, MVT::v2i64, 1},  // movsd
      {ISD::VECTOR_SHUFFLE, MVT::v2f64, 1},  // movsd

      // SSE3 doesn't have 'blendps'. The following shuffles are expanded into
      // the sequence 'shufps + pshufd'
      {ISD::VECTOR_SHUFFLE, MVT::v4i32, 2},
      {ISD::VECTOR_SHUFFLE, MVT::v4f32, 2},

      {ISD::VECTOR_SHUFFLE, MVT::v8i16, 3}, // pshufb + pshufb + or
      {ISD::VECTOR_SHUFFLE, MVT::v16i8, 3}  // pshufb + pshufb + or
    };
 
    if (ST->hasSSSE3()) {
      int Idx = CostTableLookup(SSSE3AltShuffleTbl, ISD::VECTOR_SHUFFLE, LT.second);
      if (Idx != -1)
        return LT.first * SSSE3AltShuffleTbl[Idx].Cost;
    }

    static const CostTblEntry<MVT::SimpleValueType> SSEAltShuffleTbl[] = {
      {ISD::VECTOR_SHUFFLE, MVT::v2i64, 1},  // movsd
      {ISD::VECTOR_SHUFFLE, MVT::v2f64, 1},  // movsd

      {ISD::VECTOR_SHUFFLE, MVT::v4i32, 2}, // shufps + pshufd
      {ISD::VECTOR_SHUFFLE, MVT::v4f32, 2}, // shufps + pshufd
 
      // This is expanded into a long sequence of four extract + four insert.
      {ISD::VECTOR_SHUFFLE, MVT::v8i16, 8}, // 4 x pextrw + 4 pinsrw.

      // 8 x (pinsrw + pextrw + and + movb + movzb + or)
      {ISD::VECTOR_SHUFFLE, MVT::v16i8, 48}
    };

    // Fall-back (SSE3 and SSE2). 
    int Idx = CostTableLookup(SSEAltShuffleTbl, ISD::VECTOR_SHUFFLE, LT.second);
    if (Idx != -1)
      return LT.first * SSEAltShuffleTbl[Idx].Cost;
    return TargetTransformInfo::getShuffleCost(Kind, Tp, Index, SubTp);
  }

  return TargetTransformInfo::getShuffleCost(Kind, Tp, Index, SubTp);
}

unsigned X86TTI::getCastInstrCost(unsigned Opcode, Type *Dst, Type *Src) const {
  int ISD = TLI->InstructionOpcodeToISD(Opcode);
  assert(ISD && "Invalid opcode");

  std::pair<unsigned, MVT> LTSrc = TLI->getTypeLegalizationCost(Src);
  std::pair<unsigned, MVT> LTDest = TLI->getTypeLegalizationCost(Dst);

  static const TypeConversionCostTblEntry<MVT::SimpleValueType>
  SSE2ConvTbl[] = {
    // These are somewhat magic numbers justified by looking at the output of
    // Intel's IACA, running some kernels and making sure when we take
    // legalization into account the throughput will be overestimated.
    { ISD::UINT_TO_FP, MVT::v2f64, MVT::v2i64, 2*10 },
    { ISD::UINT_TO_FP, MVT::v2f64, MVT::v4i32, 4*10 },
    { ISD::UINT_TO_FP, MVT::v2f64, MVT::v8i16, 8*10 },
    { ISD::UINT_TO_FP, MVT::v2f64, MVT::v16i8, 16*10 },
    { ISD::SINT_TO_FP, MVT::v2f64, MVT::v2i64, 2*10 },
    { ISD::SINT_TO_FP, MVT::v2f64, MVT::v4i32, 4*10 },
    { ISD::SINT_TO_FP, MVT::v2f64, MVT::v8i16, 8*10 },
    { ISD::SINT_TO_FP, MVT::v2f64, MVT::v16i8, 16*10 },
    // There are faster sequences for float conversions.
    { ISD::UINT_TO_FP, MVT::v4f32, MVT::v2i64, 15 },
    { ISD::UINT_TO_FP, MVT::v4f32, MVT::v4i32, 8 },
    { ISD::UINT_TO_FP, MVT::v4f32, MVT::v8i16, 15 },
    { ISD::UINT_TO_FP, MVT::v4f32, MVT::v16i8, 8 },
    { ISD::SINT_TO_FP, MVT::v4f32, MVT::v2i64, 15 },
    { ISD::SINT_TO_FP, MVT::v4f32, MVT::v4i32, 15 },
    { ISD::SINT_TO_FP, MVT::v4f32, MVT::v8i16, 15 },
    { ISD::SINT_TO_FP, MVT::v4f32, MVT::v16i8, 8 },
  };

  if (ST->hasSSE2() && !ST->hasAVX()) {
    int Idx =
        ConvertCostTableLookup(SSE2ConvTbl, ISD, LTDest.second, LTSrc.second);
    if (Idx != -1)
      return LTSrc.first * SSE2ConvTbl[Idx].Cost;
  }

  static const TypeConversionCostTblEntry<MVT::SimpleValueType>
  AVX512ConversionTbl[] = {
    { ISD::FP_EXTEND, MVT::v8f64,   MVT::v8f32,  1 },
    { ISD::FP_EXTEND, MVT::v8f64,   MVT::v16f32, 3 },
    { ISD::FP_ROUND,  MVT::v8f32,   MVT::v8f64,  1 },
    { ISD::FP_ROUND,  MVT::v16f32,  MVT::v8f64,  3 },

    { ISD::TRUNCATE,  MVT::v16i8,   MVT::v16i32, 1 },
    { ISD::TRUNCATE,  MVT::v16i16,  MVT::v16i32, 1 },
    { ISD::TRUNCATE,  MVT::v8i16,   MVT::v8i64,  1 },
    { ISD::TRUNCATE,  MVT::v8i32,   MVT::v8i64,  1 },
    { ISD::TRUNCATE,  MVT::v16i32,  MVT::v8i64,  4 },

    // v16i1 -> v16i32 - load + broadcast
    { ISD::SIGN_EXTEND, MVT::v16i32, MVT::v16i1,  2 },
    { ISD::ZERO_EXTEND, MVT::v16i32, MVT::v16i1,  2 },

    { ISD::SIGN_EXTEND, MVT::v16i32, MVT::v16i8,  1 },
    { ISD::ZERO_EXTEND, MVT::v16i32, MVT::v16i8,  1 },
    { ISD::SIGN_EXTEND, MVT::v16i32, MVT::v16i16, 1 },
    { ISD::ZERO_EXTEND, MVT::v16i32, MVT::v16i16, 1 },
    { ISD::SIGN_EXTEND, MVT::v8i64,  MVT::v16i32, 3 },
    { ISD::ZERO_EXTEND, MVT::v8i64,  MVT::v16i32, 3 },

    { ISD::SINT_TO_FP,  MVT::v16f32, MVT::v16i1,  3 },
    { ISD::SINT_TO_FP,  MVT::v16f32, MVT::v16i8,  2 },
    { ISD::SINT_TO_FP,  MVT::v16f32, MVT::v16i16, 2 },
    { ISD::SINT_TO_FP,  MVT::v16f32, MVT::v16i32, 1 },
    { ISD::SINT_TO_FP,  MVT::v8f64,  MVT::v8i1,   4 },
    { ISD::SINT_TO_FP,  MVT::v8f64,  MVT::v8i16,  2 },
    { ISD::SINT_TO_FP,  MVT::v8f64,  MVT::v8i32,  1 },
  };

  if (ST->hasAVX512()) {
    int Idx = ConvertCostTableLookup(AVX512ConversionTbl, ISD, LTDest.second,
                                     LTSrc.second);
    if (Idx != -1)
      return AVX512ConversionTbl[Idx].Cost;
  }
  EVT SrcTy = TLI->getValueType(Src);
  EVT DstTy = TLI->getValueType(Dst);

  // The function getSimpleVT only handles simple value types.
  if (!SrcTy.isSimple() || !DstTy.isSimple())
    return TargetTransformInfo::getCastInstrCost(Opcode, Dst, Src);

  static const TypeConversionCostTblEntry<MVT::SimpleValueType>
  AVX2ConversionTbl[] = {
    { ISD::SIGN_EXTEND, MVT::v16i16, MVT::v16i8,  1 },
    { ISD::ZERO_EXTEND, MVT::v16i16, MVT::v16i8,  1 },
    { ISD::SIGN_EXTEND, MVT::v8i32,  MVT::v8i1,   3 },
    { ISD::ZERO_EXTEND, MVT::v8i32,  MVT::v8i1,   3 },
    { ISD::SIGN_EXTEND, MVT::v8i32,  MVT::v8i8,   3 },
    { ISD::ZERO_EXTEND, MVT::v8i32,  MVT::v8i8,   3 },
    { ISD::SIGN_EXTEND, MVT::v8i32,  MVT::v8i16,  1 },
    { ISD::ZERO_EXTEND, MVT::v8i32,  MVT::v8i16,  1 },
    { ISD::SIGN_EXTEND, MVT::v4i64,  MVT::v4i1,   3 },
    { ISD::ZERO_EXTEND, MVT::v4i64,  MVT::v4i1,   3 },
    { ISD::SIGN_EXTEND, MVT::v4i64,  MVT::v4i8,   3 },
    { ISD::ZERO_EXTEND, MVT::v4i64,  MVT::v4i8,   3 },
    { ISD::SIGN_EXTEND, MVT::v4i64,  MVT::v4i16,  3 },
    { ISD::ZERO_EXTEND, MVT::v4i64,  MVT::v4i16,  3 },
    { ISD::SIGN_EXTEND, MVT::v4i64,  MVT::v4i32,  1 },
    { ISD::ZERO_EXTEND, MVT::v4i64,  MVT::v4i32,  1 },

    { ISD::TRUNCATE,    MVT::v4i8,   MVT::v4i64,  2 },
    { ISD::TRUNCATE,    MVT::v4i16,  MVT::v4i64,  2 },
    { ISD::TRUNCATE,    MVT::v4i32,  MVT::v4i64,  2 },
    { ISD::TRUNCATE,    MVT::v8i8,   MVT::v8i32,  2 },
    { ISD::TRUNCATE,    MVT::v8i16,  MVT::v8i32,  2 },
    { ISD::TRUNCATE,    MVT::v8i32,  MVT::v8i64,  4 },

    { ISD::FP_EXTEND,   MVT::v8f64,  MVT::v8f32,  3 },
    { ISD::FP_ROUND,    MVT::v8f32,  MVT::v8f64,  3 },

    { ISD::UINT_TO_FP,  MVT::v8f32,  MVT::v8i32,  8 },
  };

  static const TypeConversionCostTblEntry<MVT::SimpleValueType>
  AVXConversionTbl[] = {
    { ISD::SIGN_EXTEND, MVT::v16i16, MVT::v16i8, 4 },
    { ISD::ZERO_EXTEND, MVT::v16i16, MVT::v16i8, 4 },
    { ISD::SIGN_EXTEND, MVT::v8i32,  MVT::v8i1,  7 },
    { ISD::ZERO_EXTEND, MVT::v8i32,  MVT::v8i1,  4 },
    { ISD::SIGN_EXTEND, MVT::v8i32,  MVT::v8i8,  7 },
    { ISD::ZERO_EXTEND, MVT::v8i32,  MVT::v8i8,  4 },
    { ISD::SIGN_EXTEND, MVT::v8i32,  MVT::v8i16, 4 },
    { ISD::ZERO_EXTEND, MVT::v8i32,  MVT::v8i16, 4 },
    { ISD::SIGN_EXTEND, MVT::v4i64,  MVT::v4i1,  6 },
    { ISD::ZERO_EXTEND, MVT::v4i64,  MVT::v4i1,  4 },
    { ISD::SIGN_EXTEND, MVT::v4i64,  MVT::v4i8,  6 },
    { ISD::ZERO_EXTEND, MVT::v4i64,  MVT::v4i8,  4 },
    { ISD::SIGN_EXTEND, MVT::v4i64,  MVT::v4i16, 6 },
    { ISD::ZERO_EXTEND, MVT::v4i64,  MVT::v4i16, 3 },
    { ISD::SIGN_EXTEND, MVT::v4i64,  MVT::v4i32, 4 },
    { ISD::ZERO_EXTEND, MVT::v4i64,  MVT::v4i32, 4 },

    { ISD::TRUNCATE,    MVT::v4i8,  MVT::v4i64,  4 },
    { ISD::TRUNCATE,    MVT::v4i16, MVT::v4i64,  4 },
    { ISD::TRUNCATE,    MVT::v4i32, MVT::v4i64,  4 },
    { ISD::TRUNCATE,    MVT::v8i8,  MVT::v8i32,  4 },
    { ISD::TRUNCATE,    MVT::v8i16, MVT::v8i32,  5 },
    { ISD::TRUNCATE,    MVT::v16i8, MVT::v16i16, 4 },
    { ISD::TRUNCATE,    MVT::v8i32, MVT::v8i64,  9 },

    { ISD::SINT_TO_FP,  MVT::v8f32, MVT::v8i1,  8 },
    { ISD::SINT_TO_FP,  MVT::v8f32, MVT::v8i8,  8 },
    { ISD::SINT_TO_FP,  MVT::v8f32, MVT::v8i16, 5 },
    { ISD::SINT_TO_FP,  MVT::v8f32, MVT::v8i32, 1 },
    { ISD::SINT_TO_FP,  MVT::v4f32, MVT::v4i1,  3 },
    { ISD::SINT_TO_FP,  MVT::v4f32, MVT::v4i8,  3 },
    { ISD::SINT_TO_FP,  MVT::v4f32, MVT::v4i16, 3 },
    { ISD::SINT_TO_FP,  MVT::v4f32, MVT::v4i32, 1 },
    { ISD::SINT_TO_FP,  MVT::v4f64, MVT::v4i1,  3 },
    { ISD::SINT_TO_FP,  MVT::v4f64, MVT::v4i8,  3 },
    { ISD::SINT_TO_FP,  MVT::v4f64, MVT::v4i16, 3 },
    { ISD::SINT_TO_FP,  MVT::v4f64, MVT::v4i32, 1 },

    { ISD::UINT_TO_FP,  MVT::v8f32, MVT::v8i1,  6 },
    { ISD::UINT_TO_FP,  MVT::v8f32, MVT::v8i8,  5 },
    { ISD::UINT_TO_FP,  MVT::v8f32, MVT::v8i16, 5 },
    { ISD::UINT_TO_FP,  MVT::v8f32, MVT::v8i32, 9 },
    { ISD::UINT_TO_FP,  MVT::v4f32, MVT::v4i1,  7 },
    { ISD::UINT_TO_FP,  MVT::v4f32, MVT::v4i8,  2 },
    { ISD::UINT_TO_FP,  MVT::v4f32, MVT::v4i16, 2 },
    { ISD::UINT_TO_FP,  MVT::v4f32, MVT::v4i32, 6 },
    { ISD::UINT_TO_FP,  MVT::v4f64, MVT::v4i1,  7 },
    { ISD::UINT_TO_FP,  MVT::v4f64, MVT::v4i8,  2 },
    { ISD::UINT_TO_FP,  MVT::v4f64, MVT::v4i16, 2 },
    { ISD::UINT_TO_FP,  MVT::v4f64, MVT::v4i32, 6 },
    // The generic code to compute the scalar overhead is currently broken.
    // Workaround this limitation by estimating the scalarization overhead
    // here. We have roughly 10 instructions per scalar element.
    // Multiply that by the vector width.
    // FIXME: remove that when PR19268 is fixed.
    { ISD::UINT_TO_FP,  MVT::v2f64, MVT::v2i64, 2*10 },
    { ISD::UINT_TO_FP,  MVT::v4f64, MVT::v4i64, 4*10 },

    { ISD::FP_TO_SINT,  MVT::v8i8,  MVT::v8f32, 7 },
    { ISD::FP_TO_SINT,  MVT::v4i8,  MVT::v4f32, 1 },
    // This node is expanded into scalarized operations but BasicTTI is overly
    // optimistic estimating its cost.  It computes 3 per element (one
    // vector-extract, one scalar conversion and one vector-insert).  The
    // problem is that the inserts form a read-modify-write chain so latency
    // should be factored in too.  Inflating the cost per element by 1.
    { ISD::FP_TO_UINT,  MVT::v8i32, MVT::v8f32, 8*4 },
    { ISD::FP_TO_UINT,  MVT::v4i32, MVT::v4f64, 4*4 },
  };

  if (ST->hasAVX2()) {
    int Idx = ConvertCostTableLookup(AVX2ConversionTbl, ISD,
                                     DstTy.getSimpleVT(), SrcTy.getSimpleVT());
    if (Idx != -1)
      return AVX2ConversionTbl[Idx].Cost;
  }

  if (ST->hasAVX()) {
    int Idx = ConvertCostTableLookup(AVXConversionTbl, ISD, DstTy.getSimpleVT(),
                                     SrcTy.getSimpleVT());
    if (Idx != -1)
      return AVXConversionTbl[Idx].Cost;
  }

  return TargetTransformInfo::getCastInstrCost(Opcode, Dst, Src);
}

unsigned X86TTI::getCmpSelInstrCost(unsigned Opcode, Type *ValTy,
                                    Type *CondTy) const {
  // Legalize the type.
  std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(ValTy);

  MVT MTy = LT.second;

  int ISD = TLI->InstructionOpcodeToISD(Opcode);
  assert(ISD && "Invalid opcode");

  static const CostTblEntry<MVT::SimpleValueType> SSE42CostTbl[] = {
    { ISD::SETCC,   MVT::v2f64,   1 },
    { ISD::SETCC,   MVT::v4f32,   1 },
    { ISD::SETCC,   MVT::v2i64,   1 },
    { ISD::SETCC,   MVT::v4i32,   1 },
    { ISD::SETCC,   MVT::v8i16,   1 },
    { ISD::SETCC,   MVT::v16i8,   1 },
  };

  static const CostTblEntry<MVT::SimpleValueType> AVX1CostTbl[] = {
    { ISD::SETCC,   MVT::v4f64,   1 },
    { ISD::SETCC,   MVT::v8f32,   1 },
    // AVX1 does not support 8-wide integer compare.
    { ISD::SETCC,   MVT::v4i64,   4 },
    { ISD::SETCC,   MVT::v8i32,   4 },
    { ISD::SETCC,   MVT::v16i16,  4 },
    { ISD::SETCC,   MVT::v32i8,   4 },
  };

  static const CostTblEntry<MVT::SimpleValueType> AVX2CostTbl[] = {
    { ISD::SETCC,   MVT::v4i64,   1 },
    { ISD::SETCC,   MVT::v8i32,   1 },
    { ISD::SETCC,   MVT::v16i16,  1 },
    { ISD::SETCC,   MVT::v32i8,   1 },
  };

  static const CostTblEntry<MVT::SimpleValueType> AVX512CostTbl[] = {
    { ISD::SETCC,   MVT::v8i64,   1 },
    { ISD::SETCC,   MVT::v16i32,  1 },
    { ISD::SETCC,   MVT::v8f64,   1 },
    { ISD::SETCC,   MVT::v16f32,  1 },
  };

  if (ST->hasAVX512()) {
    int Idx = CostTableLookup(AVX512CostTbl, ISD, MTy);
    if (Idx != -1)
      return LT.first * AVX512CostTbl[Idx].Cost;
  }

  if (ST->hasAVX2()) {
    int Idx = CostTableLookup(AVX2CostTbl, ISD, MTy);
    if (Idx != -1)
      return LT.first * AVX2CostTbl[Idx].Cost;
  }

  if (ST->hasAVX()) {
    int Idx = CostTableLookup(AVX1CostTbl, ISD, MTy);
    if (Idx != -1)
      return LT.first * AVX1CostTbl[Idx].Cost;
  }

  if (ST->hasSSE42()) {
    int Idx = CostTableLookup(SSE42CostTbl, ISD, MTy);
    if (Idx != -1)
      return LT.first * SSE42CostTbl[Idx].Cost;
  }

  return TargetTransformInfo::getCmpSelInstrCost(Opcode, ValTy, CondTy);
}

unsigned X86TTI::getVectorInstrCost(unsigned Opcode, Type *Val,
                                    unsigned Index) const {
  assert(Val->isVectorTy() && "This must be a vector type");

  if (Index != -1U) {
    // Legalize the type.
    std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Val);

    // This type is legalized to a scalar type.
    if (!LT.second.isVector())
      return 0;

    // The type may be split. Normalize the index to the new type.
    unsigned Width = LT.second.getVectorNumElements();
    Index = Index % Width;

    // Floating point scalars are already located in index #0.
    if (Val->getScalarType()->isFloatingPointTy() && Index == 0)
      return 0;
  }

  return TargetTransformInfo::getVectorInstrCost(Opcode, Val, Index);
}

unsigned X86TTI::getScalarizationOverhead(Type *Ty, bool Insert,
                                            bool Extract) const {
  assert (Ty->isVectorTy() && "Can only scalarize vectors");
  unsigned Cost = 0;

  for (int i = 0, e = Ty->getVectorNumElements(); i < e; ++i) {
    if (Insert)
      Cost += TopTTI->getVectorInstrCost(Instruction::InsertElement, Ty, i);
    if (Extract)
      Cost += TopTTI->getVectorInstrCost(Instruction::ExtractElement, Ty, i);
  }

  return Cost;
}

unsigned X86TTI::getMemoryOpCost(unsigned Opcode, Type *Src, unsigned Alignment,
                                 unsigned AddressSpace) const {
  // Handle non-power-of-two vectors such as <3 x float>
  if (VectorType *VTy = dyn_cast<VectorType>(Src)) {
    unsigned NumElem = VTy->getVectorNumElements();

    // Handle a few common cases:
    // <3 x float>
    if (NumElem == 3 && VTy->getScalarSizeInBits() == 32)
      // Cost = 64 bit store + extract + 32 bit store.
      return 3;

    // <3 x double>
    if (NumElem == 3 && VTy->getScalarSizeInBits() == 64)
      // Cost = 128 bit store + unpack + 64 bit store.
      return 3;

    // Assume that all other non-power-of-two numbers are scalarized.
    if (!isPowerOf2_32(NumElem)) {
      unsigned Cost = TargetTransformInfo::getMemoryOpCost(Opcode,
                                                           VTy->getScalarType(),
                                                           Alignment,
                                                           AddressSpace);
      unsigned SplitCost = getScalarizationOverhead(Src,
                                                    Opcode == Instruction::Load,
                                                    Opcode==Instruction::Store);
      return NumElem * Cost + SplitCost;
    }
  }

  // Legalize the type.
  std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Src);
  assert((Opcode == Instruction::Load || Opcode == Instruction::Store) &&
         "Invalid Opcode");

  // Each load/store unit costs 1.
  unsigned Cost = LT.first * 1;

  // On Sandybridge 256bit load/stores are double pumped
  // (but not on Haswell).
  if (LT.second.getSizeInBits() > 128 && !ST->hasAVX2())
    Cost*=2;

  return Cost;
}

unsigned X86TTI::getAddressComputationCost(Type *Ty, bool IsComplex) const {
  // Address computations in vectorized code with non-consecutive addresses will
  // likely result in more instructions compared to scalar code where the
  // computation can more often be merged into the index mode. The resulting
  // extra micro-ops can significantly decrease throughput.
  unsigned NumVectorInstToHideOverhead = 10;

  if (Ty->isVectorTy() && IsComplex)
    return NumVectorInstToHideOverhead;

  return TargetTransformInfo::getAddressComputationCost(Ty, IsComplex);
}

unsigned X86TTI::getReductionCost(unsigned Opcode, Type *ValTy,
                                  bool IsPairwise) const {
  
  std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(ValTy);
  
  MVT MTy = LT.second;
  
  int ISD = TLI->InstructionOpcodeToISD(Opcode);
  assert(ISD && "Invalid opcode");
 
  // We use the Intel Architecture Code Analyzer(IACA) to measure the throughput 
  // and make it as the cost. 
 
  static const CostTblEntry<MVT::SimpleValueType> SSE42CostTblPairWise[] = {
    { ISD::FADD,  MVT::v2f64,   2 },
    { ISD::FADD,  MVT::v4f32,   4 },
    { ISD::ADD,   MVT::v2i64,   2 },      // The data reported by the IACA tool is "1.6".
    { ISD::ADD,   MVT::v4i32,   3 },      // The data reported by the IACA tool is "3.5".
    { ISD::ADD,   MVT::v8i16,   5 },
  };
 
  static const CostTblEntry<MVT::SimpleValueType> AVX1CostTblPairWise[] = {
    { ISD::FADD,  MVT::v4f32,   4 },
    { ISD::FADD,  MVT::v4f64,   5 },
    { ISD::FADD,  MVT::v8f32,   7 },
    { ISD::ADD,   MVT::v2i64,   1 },      // The data reported by the IACA tool is "1.5".
    { ISD::ADD,   MVT::v4i32,   3 },      // The data reported by the IACA tool is "3.5".
    { ISD::ADD,   MVT::v4i64,   5 },      // The data reported by the IACA tool is "4.8".
    { ISD::ADD,   MVT::v8i16,   5 },
    { ISD::ADD,   MVT::v8i32,   5 },
  };

  static const CostTblEntry<MVT::SimpleValueType> SSE42CostTblNoPairWise[] = {
    { ISD::FADD,  MVT::v2f64,   2 },
    { ISD::FADD,  MVT::v4f32,   4 },
    { ISD::ADD,   MVT::v2i64,   2 },      // The data reported by the IACA tool is "1.6".
    { ISD::ADD,   MVT::v4i32,   3 },      // The data reported by the IACA tool is "3.3".
    { ISD::ADD,   MVT::v8i16,   4 },      // The data reported by the IACA tool is "4.3".
  };
  
  static const CostTblEntry<MVT::SimpleValueType> AVX1CostTblNoPairWise[] = {
    { ISD::FADD,  MVT::v4f32,   3 },
    { ISD::FADD,  MVT::v4f64,   3 },
    { ISD::FADD,  MVT::v8f32,   4 },
    { ISD::ADD,   MVT::v2i64,   1 },      // The data reported by the IACA tool is "1.5".
    { ISD::ADD,   MVT::v4i32,   3 },      // The data reported by the IACA tool is "2.8".
    { ISD::ADD,   MVT::v4i64,   3 },
    { ISD::ADD,   MVT::v8i16,   4 },
    { ISD::ADD,   MVT::v8i32,   5 },
  };
  
  if (IsPairwise) {
    if (ST->hasAVX()) {
      int Idx = CostTableLookup(AVX1CostTblPairWise, ISD, MTy);
      if (Idx != -1)
        return LT.first * AVX1CostTblPairWise[Idx].Cost;
    }
  
    if (ST->hasSSE42()) {
      int Idx = CostTableLookup(SSE42CostTblPairWise, ISD, MTy);
      if (Idx != -1)
        return LT.first * SSE42CostTblPairWise[Idx].Cost;
    }
  } else {
    if (ST->hasAVX()) {
      int Idx = CostTableLookup(AVX1CostTblNoPairWise, ISD, MTy);
      if (Idx != -1)
        return LT.first * AVX1CostTblNoPairWise[Idx].Cost;
    }
    
    if (ST->hasSSE42()) {
      int Idx = CostTableLookup(SSE42CostTblNoPairWise, ISD, MTy);
      if (Idx != -1)
        return LT.first * SSE42CostTblNoPairWise[Idx].Cost;
    }
  }

  return TargetTransformInfo::getReductionCost(Opcode, ValTy, IsPairwise);
}

/// \brief Calculate the cost of materializing a 64-bit value. This helper
/// method might only calculate a fraction of a larger immediate. Therefore it
/// is valid to return a cost of ZERO.
unsigned X86TTI::getIntImmCost(int64_t Val) const {
  if (Val == 0)
    return TCC_Free;

  if (isInt<32>(Val))
    return TCC_Basic;

  return 2 * TCC_Basic;
}

unsigned X86TTI::getIntImmCost(const APInt &Imm, Type *Ty) const {
  assert(Ty->isIntegerTy());

  unsigned BitSize = Ty->getPrimitiveSizeInBits();
  if (BitSize == 0)
    return ~0U;

  // Never hoist constants larger than 128bit, because this might lead to
  // incorrect code generation or assertions in codegen.
  // Fixme: Create a cost model for types larger than i128 once the codegen
  // issues have been fixed.
  if (BitSize > 128)
    return TCC_Free;

  if (Imm == 0)
    return TCC_Free;

  // Sign-extend all constants to a multiple of 64-bit.
  APInt ImmVal = Imm;
  if (BitSize & 0x3f)
    ImmVal = Imm.sext((BitSize + 63) & ~0x3fU);

  // Split the constant into 64-bit chunks and calculate the cost for each
  // chunk.
  unsigned Cost = 0;
  for (unsigned ShiftVal = 0; ShiftVal < BitSize; ShiftVal += 64) {
    APInt Tmp = ImmVal.ashr(ShiftVal).sextOrTrunc(64);
    int64_t Val = Tmp.getSExtValue();
    Cost += getIntImmCost(Val);
  }
  // We need at least one instruction to materialze the constant.
  return std::max(1U, Cost);
}

unsigned X86TTI::getIntImmCost(unsigned Opcode, unsigned Idx, const APInt &Imm,
                               Type *Ty) const {
  assert(Ty->isIntegerTy());

  unsigned BitSize = Ty->getPrimitiveSizeInBits();
  // There is no cost model for constants with a bit size of 0. Return TCC_Free
  // here, so that constant hoisting will ignore this constant.
  if (BitSize == 0)
    return TCC_Free;

  unsigned ImmIdx = ~0U;
  switch (Opcode) {
  default: return TCC_Free;
  case Instruction::GetElementPtr:
    // Always hoist the base address of a GetElementPtr. This prevents the
    // creation of new constants for every base constant that gets constant
    // folded with the offset.
    if (Idx == 0)
      return 2 * TCC_Basic;
    return TCC_Free;
  case Instruction::Store:
    ImmIdx = 0;
    break;
  case Instruction::Add:
  case Instruction::Sub:
  case Instruction::Mul:
  case Instruction::UDiv:
  case Instruction::SDiv:
  case Instruction::URem:
  case Instruction::SRem:
  case Instruction::And:
  case Instruction::Or:
  case Instruction::Xor:
  case Instruction::ICmp:
    ImmIdx = 1;
    break;
  // Always return TCC_Free for the shift value of a shift instruction.
  case Instruction::Shl:
  case Instruction::LShr:
  case Instruction::AShr:
    if (Idx == 1)
      return TCC_Free;
    break;
  case Instruction::Trunc:
  case Instruction::ZExt:
  case Instruction::SExt:
  case Instruction::IntToPtr:
  case Instruction::PtrToInt:
  case Instruction::BitCast:
  case Instruction::PHI:
  case Instruction::Call:
  case Instruction::Select:
  case Instruction::Ret:
  case Instruction::Load:
    break;
  }

  if (Idx == ImmIdx) {
    unsigned NumConstants = (BitSize + 63) / 64;
    unsigned Cost = X86TTI::getIntImmCost(Imm, Ty);
    return (Cost <= NumConstants * TCC_Basic)
      ? static_cast<unsigned>(TCC_Free)
      : Cost;
  }

  return X86TTI::getIntImmCost(Imm, Ty);
}

unsigned X86TTI::getIntImmCost(Intrinsic::ID IID, unsigned Idx,
                               const APInt &Imm, Type *Ty) const {
  assert(Ty->isIntegerTy());

  unsigned BitSize = Ty->getPrimitiveSizeInBits();
  // There is no cost model for constants with a bit size of 0. Return TCC_Free
  // here, so that constant hoisting will ignore this constant.
  if (BitSize == 0)
    return TCC_Free;

  switch (IID) {
  default: return TCC_Free;
  case Intrinsic::sadd_with_overflow:
  case Intrinsic::uadd_with_overflow:
  case Intrinsic::ssub_with_overflow:
  case Intrinsic::usub_with_overflow:
  case Intrinsic::smul_with_overflow:
  case Intrinsic::umul_with_overflow:
    if ((Idx == 1) && Imm.getBitWidth() <= 64 && isInt<32>(Imm.getSExtValue()))
      return TCC_Free;
    break;
  case Intrinsic::experimental_stackmap:
    if ((Idx < 2) || (Imm.getBitWidth() <= 64 && isInt<64>(Imm.getSExtValue())))
      return TCC_Free;
    break;
  case Intrinsic::experimental_patchpoint_void:
  case Intrinsic::experimental_patchpoint_i64:
    if ((Idx < 4) || (Imm.getBitWidth() <= 64 && isInt<64>(Imm.getSExtValue())))
      return TCC_Free;
    break;
  }
  return X86TTI::getIntImmCost(Imm, Ty);
}