Commits

Branch selector

 main

User selector

RKSimon

Datepicker

All time

Commit History

Commits on May 30, 2025

• [X86] combineEXTRACT_SUBVECTOR - generalize extract_subvector(broadcast(x),c) fold with IsElementEquivalent (#141963)

  Show description for 37edd2c
  RKSimon

  authored
  37edd2c

  Copy full SHA for 37edd2c

Commits on May 28, 2025

• [X86] X86FixupVectorConstantsPass - use scheduler model to avoid regressions (#140028)

  Show description for d1592a9
  RKSimon

  authored
  d1592a9

  Copy full SHA for d1592a9

Commits on May 27, 2025

• [X86] combineTargetShuffle - fold (vzmovl (shift x, y)) -> (shift (vzmovl x), y) (#141579)

  Show description for 344f594
  RKSimon

  authored
  344f594

  Copy full SHA for 344f594

• [X86] Add test coverage for #141475

  RKSimon

  committed
  7462da1

  Copy full SHA for 7462da1

• [X86] combineTruncatedArithmetic - move more of fold inside combinei64TruncSrlConstant

  Show description for 6f3efd8
  RKSimon

  committed
  6f3efd8

  Copy full SHA for 6f3efd8

• [X86] combinei64TruncSrlConstant - sink EVT::getIntegerVT to ISD::ADD case only. NFC.

  Show description for 6da8f3b
  RKSimon

  committed
  6da8f3b

  Copy full SHA for 6da8f3b

Commits on May 25, 2025

• VPlanRecipes.cpp - fix "not all control paths return a value" MSVC warning

  RKSimon

  committed
  63eb004

  Copy full SHA for 63eb004

• OffloadBundle.h - fix "not all control paths return a value" MSVC warning

  RKSimon

  committed
  6680772

  Copy full SHA for 6680772

Commits on May 23, 2025

• [X86] vector-shuffle-combining-avx512bwvl.ll - regenerate test check

  RKSimon

  committed
  23a341e

  Copy full SHA for 23a341e

• [X86] lowerVECTOR_SHUFFLE - canonicalize zeros/ones/fp splat constants to ensure no undefs (#141214)

  Show description for 6e574a4
  RKSimon

  authored
  6e574a4

  Copy full SHA for 6e574a4

• [X86] IsElementEquivalent - add handling for ISD::BITCASTS from smaller vector elements (#139741)

  Show description for 34deb76
  RKSimon

  authored
  34deb76

  Copy full SHA for 34deb76

Commits on May 22, 2025

• [APInt] APInt::clearBitsSlowCase - fix cut+paste typo when merging lo/himasks (#141108)

  Show description for 23f0fbf
  RKSimon

  authored
  23f0fbf

  Copy full SHA for 23f0fbf

• [X86] lowerV4F64Shuffle - prefer BLEND before UNPCK shuffle matching (#141073)

  Show description for 9a77af3
  RKSimon

  authored
  9a77af3

  Copy full SHA for 9a77af3

• [X86] lowerShuffleAsSplitOrBlend - prefer splitting AVX1-only shuffles if the operands can be freely split (#141055)

  Show description for 03cc50f
  RKSimon

  authored
  03cc50f

  Copy full SHA for 03cc50f

• [X86] lowerShuffleAsSplitOrBlend - use isElementEquivalent to help identify hidden splat/broadcasts (#141035)

  Show description for b9a7098
  RKSimon

  authored
  b9a7098

  Copy full SHA for b9a7098

• [X86] icmp-abs-C-vec.ll - regenerate test checks for TERNLOG comments

  RKSimon

  committed
  c42c91c

  Copy full SHA for c42c91c

Commits on May 21, 2025

• [X86] combineINSERT_SUBVECTOR - use concatSubVectors instead of direct fold to X86ISD::SUBV_BROADCAST_LOAD (#140919)

  Show description for f1d03de
  RKSimon

  authored
  f1d03de

  Copy full SHA for f1d03de

• [X86] combineINSERT_SUBVECTOR - simplify aligned index assertion to avoid signed/unsigned warning. NFC.

  RKSimon

  committed
  32f8132

  Copy full SHA for 32f8132

• [X86] lowerV8F32Shuffle - use lowerShufflePairAsUNPCKAndPermute on AVX1 targets (#140881)

  Show description for 4091b5d
  RKSimon

  authored
  4091b5d

  Copy full SHA for 4091b5d

• [X86] combineINSERT_SUBVECTOR - generalise insert_subvector(x,extract(broadcast)) -> blend (#140516)

  Show description for dec214d
  RKSimon

  authored
  dec214d

  Copy full SHA for dec214d

Commits on May 20, 2025

• [X86] matchBinaryPermuteShuffle - match AVX512 "cross lane" SHLDQ/SRLDQ style patterns using VALIGN (#140538)

  Show description for 09fd8f0
  RKSimon

  authored
  09fd8f0

  Copy full SHA for 09fd8f0

• [X86] combineAdd - use SDPatternMatch to simplify "(add (zext (vXi1 X)), Y) -> (sub Y, (sext (vXi1 X)))" matching. (#140731)

  RKSimon

  authored
  621a5a9

  Copy full SHA for 621a5a9

• Fix MSVC "not all control paths return a value" warning. NFC.

  RKSimon

  committed
  4fa2c62

  Copy full SHA for 4fa2c62

• [X86] combineSCALAR_TO_VECTOR - fold (v4i32 (scalar_to_vector (i32 (bitcast (float))))) to MOVD/SS (#140704)

  Show description for 01d9c7f
  RKSimon

  authored
  01d9c7f

  Copy full SHA for 01d9c7f

• [X86] Add #140693 test coverage

  RKSimon

  committed
  4b54d1a

  Copy full SHA for 4b54d1a

• [X86] scalarize-fp.ll - add AVX512 test coverage

  RKSimon

  committed
  a17ff7f

  Copy full SHA for a17ff7f

• [X86] SimplifyDemandedVectorEltsForTargetNode - replace packed fcmp node with scalar fcmp node if only element0 is demanded (#140563)

  Show description for 57aa7fa
  RKSimon

  authored
  57aa7fa

  Copy full SHA for 57aa7fa

Commits on May 19, 2025

• [X86] Add test coverage for #140534

  RKSimon

  committed
  f3d36b1

  Copy full SHA for f3d36b1

• [X86] Add tests showing failure to combine AVX512 shuffles to match "cross lane" SHLDQ/SRLDQ patterns using VALIGN

  Show description for 52e1995
  RKSimon

  committed
  52e1995

  Copy full SHA for 52e1995

• [X86] IsElementEquivalent - add basic handling for X86ISD::SUBV_BROADCAST_LOAD (#139727)

  Show description for 23e8480
  RKSimon

  authored
  23e8480

  Copy full SHA for 23e8480

Commits on May 18, 2025

• [X86] Move X86ISD::FP_TO_SINT_SAT/FP_TO_UINT_SAT outside of the MEMORY_OPCODE enum block (#140272)

  Show description for ad31366
  RKSimon

  authored
  ad31366

  Copy full SHA for ad31366

• [AMDGPU] select-undef.ll - regenerate test checks

  Show description for 01bf2d7
  RKSimon

  committed
  01bf2d7

  Copy full SHA for 01bf2d7

Commits on May 16, 2025

• [X86] SimplifyDemandedVectorEltsForTargetNode - add X86ISD::MOVDDUP handling (#140237)

  Show description for 2b5222b
  RKSimon

  authored
  2b5222b

  Copy full SHA for 2b5222b

• [X86] combineGatherScatter - ensure index / pointer sizes match when converting index shift to scale (#140110)

  Show description for 882a4c5
  RKSimon

  authored
  882a4c5

  Copy full SHA for 882a4c5

• [X86] Add test coverage for #140234

  RKSimon

  committed
  b712590

  Copy full SHA for b712590

Pagination

Previous

Next