Commits · e781620dee4415018bdf3769a97c2ec31a85ff68 · Roger Ferrer / llvm-epi · GitLab

Jan 18, 2022
- [LoopVectorize][AArch64] Use get.active.lane.mask intrinsic when SVE is enabled · e781620d
  David Sherwood authored Dec 06, 2021
  e781620d
Jan 17, 2022
- [LV] Add test coverage for reductions with odd interleave counts. · 524150fe
  Florian Hahn authored Jan 17, 2022
  524150fe
- [LV] Make test more robust by adding users of inductions. · 4a6f4754
  Florian Hahn authored Jan 15, 2022
  4a6f4754
- [RISCV] Add initial support for getRegUsageForType and getNumberOfRegisters · cc35161d
  Kito Cheng authored Jan 09, 2022
  cc35161d
Jan 16, 2022
- [LV] Restore metadata to disable runtime unrolling for epilogue loop. · 070d1034
  Florian Hahn authored Jan 16, 2022
  070d1034
Jan 15, 2022
- [IRBuilder] Migrate select-folding to value-based FoldSelect. · ba3198cf
  Florian Hahn authored Jan 15, 2022
  ba3198cf
Jan 14, 2022
- Recommit "[LV] Inline CreateSplatIV call for scalar VFs." · 42b34fac
  Florian Hahn authored Jan 14, 2022
  42b34fac
- [LV] Add test with an integer induction based on a ptr one. · 5e4966cb
  Florian Hahn authored Jan 14, 2022
  5e4966cb
- Revert "[LV] Inline CreateSplatIV call for scalar VFs (NFC)." · 073c27b5
  James Y Knight authored Jan 14, 2022
  073c27b5
Jan 13, 2022
- [LV] Extend check lines for pr34681.ll to cover foldable select. · 7b9f5cbf
  Florian Hahn authored Jan 13, 2022
  
  7b9f5cbf
- [VPlan] Make IV operand explicit for VPWidenCanonicalIVRecipe (NFC). · 3f2fb767
  Florian Hahn authored Jan 13, 2022
  3f2fb767
- [LV] Inline CreateSplatIV call for scalar VFs (NFC). · 7ce48be0
  Florian Hahn authored Jan 13, 2022
  7ce48be0
Jan 12, 2022
- [VPlan] Introduce and use BranchOnCount VPInstruction. · d4a8fc3a
  Florian Hahn authored Jan 12, 2022
  d4a8fc3a
- [LoopVectorize] Pass a vector type to isLegalMaskedGather/Scatter · 552eb372
  Rosie Sumpter authored Dec 06, 2021
  552eb372
- [IRBuilder] Migrate icmp-folding to value-based FoldICmp. · 138fcc5f
  Florian Hahn authored Jan 12, 2022
  138fcc5f
- [IRBuilder] Migrate add-folding to value-based FoldAdd. · 7e680613
  Florian Hahn authored Jan 12, 2022
  7e680613
Jan 11, 2022
- [IRBuilder] Introduce folder using inst-simplify, use for Or fold. · f0ef1ea6
  Florian Hahn authored Jan 11, 2022
  f0ef1ea6
Jan 10, 2022
- [LoopVectorize] Make VPWidenCanonicalIVRecipe::execute work for scalable vectors · b0922a9d
  David Sherwood authored Nov 03, 2021
  b0922a9d
- [SCEVExpander] Only create trunc when needed. · aecad582
  Florian Hahn authored Jan 10, 2022
  aecad582
- [LoopVectorize] Add support for tail folding using scalable vectors · e3c84fb9
  David Sherwood authored Oct 26, 2021
  e3c84fb9
Jan 09, 2022
- [SCEVExpander] Only check overflow if it is needed. · 7f1bf68d
  Florian Hahn authored Jan 06, 2022
  7f1bf68d
Jan 08, 2022
- [LV] Improve check lines in existing tests. · 3b7b1a75
  Florian Hahn authored Jan 08, 2022
  3b7b1a75
- [LV] Make tests more robust by removing undef. · daa5e263
  Florian Hahn authored Jan 07, 2022
  daa5e263
- [SCEVExpander] Skip creating <u 0 check, which is always false. · 9345ab3a
  Florian Hahn authored Jan 08, 2022
  9345ab3a
- [RISCV] Add a command line option to control the LMUL used by TTI's getRegisterBitWidth. · 042394b6
  Craig Topper authored Jan 07, 2022
  042394b6
Jan 07, 2022
- [AArch64] Update addo and subo costs · bc615e43
  David Green authored Jan 07, 2022
  bc615e43
- [SCEVExpand] Only create required predicate checks. · f395a4f8
  Florian Hahn authored Jan 07, 2022
  f395a4f8
Jan 06, 2022
- [SCEVExpand] Do not create redundant 'or false' for pred expansion. · 86d113a8
  Florian Hahn authored Jan 06, 2022
  86d113a8
Jan 05, 2022
- [LV] Remove what seems like stale code in collectElementTypesForWidening. · 95a93722
  Sander de Smalen authored Jan 04, 2022
  95a93722
- [VPlan] Add VPCanonicalIVPHIRecipe, partly retire createInductionVariable. · 65c4d619
  Florian Hahn authored Jan 05, 2022
  65c4d619
Jan 04, 2022
- [LoopVectorize][CostModel] Choose smaller VFs for in-loop reductions without loads/stores · 961f51fd
  Rosie Sumpter authored Nov 08, 2021
  961f51fd
Jan 02, 2022
- [VPlan] Don't consider VPWidenCanonicalIVRecipe phi-like. · b1a333f0
  Florian Hahn authored Jan 02, 2022
  b1a333f0
Dec 30, 2021
- [InstCombine] fold opposite shifts around an add · 0c6979b2
  Sanjay Patel authored Dec 30, 2021
  0c6979b2
- Revert "[InstCombine] fold opposite shifts around an add" · fd9cd340
  Sanjay Patel authored Dec 30, 2021
  fd9cd340
- [InstCombine] fold opposite shifts around an add · 2e3e0a5c
  Sanjay Patel authored Dec 30, 2021
  2e3e0a5c
Dec 23, 2021
- [RISCV] Disable interleaving scalar loops in the loop vectorizer. · a9486a40
  Craig Topper authored Dec 23, 2021
  a9486a40
Dec 22, 2021
- [VPlan] Create header & latch blocks for skeleton up front (NFC). · ede7c243
  Florian Hahn authored Dec 22, 2021
  ede7c243
Dec 20, 2021
- Fix buildbot failure caused by D115651 · 290ae657
  Sander de Smalen authored Dec 20, 2021
  290ae657
- [LV] Enable scalable vectorization by default for SVE cores. · b1ff20fd
  Sander de Smalen authored Dec 13, 2021
  b1ff20fd
- [VPlan] Add Debugloc to VPInstruction. · 5b362e4c
  Florian Hahn authored Dec 20, 2021
  5b362e4c