Commits · 8e644c08ee6e921430e1a1f3e9a8b9dab6c0f8f9 · Roger Ferrer / llvm-epi-0.8 · GitLab

This project is archived. Its data is read-only.

Jan 30, 2017
- [InstCombine] fixed to propagate 'exact' on lshr · 8e644c08
  Sanjay Patel authored Jan 30, 2017
  8e644c08
- [InstCombine] enable lshr(shl X, C1), C2 folds for vectors with splat constants · 1196d7cd
  Sanjay Patel authored Jan 30, 2017
  1196d7cd
Jan 29, 2017
- [InstCombine] enable (X >>?,exact C1) << C2 --> X << (C2 - C1) for vectors with splats · 062adaab
  Sanjay Patel authored Jan 29, 2017
  062adaab
Jan 28, 2017
- [InstCombine] move icmp transforms that might be recognized as min/max and inf-loop (PR31751) · febcb9ce
  Sanjay Patel authored Jan 27, 2017
  febcb9ce
Jan 27, 2017
- [NVPTX] [InstCombine] Add llvm_unreachable to appease MSVC. · 25ebe2d7
  Justin Lebar authored Jan 27, 2017
  25ebe2d7
- [NVPTX] Fix use-after-stack-free bug in InstCombineCalls. · e3ac0fb9
  Justin Lebar authored Jan 27, 2017
  e3ac0fb9
- [NVPTX] Upgrade NVVM intrinsics in InstCombineCalls. · 698c31b8
  Justin Lebar authored Jan 27, 2017
  698c31b8
- Revert a couple of InstCombine/Guard checkins · 7516192a
  Sanjoy Das authored Jan 26, 2017
  7516192a
Jan 26, 2017
- [InstCombine] fold (X >>u C) << C --> X & (-1 << C) · 50753f02
  Sanjay Patel authored Jan 26, 2017
  50753f02
- [InstCombine] use m_APInt to allow (X << C) >>u C --> X & (-1 >>u C) with splat vectors · b0d96d32
  Sanjay Patel authored Jan 26, 2017
  b0d96d32
- [X86] Add demanded elts support for the inputs to pclmul intrinsic · b6122122
  Craig Topper authored Jan 26, 2017
  b6122122
Jan 25, 2017
- [InstCombine] Canonicalize guards for NOT OR condition · b85f7a5d
  Artur Pilipenko authored Jan 25, 2017
  b85f7a5d
- [InstCombine][SSE] Add support for PACKSS/PACKUS constant folding · 6f6b2791
  Simon Pilgrim authored Jan 25, 2017
  6f6b2791
- [InstCombine] Canonicalize guards for AND condition · 4df4c4a4
  Artur Pilipenko authored Jan 25, 2017
  4df4c4a4
- [InstCombine] Allow InstrCombine to remove one of adjacent guards if they are equivalent · e812ca00
  Artur Pilipenko authored Jan 25, 2017
  e812ca00
Jan 24, 2017
- Use InstCombine's builder in foldSelectCttzCtlz instead of creating a new one. · d90f5f66
  Amaury Sechet authored Jan 24, 2017
  d90f5f66
- Fix formating in foldSelectCttzCtlz. NFC · 5da456e6
  Amaury Sechet authored Jan 24, 2017
  5da456e6
- [InstCombine][X86] MULDQ/MULUDQ undef -> zero · 78f8630a
  Simon Pilgrim authored Jan 24, 2017
  78f8630a
- SimplifyLibCalls: Replace more unary libcalls with intrinsics · 954a624f
  Matt Arsenault authored Jan 23, 2017
  954a624f
Jan 23, 2017
- [InstCombine][X86] Add MULDQ/MULUDQ constant folding support · f6f3a361
  Simon Pilgrim authored Jan 23, 2017
  f6f3a361
- [InstCombine][X86] MULDQ/MULUDQ undef -> zero · bb13fdab
  Simon Pilgrim authored Jan 23, 2017
  bb13fdab
Jan 21, 2017
- [InstCombine] use m_APInt to allow ashr folds for vectors with splat constants · 478a83c9
  Sanjay Patel authored Jan 21, 2017
  478a83c9
Jan 20, 2017
- [InstCombine][X86] Add MULDQ/MULUDQ undef handling · a50a93fc
  Simon Pilgrim authored Jan 20, 2017
  a50a93fc
- [InstCombine][SSE] Add DemandedElts support for PACKSS/PACKUS instructions · 51b3b98e
  Simon Pilgrim authored Jan 20, 2017
  51b3b98e
Jan 19, 2017
- [InstCombine] Simplify gep (gep p, a), (b-a) · 2ef8c4e7
  Davide Italiano authored Jan 19, 2017
  2ef8c4e7
- [InstCombine] icmp Pred (shl nsw X, C1), C0 --> icmp Pred X, C0 >> C1 · 291c3d8f
  Sanjay Patel authored Jan 19, 2017
  291c3d8f
Jan 18, 2017
- [InstCombine] add an assert to make a shl+icmp transform assumption explicit; NFCI · ae23d65a
  Sanjay Patel authored Jan 18, 2017
  ae23d65a
- [InstCombine] remove a redundant check; NFCI · 589de5ea
  Sanjay Patel authored Jan 18, 2017
  589de5ea
- [InstCombine][AVX2] Add DemandedElts support for VPERMD/VPERMPS shuffles · fe2c0ed4
  Simon Pilgrim authored Jan 18, 2017
  fe2c0ed4
- [InstCombine] Remove unnecessary intrinsics demanded elts handling · a22c3a1c
  Simon Pilgrim authored Jan 18, 2017
  a22c3a1c
Jan 17, 2017
- [InstCombine] refactor foldICmpShlConstant(); NFCI · 14715b3c
  Sanjay Patel authored Jan 17, 2017
  14715b3c
- [InstCombine] Fold ((C1 OP zext(X)) & C2) -> zext((C1 OP X) & C2) · de55c606
  David Majnemer authored Jan 17, 2017
  de55c606
- [InstCombine] reduce indent; NFCI · 5424bd26
  Sanjay Patel authored Jan 17, 2017
  5424bd26
- [InstCombine][X86][AVX] Add DemandedElts support for VPERMILPD/VPERMILPS instructions · d4eb800b
  Simon Pilgrim authored Jan 17, 2017
  d4eb800b
- [InstCombine] Don't DSE across readnone functions that may throw · 679bc32c
  Sanjoy Das authored Jan 17, 2017
  679bc32c
- [InstCombine] Fold ((C1-zext(X)) & C2) -> zext((C1-X) & C2) · 36d382b7
  David Majnemer authored Jan 17, 2017
  36d382b7
- SimplifyLibCalls: Replace fabs libcalls with intrinsics · 7233344c
  Matt Arsenault authored Jan 17, 2017
  7233344c
Jan 16, 2017
- [InstCombine] use m_APInt instead of faking it · da5682af
  Sanjay Patel authored Jan 16, 2017
  da5682af
- [InstCombine] fix names in canEvaluateShiftedShift(); NFC · 65cce20c
  Sanjay Patel authored Jan 16, 2017
  65cce20c
- [InstCombine] use m_APInt to allow shift-shift folds for vectors with splat constants · ab8b32de
  Sanjay Patel authored Jan 16, 2017
  ab8b32de