Commits · d07c731d86d15d52bc3b8be028393cb11a76374a · Roger Ferrer / llvm-epi · GitLab

Oct 31, 2016
- Improved cost model for FDIV and FSQRT, by Andrew Tischenko · d07c731d
  Alexey Bataev authored Oct 31, 2016
  d07c731d
- Add triple to test so it does not fail on windows. · 7c41f20a
  Manuel Klimek authored Oct 31, 2016
  7c41f20a
- Delete .s file that did not test anything, and check in test that works. · bab67d2a
  Manuel Klimek authored Oct 31, 2016
  bab67d2a
- [AVX-512] Add missing patterns for selecting masked vector extracts that started from shuffles. · d4e58070
  Craig Topper authored Oct 31, 2016
  d4e58070
Oct 30, 2016
- [DAG] x | x --> x · 339a51ac
  Sanjay Patel authored Oct 30, 2016
  339a51ac
- [DAG] x & x --> x · 13aee345
  Sanjay Patel authored Oct 30, 2016
  13aee345
- [x86] add tests for basic logic op folds · 8a5f9810
  Sanjay Patel authored Oct 30, 2016
  8a5f9810
- Revert r285517 due to build failures. · 06903d16
  Dorit Nuzman authored Oct 30, 2016
  06903d16
- [LoopVectorize] Make interleaved-accesses analysis less conservative about · 3c1c658f
  Dorit Nuzman authored Oct 30, 2016
  3c1c658f
- [ThinLTO] Correctly resolve linkonce when importing aliasee · 3bc8abdf
  Teresa Johnson authored Oct 30, 2016
  3bc8abdf
Oct 29, 2016
- [ValueTracking] recognize more variants of smin/smax · 36eeb6d6
  Sanjay Patel authored Oct 29, 2016
  36eeb6d6
- [x86] add tests for smin/smax matchSelPattern (D26091) · e9fa95e5
  Sanjay Patel authored Oct 29, 2016
  e9fa95e5
- [InstCombine] re-use bitcasted compare operands in selects (PR28001) · 978f827d
  Sanjay Patel authored Oct 29, 2016
  978f827d
- [DAGCombiner] (REAPPLIED) Add vector demanded elements support to computeKnownBits · 75a697a1
  Simon Pilgrim authored Oct 29, 2016
  75a697a1
- Fixed FMA + FNEG combine. · 519b4ccd
  Elena Demikhovsky authored Oct 29, 2016
  519b4ccd
- AMDGPU: Use 1/2pi inline imm on VI · c88ba36e
  Matt Arsenault authored Oct 29, 2016
  c88ba36e
- Do not print out Flags field twice. · c95b4644
  Rui Ueyama authored Oct 28, 2016
  c95b4644
- [DAGCombiner] Fix a crash visiting `AND` nodes. · 86168b23
  Davide Italiano authored Oct 28, 2016
  86168b23
- AMDGPU/SI: Don't use non-0 waitcnt values when waiting on Flat instructions · 6695ba04
  Tom Stellard authored Oct 28, 2016
  6695ba04
Oct 28, 2016
- Add missing lit.local.cfg to llvm/test/Transforms/CodeGenPrepare/NVPTX. · 1535a5e9
  Justin Lebar authored Oct 28, 2016
  1535a5e9
- AMDGPU: Add definitions for scalar store instructions · 7b647556
  Matt Arsenault authored Oct 28, 2016
  7b647556
- [NVPTX] Compute 'rem' using the result of 'div', if possible. · f0a80ba3
  Justin Lebar authored Oct 28, 2016
  f0a80ba3
- Don't leave unused divs/rems sitting around in BypassSlowDivision. · 0ede5fb1
  Justin Lebar authored Oct 28, 2016
  0ede5fb1
- Don't claim the udiv created in BypassSlowDivision is exact. · 468bf732
  Justin Lebar authored Oct 28, 2016
  468bf732
- AMDGPU: Change check prefix in test · b5f2bb1a
  Matt Arsenault authored Oct 28, 2016
  b5f2bb1a
- AMDGPU: Diagnose using too many SGPRs · 4eae3019
  Matt Arsenault authored Oct 28, 2016
  4eae3019
- Handle non-~0 lane masks on live-in registers in LivePhysRegs · 2717175c
  Krzysztof Parzyszek authored Oct 28, 2016
  2717175c
- SpeculativeExecution: Allow speculating more inst types · ef002834
  Matt Arsenault authored Oct 28, 2016
  ef002834
- AMDGPU: Fix using incorrect private resource with no allocation · 08906a3c
  Matt Arsenault authored Oct 28, 2016
  08906a3c
- Implement vector count leading/trailing bytes with zero lsb and vector parity · e28a0fc7
  Nemanja Ivanovic authored Oct 28, 2016
  e28a0fc7
- Make swift calling convention test specific to armv7 · 6200b2b6
  Arnold Schwaighofer authored Oct 28, 2016
  6200b2b6
- [x86] add tests for missed umin/umax · 03a585e8
  Sanjay Patel authored Oct 28, 2016
  03a585e8
- More swift calling convention tests · 7f4b31c0
  Arnold Schwaighofer authored Oct 28, 2016
  7f4b31c0
- [InstCombine] move/add tests for smin/smax folds · 19ace1d5
  Sanjay Patel authored Oct 28, 2016
  19ace1d5
- [Hexagon] Maintain kill flags through splitting in expand-condsets · 87a47be0
  Krzysztof Parzyszek authored Oct 28, 2016
  87a47be0
- [Loads] Fix crash in is isDereferenceableAndAlignedPointer() · 13068995
  Tom Stellard authored Oct 28, 2016
  13068995
- [LV] Correct misleading comments in test (NFC) · 9b675536
  Matthew Simpson authored Oct 28, 2016
  9b675536
- Revert "[DAGCombiner] Add vector demanded elements support to computeKnownBits" · 5cee232b
  Juergen Ributzka authored Oct 28, 2016
  5cee232b
- [Reassociate] Removing instructions mutates the IR. · 631cd27f
  Davide Italiano authored Oct 28, 2016
  631cd27f
- [ConstantFold] Get the correct vector type when folding a getelementptr. · 30665147
  Davide Italiano authored Oct 28, 2016
  30665147