Commits · eb8f1e27d9b68c4bfbbfbcf3741434722e5f27ef · Roger Ferrer / llvm-epi · GitLab

Jun 25, 2017
- [AVX2] [TTI CostModel] Add cost of interleaved loads/stores for AVX2 · e0e0f1dd
  Dorit Nuzman authored Jun 25, 2017
  e0e0f1dd
Jun 20, 2017
- [CostModel][X86] Add scalar arithmetic cost tests · 68204b83
  Simon Pilgrim authored Jun 20, 2017
  68204b83
- [CostModel][X86] Declare costs variables based on type · 36c17935
  Simon Pilgrim authored Jun 20, 2017
  36c17935
May 24, 2017
- Revert r291254: [AArch64] Reduce vector insert/extract cost for Falkor · 6349380f
  Matthew Simpson authored May 24, 2017
  6349380f
May 19, 2017
- Fix line-endings. · c74e7f0a
  Simon Pilgrim authored May 19, 2017
  c74e7f0a
May 18, 2017
- [X86][AVX512] Add 512-bit vector ctpop costs + tests · 6bba6068
  Simon Pilgrim authored May 18, 2017
  6bba6068
May 17, 2017
- [X86][AVX512] Add 512-bit vector ctlz costs + tests · 23ef2672
  Simon Pilgrim authored May 17, 2017
  23ef2672
- [X86][AVX512] Add 512-bit vector cttz costs + tests · d0365967
  Simon Pilgrim authored May 17, 2017
  d0365967
- [X86] Split ctpop/ctlz/cttz cost tests · 91b46c99
  Simon Pilgrim authored May 17, 2017
  91b46c99
- [X86][AVX512] Add 512-bit vector bitreverse costs + tests · a9a92a1a
  Simon Pilgrim authored May 17, 2017
  a9a92a1a
- [SystemZ] Modelling of costs of divisions with a constant power of 2. · 8722ade7
  Jonas Paulsson authored May 17, 2017
  8722ade7
May 14, 2017
- [X86][AVX1] Account for cost of extract/insert of 256-bit shifts · d0ef9d8e
  Simon Pilgrim authored May 14, 2017
  d0ef9d8e
- [X86][AVX2] Fix costs for v4i64 ashr by splat · f96b4ab9
  Simon Pilgrim authored May 14, 2017
  f96b4ab9
- [X86][AVX1] Account for cost of extract/insert of 256-bit shifts by splat · de4467b1
  Simon Pilgrim authored May 14, 2017
  de4467b1
- [X86][AVX1] Account for cost of extract/insert of 256-bit SDIV/UDIV by mul sequences · d3f0d03c
  Simon Pilgrim authored May 14, 2017
  d3f0d03c
- [X86][XOP] XOP's general v16i8 shifts will be used instead of v8i16 shift + mask. · 5bef9c62
  Simon Pilgrim authored May 14, 2017
  5bef9c62
- [X86][SSE] Account for cost of extract/insert of v32i8 vector shifts · aa8dffb6
  Simon Pilgrim authored May 14, 2017
  aa8dffb6
- [X86][XOP] Account for cost of extract/insert of 256-bit vector shifts · 4599eaa0
  Simon Pilgrim authored May 14, 2017
  4599eaa0
May 10, 2017
- AMDGPU: Make some packed shuffles free · 3c5e4237
  Matt Arsenault authored May 10, 2017
  3c5e4237
May 09, 2017
- [AArch64] Consider widening instructions in cost calculations · 78fd46b2
  Matthew Simpson authored May 09, 2017
  78fd46b2
May 07, 2017
- [X86][AVX1] Improve 256-bit vector costs for integer unary intrinsics. · 2d1c6d6e
  Simon Pilgrim authored May 07, 2017
  2d1c6d6e
May 03, 2017
- Support arbitrary address space pointers in masked gather/scatter intrinsics. · ef5798ac
  Elad Cohen authored May 03, 2017
  ef5798ac
Apr 12, 2017
- [SystemZ] Fix target specific tests · ab85113a
  Renato Golin authored Apr 12, 2017
  ab85113a
- [SystemZ] Updated test fp-cast.ll · 9b487543
  Jonas Paulsson authored Apr 12, 2017
  9b487543
- [SystemZ] TargetTransformInfo cost functions implemented. · fccc7d66
  Jonas Paulsson authored Apr 12, 2017
  fccc7d66
Mar 21, 2017
- AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel · 3dbeefa9
  Matt Arsenault authored Mar 21, 2017
  3dbeefa9
Mar 16, 2017
- [BasicTTIImpl] Bugfix in getIntrinsicInstrCost() · facc4c5c
  Jonas Paulsson authored Mar 16, 2017
  facc4c5c
Mar 15, 2017
- [X86] Add missing BITREVERSE costs for SSE2 vectors and i8/i16/i32/i64 scalars · 06c70adc
  Simon Pilgrim authored Mar 15, 2017
  06c70adc
Mar 14, 2017
- [TargetTransformInfo] getIntrinsicInstrCost() scalarization estimation improved · a48ea231
  Jonas Paulsson authored Mar 14, 2017
  a48ea231
Feb 17, 2017
- [PPC] Give unaligned memory access lower cost on processor that supports it · 7ec2c720
  Guozhi Wei authored Feb 17, 2017
  7ec2c720
Feb 02, 2017
- [X86] Add costs for non-AVX512 single-source permutation integer shuffles · e6d59fdc
  Michael Kuperstein authored Feb 02, 2017
  e6d59fdc
Feb 01, 2017
- [X86] Extend single-source shuffle cost test to test more arches. NFC. · cb4ceeda
  Michael Kuperstein authored Feb 01, 2017
  cb4ceeda
Jan 25, 2017
- Revert "[PPC] Give unaligned memory access lower cost on processor that supports it" · 65144c85
  Daniel Jasper authored Jan 25, 2017
  65144c85
Jan 21, 2017
- [PPC] Give unaligned memory access lower cost on processor that supports it · a5c6ed5a
  Guozhi Wei authored Jan 20, 2017
  a5c6ed5a
Jan 15, 2017
- [CostModel][X86] Fix AVX512BW vector shift costs for vXi16 types · 6ed996cd
  Simon Pilgrim authored Jan 15, 2017
  6ed996cd
- [CostModel][X86] Drop separate AVX512VL checks - they match existing AVX512 costs · a6c1974e
  Simon Pilgrim authored Jan 15, 2017
  a6c1974e
- [CostModel][X86] Update vector shift tests to correctly check by non-constant uniform values. · 9b169e3c
  Simon Pilgrim authored Jan 15, 2017
  9b169e3c
Jan 14, 2017
- [CostModel][X86] Updated vXi64 ASHR costs on AVX512 targets now that D28604 has landed · d419b73a
  Simon Pilgrim authored Jan 14, 2017
  d419b73a
Jan 11, 2017
- [X86][AVX512BW] Vectorize v64i8 vector shifts · 5a81fefa
  Simon Pilgrim authored Jan 11, 2017
  5a81fefa
- Fix line endings · c22c889f
  Simon Pilgrim authored Jan 11, 2017
  c22c889f