Commits · bce1f6b49196e174f299e577a42151e2adbf49d8 · Roger Ferrer / llvm-epi-0.8 · GitLab

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

Oct 02, 2016
- [X86][AVX2] Missed opportunities to combine to VPERMD/VPERMPS · bce1f6b4
  Simon Pilgrim authored Oct 02, 2016
  bce1f6b4
- [X86][AVX2] Fix typo in test names · b5200971
  Simon Pilgrim authored Oct 02, 2016
  b5200971
- [x86] remove 'nan' strings from copysign assertions; NFC · 170d7eb3
  Sanjay Patel authored Oct 02, 2016
  170d7eb3
- [x86] add test to show unnecessary scalarization of copysign intrinsics (PR30433) · dfbbbcd6
  Sanjay Patel authored Oct 02, 2016
  dfbbbcd6
- [X86][AVX] Ensure broadcast loads respect dependencies · 03afbe78
  Simon Pilgrim authored Oct 02, 2016
  03afbe78
- [PowerPC] Refactor soft-float support, and enable PPC64 soft float · a9321059
  Hal Finkel authored Oct 02, 2016
  a9321059
Oct 01, 2016
- [X86][SSE] Add support for combining target shuffles to binary BLEND · 1638d49f
  Simon Pilgrim authored Oct 01, 2016
  1638d49f
- [X86][SSE] Always combine target shuffles to MOVSD/MOVSS · ae17cf20
  Simon Pilgrim authored Oct 01, 2016
  ae17cf20
- [X86][SSE] Enable commutation from MOVSD/MOVSS to BLENDPD/BLENDPS on SSE41+ targets · ccdd1ff4
  Simon Pilgrim authored Oct 01, 2016
  ccdd1ff4
- [X86][SSE] Regenerate vselect tests and improve AVX1/AVX2 coverage · f1c575ba
  Simon Pilgrim authored Oct 01, 2016
  f1c575ba
- [X86] Cleanup patterns for using VMOVDDUP for broadcasts. · 5eb5ade8
  Craig Topper authored Oct 01, 2016
  5eb5ade8
- [AVX-512] Add VLX command lines to 128 and 256-bit shufffle tests. · 8aca9050
  Craig Topper authored Oct 01, 2016
  8aca9050
- Revert "AMDGPU: Don't use offen if it is 0" · 86eeda8e
  Mehdi Amini authored Oct 01, 2016
  86eeda8e
- AMDGPU: Don't use offen if it is 0 · 3070fdf7
  Matt Arsenault authored Oct 01, 2016
  3070fdf7
- [SEH] Emit the parent frame offset label even if there are no funclets · 9cb915b7
  Reid Kleckner authored Sep 30, 2016
  9cb915b7
Sep 30, 2016
- X86: Allow conditional tail calls in Win64 "leaf" functions (PR26302) · b5643b47
  Hans Wennborg authored Sep 30, 2016
  b5643b47
- [WebAssembly] Make register stackification more conservative · e9e6891b
  Derek Schuff authored Sep 30, 2016
  e9e6891b
- Revert "[RegAllocGreedy] Attempt to split unspillable live intervals" · 309eba75
  Dylan McKay authored Sep 30, 2016
  309eba75
- [RegAllocGreedy] Attempt to split unspillable live intervals · 2a80cc68
  Dylan McKay authored Sep 30, 2016
  2a80cc68
- Revert r282835 "[AVX-512] Always use the full 32 register vector classes for... · 3f37a418
  Craig Topper authored Sep 30, 2016
  3f37a418
- [AVX-512] Always use the full 32 register vector classes for addRegisterClass... · bc6e97b8
  Craig Topper authored Sep 30, 2016
  bc6e97b8
- AMDGPU: Use unsigned compare for eq/ne · 5d8eb25e
  Matt Arsenault authored Sep 30, 2016
  5d8eb25e
- [X86] Don't preserve Win64 SSE CSRs when SSE is disabled · 147f91c8
  Reid Kleckner authored Sep 30, 2016
  147f91c8
Sep 29, 2016
- [X86] Add explicit test triple to make windows/msvc builds happier · d72330dd
  Simon Pilgrim authored Sep 29, 2016
  d72330dd
- [X86] Really fix the FileCheck line from r282690. · bd74f756
  Craig Topper authored Sep 29, 2016
  bd74f756
- [AVX-512] Fix a check line from r282690. · 1b60e9d7
  Craig Topper authored Sep 29, 2016
  1b60e9d7
- [AVX-512] Support spills of XMM16-31 and YMM16-31 when VLX isn't available. · d875d6b9
  Craig Topper authored Sep 29, 2016
  d875d6b9
- [X86] Remove extra FileCheck lines that got left behind in r282688. · f91830e6
  Craig Topper authored Sep 29, 2016
  f91830e6
- [AVX-512] Replicate pattern from AVX to select VMOVDDUP for (v2f64... · 7eb0e7ce
  Craig Topper authored Sep 29, 2016
  7eb0e7ce
- [X86] Add EVEX encoded VBROADCASTSS/SD and VPBROADCASTD/Q to execution domain fixing table. · e7f26111
  Craig Topper authored Sep 29, 2016
  e7f26111
- [X86] Add 512-bit VPBROADCASTB and VPBROADCASTW tests. · 7da04650
  Craig Topper authored Sep 29, 2016
  7da04650
- [X86] Add VBROADCASTF128/VBROADCASTI128 to execution domain fixing tables. · 816a1d77
  Craig Topper authored Sep 29, 2016
  816a1d77
- AMDGPU: Partially fix control flow at -O0 · e6740754
  Matt Arsenault authored Sep 29, 2016
  e6740754
Sep 28, 2016
- IfConversion: Add implicit uses for redefined regs with live subregisters · dcb1bcae
  Krzysztof Parzyszek authored Sep 28, 2016
  dcb1bcae
- [AMDGPU] Promote uniform i16 ops to i32 ops for targets that have 16 bit instructions · e14df4b2
  Konstantin Zhuravlyov authored Sep 28, 2016
  e14df4b2
- [X86][AVX] Add test showing that VBROADCAST loads don't correctly respect dependencies · fea5c7a0
  Simon Pilgrim authored Sep 28, 2016
  fea5c7a0
- Teach LiveDebugValues about lexical scopes. · 7f5866c2
  Adrian Prantl authored Sep 28, 2016
  7f5866c2
- [NVPTX] Added intrinsics for atom.gen.{sys|cta}.* instructions. · 3e121158
  Artem Belevich authored Sep 28, 2016
  3e121158
- Revert "In visitSTORE, always use FindBetterChain, rather than only when UseAA is enabled." · e524f508
  Nirav Dave authored Sep 28, 2016
  e524f508
- In visitSTORE, always use FindBetterChain, rather than only when UseAA is enabled. · e17e055b
  Nirav Dave authored Sep 28, 2016
  e17e055b