Commits · f57ae33381f27445d2b174bfae2cd94d49852398 · Roger Ferrer / llvm-epi · GitLab

Mar 25, 2017
- [AMDGPU] Switch data layout by triple environment amdgiz · 14834c3e
  Yaxun Liu authored Mar 25, 2017
  14834c3e
Mar 24, 2017
- AMDGPU: Fix annotating loops with nested loop conditions · 0607a442
  Matt Arsenault authored Mar 24, 2017
  0607a442
- AMDGPU: Implement f16 fround · b5d23271
  Matt Arsenault authored Mar 24, 2017
  b5d23271
- AMDGPU: Unify divergent function exits. · b8f8dbc2
  Matt Arsenault authored Mar 24, 2017
  b8f8dbc2
- [AMDGPU] Fold V_CNDMASK with identical source operands · 70603dce
  Stanislav Mekhanoshin authored Mar 24, 2017
  70603dce
- [AMDGPU] Rename Kind to ValueKind in metadata to be consistent · 4986d9fb
  Konstantin Zhuravlyov authored Mar 24, 2017
  4986d9fb
- [AMDGPU] Add AMDGPUAliasAnalysis to opt pipeline · a27b2cac
  Stanislav Mekhanoshin authored Mar 24, 2017
  a27b2cac
- [AMDGPU] Don't enforce constexpr, there are still old standard libraries... · 80e3d5bb
  Benjamin Kramer authored Mar 24, 2017
  80e3d5bb
- [AMDGPU] Remove double map lookups in SI scheduler · e2419dc9
  Valery Pykhtin authored Mar 24, 2017
  e2419dc9
- [AMDGPU] Fix SGPR usage count in SI scheduler · f7d1023a
  Valery Pykhtin authored Mar 24, 2017
  f7d1023a
- [AMDGPU] Add a new line after a debug message · 57ab6999
  Valery Pykhtin authored Mar 24, 2017
  57ab6999
- Don't build up std::vectors with constant sizes when an array suffices. · c06d672a
  Benjamin Kramer authored Mar 24, 2017
  c06d672a
Mar 23, 2017
- [AMDGPU] Do not emit isa info as code object metadata · 4cbb6895
  Konstantin Zhuravlyov authored Mar 22, 2017
  4cbb6895
- [AMDGPU] Emit kernel debug properties as code object metadata · a780ffaa
  Konstantin Zhuravlyov authored Mar 22, 2017
  a780ffaa
Mar 22, 2017
- [AMDGPU] Emit kernel code properties as code object metadata · ca0e7f64
  Konstantin Zhuravlyov authored Mar 22, 2017
  ca0e7f64
- [AMDGPU] Restructure code object metadata creation · 7498cd61
  Konstantin Zhuravlyov authored Mar 22, 2017
  7498cd61
- [AMDGPU] Fix bug 31610 · eb685e5f
  Konstantin Zhuravlyov authored Mar 22, 2017
  eb685e5f
- [AMDGPU][MC] Fix for Bug 28204 + LIT tests · 895d377d
  Dmitry Preobrazhensky authored Mar 22, 2017
  895d377d
Mar 21, 2017
- AMDGPU: Remove hasSideEffects from SI_RETURN_TO_EPILOG · 513cb7a8
  Matt Arsenault authored Mar 21, 2017
  513cb7a8
- AMDGPU: Rename SI_RETURN · 5b20fbb7
  Matt Arsenault authored Mar 21, 2017
  5b20fbb7
- Let llvm.objectsize be conservative with null pointers · 56c7e88c
  George Burgess IV authored Mar 21, 2017
  56c7e88c
- AMDGPU: Buffer descriptor changes for GFX9 · 5c7a61d2
  Marek Olsak authored Mar 21, 2017
  5c7a61d2
- AMDGPU: Always use VGPR indexing on GFX9 · e22fdb9c
  Marek Olsak authored Mar 21, 2017
  e22fdb9c
- Rename AttributeSet to AttributeList · b518054b
  Reid Kleckner authored Mar 21, 2017
  b518054b
- AMDGPU: Fix not including v2i16/v2f16 in register class · 5af82a7a
  Matt Arsenault authored Mar 21, 2017
  5af82a7a
- AMDGPU: Fix asserting on 0 dmask for image intrinsics · f8fb605a
  Matt Arsenault authored Mar 21, 2017
  f8fb605a
- [AMDGPU] Iterative scheduling infrastructure + minimal registry scheduler · fd4c410f
  Valery Pykhtin authored Mar 21, 2017
  fd4c410f
- [ADMGPU] SDWA peephole optimization pass. · f60ad58d
  Sam Kolton authored Mar 21, 2017
  f60ad58d
Mar 20, 2017
- [AMDGPU] Run always inliner early in opt · 2534bc07
  Konstantin Zhuravlyov authored Mar 20, 2017
  2534bc07
- [AMDGPU][MC] Fix for Bugs 28201, 28199, 28170 + LIT tests · 1e124e18
  Dmitry Preobrazhensky authored Mar 20, 2017
  1e124e18
- [AMDGPU][MC] Fix for Bugs 28200, 28202 + LIT tests · 40af9c35
  Dmitry Preobrazhensky authored Mar 20, 2017
  40af9c35
- Revert "[AMDGPU] Run always inliner early in opt" · 8a67eb14
  Konstantin Zhuravlyov authored Mar 20, 2017
  8a67eb14
Mar 19, 2017
- Fix MSVC warning: "switch statement contains 'default' but no 'case' labels". NFCI. · 5fa1b9a1
  Simon Pilgrim authored Mar 19, 2017
  5fa1b9a1
Mar 18, 2017
- [AMDGPU] Add address space based alias analysis pass · 8e45acfc
  Stanislav Mekhanoshin authored Mar 17, 2017
  8e45acfc
Mar 17, 2017
- AMDGPU: Fix broken condition in hazard recognizer · 59ece95f
  Matt Arsenault authored Mar 17, 2017
  59ece95f
- AMDGPU: Fix handling of constant phi input loop conditions · e70d5dcf
  Matt Arsenault authored Mar 17, 2017
  e70d5dcf
- AMDGPU: Cleanup control flow intrinsics · c5b641ac
  Matt Arsenault authored Mar 17, 2017
  c5b641ac
- Only unswitch loops with uniform conditions · ee2dd785
  Stanislav Mekhanoshin authored Mar 17, 2017
  ee2dd785
Mar 16, 2017
- [AMDGPU] Run always inliner early in opt · f8050797
  Stanislav Mekhanoshin authored Mar 16, 2017
  f8050797
- AMDGPU: Allow sinking of addressing modes for atomic_inc/dec · 7dc01c96
  Matt Arsenault authored Mar 15, 2017
  7dc01c96