Commits · 87da5f47ac0b459de051a55e076157ec054bf1db · Lorenzo Albano / LLVM bpEVL · GitLab

Jul 25, 2016
- [Hexagon] Add target feature to generate long calls · 080bebd2
  Krzysztof Parzyszek authored Jul 25, 2016
  080bebd2
- [ARM] Improve longMAC codegen test · d5ca0a65
  Sam Parker authored Jul 25, 2016
  d5ca0a65
- [mips] Optimize materialization of i64 constants · 61897520
  Simon Dardis authored Jul 25, 2016
  61897520
- [ARM] Small refactor of Thumb2 SMLA insts · 73f5c785
  Sam Parker authored Jul 25, 2016
  73f5c785
- [ARM] Enable ISel of SMMLS for ARM and Thumb2 · 68c71cd1
  Sam Parker authored Jul 25, 2016
  68c71cd1
- [AVX512] Add load folding support for the unmasked forms of the FMA instructions. · ce415ff9
  Craig Topper authored Jul 25, 2016
  ce415ff9
- [AVX512] Add some additional patterns so that we can fold broadcast loads in... · 318e40b6
  Craig Topper authored Jul 25, 2016
  318e40b6
- [AVX512] Cleanup FMA operand order in patterns to match the VEX versions and... · 6bcbf533
  Craig Topper authored Jul 25, 2016
  6bcbf533
Jul 24, 2016
- [X86] Add 'FeatureSlowSHLD' to cpu 'bdver4' · 381a0ade
  Simon Pilgrim authored Jul 24, 2016
  381a0ade
- [X86] Make the FMA3 instruction names consistent between VEX and EVEX encoded versions. · 2dca3b28
  Craig Topper authored Jul 24, 2016
  2dca3b28
- [X86] Replace CodeGenOnly VPSRAVW/D/Q_Int instructions with patterns since the... · 05629d05
  Craig Topper authored Jul 24, 2016
  05629d05
Jul 23, 2016
- [X86] Fix typo in comment. · 8152b9cd
  Craig Topper authored Jul 23, 2016
  8152b9cd
- Fix a GCC error due to this member name also being a type name. This · 488cb137
  Chandler Carruth authored Jul 23, 2016
  488cb137
- [AVX512] Implement commuting support for EVEX encoded FMA3 instructions. · b6519db9
  Craig Topper authored Jul 23, 2016
  b6519db9
- [X86] Make one of the FMA3 commuting methods static. Remove a call to isFMA3... · 6172b0b3
  Craig Topper authored Jul 23, 2016
  6172b0b3
- [X86] Fix switch statement indentation per coding standards. · ca8f5f30
  Craig Topper authored Jul 23, 2016
  ca8f5f30
- AMDGPU: Delete dead code · b40d8600
  Matt Arsenault authored Jul 23, 2016
  b40d8600
- Revert "[AMDGPU] Emit read-only data to .rodata for hsa" · b8253c88
  Tom Stellard authored Jul 22, 2016
  b8253c88
Jul 22, 2016
- Fix include case. NFC. · 8a457ddf
  George Burgess IV authored Jul 22, 2016
  8a457ddf
- GlobalISel: implement legalization pass, with just one transformation. · 33b07d67
  Tim Northover authored Jul 22, 2016
  33b07d67
- [Hexagon] Make HexagonCodeGen depend on Scalar · 3c89bb09
  Krzysztof Parzyszek authored Jul 22, 2016
  3c89bb09
- AMDGPU: Fix groupstaticsize for large LDS · 3c07c813
  Matt Arsenault authored Jul 22, 2016
  3c07c813
- AMDGPU: Add HSA dispatch id intrinsic · 8d718dcf
  Matt Arsenault authored Jul 22, 2016
  8d718dcf
- AMDGPU: Delete more dead code · f9245b75
  Matt Arsenault authored Jul 22, 2016
  f9245b75
- AMDGPU: Fix i1 fp_to_int · 7fb961f3
  Matt Arsenault authored Jul 22, 2016
  7fb961f3
- AMDGPU: Don't reinvent transferSuccessorsAndUpdatePHIs · d40ded66
  Matt Arsenault authored Jul 22, 2016
  d40ded66
- [RDF] Make the graph construction/use less expensive · 047149f7
  Krzysztof Parzyszek authored Jul 22, 2016
  047149f7
- [Hexagon] Use loop data prefetch on Hexagon · d3d0a4bd
  Krzysztof Parzyszek authored Jul 22, 2016
  d3d0a4bd
- [X86][AVX] Added support for lowering to VBROADCASTF128/VBROADCASTI128 (reapplied) · ea0d4f99
  Simon Pilgrim authored Jul 22, 2016
  ea0d4f99
- Revert "[X86][AVX] Added support for lowering to VBROADCASTF128/VBROADCASTI128" · 5ba0e203
  Benjamin Kramer authored Jul 22, 2016
  5ba0e203
- This refactoring of ARM machine block size computation creates two utility · 5c0ef833
  Sjoerd Meijer authored Jul 22, 2016
  5c0ef833
- [mips][microMIPS] Implement SLT, SLTI, SLTIU, SLTU microMIPS32r6 instructions · 2db00ce4
  Hrvoje Varga authored Jul 22, 2016
  2db00ce4
- [AVX512] Add ExeDomain to vector extend and truncate instructions. · 52e2e838
  Craig Topper authored Jul 22, 2016
  52e2e838
- [AVX512] Add initial support for the Execution Domain fixing pass to change some EVEX instructions. · f4151bea
  Craig Topper authored Jul 22, 2016
  f4151bea
- [AVX512] Fix the ExeDomain for some packed fp instructions. · 5ec33a94
  Craig Topper authored Jul 22, 2016
  5ec33a94
- [AVX512] Add load folding for some AVX512VL logic and arithmetic instructions. · 0b90756b
  Craig Topper authored Jul 22, 2016
  0b90756b
- [AVX512] Update X86InstrInfo::foldMemoryOperandCustom to handle the EVEX encoded instructions too. · ab13b33d
  Craig Topper authored Jul 22, 2016
  ab13b33d
- [AArch64] Cleanup sign extend in genAlternativeCodeSequence · 1182dd8e
  David Majnemer authored Jul 21, 2016
  1182dd8e
- [Sparc]: Fix bug in LowerSTORE due to r275592 · 26cfb6a2
  Douglas Katzman authored Jul 21, 2016
  26cfb6a2
- [X86] Do not use AND8ri8 in AVX512 pattern · c523333b
  Michael Kuperstein authored Jul 21, 2016
  c523333b