Commits · fbb8e772ec501a1b71643db90e9c6445e17d7cac · Roger Ferrer / llvm-epi · GitLab

Aug 19, 2021
- [CostModel][X86] Add isnan half/float/double costs tests · 72ebcd31
  Simon Pilgrim authored Aug 19, 2021
  
  72ebcd31
- [CostModel][X86] Add VPOPCNTDQ/BITALG ctpop costs · 9419729b
  Simon Pilgrim authored Aug 19, 2021
  9419729b
- [CostModel][X86] Add VPOPCNT/BITALG test coverage for ctpop/cttz costs · 2d60fdd7
  Simon Pilgrim authored Aug 19, 2021
  
  2d60fdd7
- [AArch64][SVE] Teach cost model that masked loads/stores are cheap · 734708e0
  Matthew Devereau authored Aug 19, 2021
  734708e0
- StackLifetime: Remove asserts for multiple lifetime intrinsics. · 6f85225e
  Peter Collingbourne authored Aug 18, 2021
  6f85225e
Aug 18, 2021
- [LICM] Remove AST-based implementation · 3dd8c917
  Nikita Popov authored Aug 17, 2021
  3dd8c917
- [Analysis][AArch64] Make fixed-width ordered reductions slightly more expensive · 219d4518
  David Sherwood authored Aug 18, 2021
  219d4518
Aug 17, 2021
- [SVE] Remove usage of getMaxVScale for AArch64, in favour of IR Attribute · ef198cd9
  Dylan Fleming authored Aug 17, 2021
  ef198cd9
Aug 16, 2021
- [MemorySSA] Remove -enable-mssa-loop-dependency option · 735a5904
  Nikita Popov authored Aug 14, 2021
  735a5904
Aug 14, 2021
- [Tests] Remove explicit -enable-mssa-loop-dependency options (NFC) · e11354c0
  Nikita Popov authored Aug 14, 2021
  e11354c0
Aug 12, 2021
- Recommit "[Matrix] Overload stride arg in matrix.columnwise.load/store." · f9993128
  Florian Hahn authored Aug 12, 2021
  f9993128
- Revert "[Matrix] Update column.major.load call in PPC test." · a72cd635
  Florian Hahn authored Aug 12, 2021
  a72cd635
- [Matrix] Update column.major.load call in PPC test. · 1f87c7c4
  Florian Hahn authored Aug 12, 2021
  1f87c7c4
Aug 10, 2021
- [NFC][X86] New Test Requires Asserts · b764b1ef
  Archibald Elliott authored Aug 10, 2021
  b764b1ef
- [X86] AVX512FP16 instructions enabling 1/6 · 6f7f5b54
  Wang, Pengfei authored Aug 10, 2021
  6f7f5b54
Aug 07, 2021
- [MemCpyOpt] Remove MemDepAnalysis-based implementation · 88003cea
  Nikita Popov authored May 08, 2021
  88003cea
Aug 06, 2021
- [LoopCacheAnalysis]: handle mismatch type for Numerator and CacheLineSize · 30b0c455
  Zheng Chen authored Aug 06, 2021
  30b0c455
- [AArch64] Expand the SVE min/max reduction costs to NEON · 649cf451
  David Green authored Aug 05, 2021
  649cf451
Aug 05, 2021
- [DA] control compile-time spent by MIV tests · 0e08891e
  Bardia Mahjour authored Aug 05, 2021
  0e08891e
Aug 02, 2021
- [AArch64] Optimise min/max lowering in ISel · b01417d3
  Irina Dobrescu authored Jul 22, 2021
  b01417d3
- [CostModel][AArch64] Add some shuffle concat tests. NFC. · 46a861af
  Sjoerd Meijer authored Aug 02, 2021
  46a861af
- [CostModel] Treat 'widen subvector' patterns as zero cost · 872a9500
  Simon Pilgrim authored Aug 02, 2021
  872a9500
- [TTI] Add basic SK_InsertSubvector shuffle mask recognition · 7397dcb4
  Simon Pilgrim authored Aug 02, 2021
  7397dcb4
Jul 27, 2021
- [AArch64] Update and expand min-max cost model test. NFC · 098984a8
  David Green authored Jul 27, 2021
  098984a8
Jul 26, 2021
- [Analysis] Fix getOrderedReductionCost to call target's getArithmeticInstrCost implementation · 77c5e6ba
  Simon Pilgrim authored Jul 26, 2021
  77c5e6ba
- [Analysis] Add simple cost model for strict (in-order) reductions · 0aff1798
  David Sherwood authored Jul 07, 2021
  0aff1798
Jul 24, 2021
- [BasicTTI] Set scalarization cost of scalable vector casts to Invalid. · c3277a88
  Sander de Smalen authored Jul 24, 2021
  c3277a88
- [tests] SCEV trip count w/ neg step and varying rhs · e9d4bb43
  Philip Reames authored Jul 23, 2021
  
  e9d4bb43
- [SCEV] Fix bug involving zero step and non-invariant RHS in trip count logic · 4a3dc7dc
  Philip Reames authored Jul 23, 2021
  4a3dc7dc
Jul 23, 2021
- [AArch64] Add worst case shuffle costs · 38986c67
  David Green authored Jul 23, 2021
  38986c67
Jul 22, 2021
- [CostModel][X86] Adjust shift SSE4 legalized costs based on llvm-mca reports. · 4185c550
  Simon Pilgrim authored Jul 22, 2021
  4185c550
- [CostModel][X86] Fix funnel shift check prefixes · 2657fe17
  Simon Pilgrim authored Jul 22, 2021
  2657fe17
- [AArch64] Adjust the cost of integer sum reductions · c9cebda7
  David Green authored Jul 22, 2021
  c9cebda7
- [CostModel][X86] Adjust shift SSE legalized costs based on llvm-mca reports. · e1bdb579
  Simon Pilgrim authored Jul 22, 2021
  e1bdb579
- [AArch64] Add and update reduction and shuffle costs. NFC · a92974bf
  David Green authored Jul 22, 2021
  
  a92974bf
Jul 20, 2021
- [tests] Add a couple of tests for zero stride trip counts w/loop varying exit values · 4c40cfc2
  Philip Reames authored Jul 19, 2021
  
  4c40cfc2
- [ScalarEvolution] Refine computeMaxBECountForLT to be accurate in more cases. · de3ea51b
  Eli Friedman authored Jul 16, 2021
  de3ea51b
Jul 19, 2021
- [CostModel][X86] Add fast math tests for float reductions · 5939c642
  Simon Pilgrim authored Jul 19, 2021
  5939c642
Jul 17, 2021
- [ScalarEvolution] Fix overflow in computeBECount. · cbba71bf
  Eli Friedman authored Jul 09, 2021
  cbba71bf
Jul 16, 2021
- [tests] Precommit test for D104140 · 5c57600b
  Philip Reames authored Jul 16, 2021
  
  5c57600b