Google Git
Sign in
opensecura / 3p / openxla / iree / b5b4ab71eb14f489085ac5f0472804a27d2f4f34
  1. b5b4ab7 Bump cpubuilder dockerfile image to newer multi-arch version. (#18558) by Scott Todd · 6 months ago
  2. 8872710 Revert "Removed the iree_thread_join in the cleanup of deferred_work_queue.c" (#18616) by Andrew Woloszyn · 6 months ago
  3. 14728a7 Bump torch-mlir to 9938abf25e1e7526ca7f43a8c49e9078c14fc55c (#18615) by Vivek Khandelwal · 6 months ago
  4. 76c3e61 [CodeGen] Fix the argument replacements in scf.forall op lowering. (#18613) by Han-Chung Wang · 6 months ago
  5. 66d0c31 [DispatchCreation] Disable batch mmt4d fusion as its not supported by backends (#18611) by Nirvedh Meshram · 6 months ago
  6. 32a44bd [DispatchCreation][NFC] Simplify the logic of dumping CollapseInfo. (#18610) by Han-Chung Wang · 6 months ago
  7. 7db91ce [Codegen][GPU] Change the location of barriers in forall fusion (#18542) by Quinn Dawkins · 6 months ago
  8. c3fa4d0 Fix `task_executor_initialize` in resource exhausted scenario. (#18609) by Stella Laurenzo · 6 months ago
  9. 97896ce Integrate LLVM at llvm/llvm-project@24d707e215a1e2 (#18606) by Han-Chung Wang · 6 months ago
  10. d18064f [COMMON] Select the last compute op that has workgroup tilin… (#18604) by Prashant Kumar · 6 months ago
  11. 5a2dd56 Removed the iree_thread_join in the cleanup of deferred_work_queue.c (#18605) by Andrew Woloszyn · 6 months ago
  12. b13d38b [DT] Collapse matmul_narrow_M/N field into round_dims_to attribute. (#18599) by Han-Chung Wang · 6 months ago
  13. d583958 simplify `DataTiledMMAAttr::buildMmaOperation` (#18597) by Benoit Jacob · 6 months ago
  14. 672ae82 Attach Fusion interface to `linalg.softmax` (#18550) by Ian Wood · 6 months ago
  15. 6634f0f Integrate LLVM at llvm/llvm-project@cebb7c010854e3 (#18596) by Han-Chung Wang · 6 months ago
  16. 0b29f7b [GPU][DT] Add support for GPU data-tiling E2E tests. (#18591) by Han-Chung Wang · 6 months ago
  17. 7290283 Add F16 support for benchmark. (#18580) by erman-gurses · 6 months ago
  18. 3773a48 Lower data-tiled multi_mma to intrinsics. (#18547) by Benoit Jacob · 6 months ago
  19. 129ad45 Fixing cconv printing of function signatures with multiple tuples. (#18595) by Ben Vanik · 6 months ago
  20. 9158a90 GPU data tiling: Refine tile dimensions, more preparation for thread distribution. (#18556) by Benoit Jacob · 6 months ago
  21. b2dd6db [Encoding] Retire original_type field. (#18586) by Han-Chung Wang · 6 months ago
  22. 863ca01 Integrate LLVM at llvm/llvm-project@3fbf6f8bb183ad (#18590) by Han-Chung Wang · 6 months ago
  23. e6cf5bb [GPU][DT] Add support for materializing encoding ops with dynamic shape. (#18585) by Han-Chung Wang · 6 months ago
  24. e19950c Integrate LLVM at llvm/llvm-project@40d6497a97a61e (#18581) by Han-Chung Wang · 6 months ago
  25. ae6e5d3 [EmitC] Fix Windows builds (#18546) by Simon Camphausen · 6 months ago
  26. c0909a4 [gpu] Use clustered gpu.subgroup_reduce for nested layout distribution (#18515) by Andrea Faulds · 6 months ago
  27. 0d9c5a8 [GPU][DT] Add support for materializing tensor.empty and linalg.fill ops (#18563) by Han-Chung Wang · 6 months ago
  28. 9d7eb9f [docs] Fix AMDGPU to target chip mapping (#18584) by Jakub Kuderski · 6 months ago
  29. bddda85 Fix iree-compile command line call (#18583) by Marius Brehler · 6 months ago
  30. 070ec4a Add missing dep in LLVMGPU package to fix Bazel build. (#18582) by Scott Todd · 6 months ago
  31. ac03e05 [NFC][Codegen] Move LLVMCPUDropVectorUnitDims to Common (#18578) by Quinn Dawkins · 6 months ago
  32. 6fd9697 Remove build_tools/docker/ files. (#18566) by Scott Todd · 6 months ago
  33. 51329bf Migrate ci_linux_arm64_clang to new dockerfile. (#18569) by Scott Todd · 6 months ago
  34. b08cf02 [torchmlir-bump] Bump torch-mlir to 99848265c388 (#18579) by Gaurav Shukla · 6 months ago
  35. 328c32a Integrate LLVM at llvm/llvm-project@f264d9a9d56f (#18577) by Prashant Kumar · 6 months ago
  36. eef4623 [LLVMGPU][ROCm] Move kernel annotation before serialization (#18573) by Jakub Kuderski · 6 months ago
  37. 5a6bd8d [Codegen][GPU] Use I64ArrayAttr for tile sizes for simpler printing (#18575) by Quinn Dawkins · 6 months ago
  38. 9ee061d [LinalgExt] Masked Attention Implementation (#18525) by rohan-tan-bhowmik · 6 months ago
  39. 891f438 [Codegen] Add control options in pack unpack decomposition (#18469) by Max191 · 6 months ago
  40. d834aa7 [GPU] Add workgroup/subgroup scope specification to mma attr interface (#18548) by Max191 · 6 months ago
  41. 546d862 Fix experimental/web/ samples after recent changes. (#18567) by Scott Todd · 6 months ago
  42. 914858f [VectorDistribution] Reuse intrinsic layout in chained gemm (#18505) by Kunwar Grover · 6 months ago
  43. 0f15c8d [LLVMGPU][ROCm] Add validation on finalized llvm bitcode (#18552) by Jakub Kuderski · 6 months ago
  44. a5f63cc Move `linux_x64_bazel` job back to running on every commit. (#18560) by Scott Todd · 6 months ago
  45. 9588e7f Drop unused header from CombineBarrierRegions.cpp. (#18559) by Scott Todd · 6 months ago
  46. 782f372 [Codegen] Check for workgroup level tile sizes in workgroup tiling (#18538) by Max191 · 6 months ago
  47. 73ffafb [Codegen][GPU] Add support for bufferizing iree_gpu.barrier_region (#18497) by Quinn Dawkins · 6 months ago
  48. 75d5aab [Codegen][GPU] Add pass to combine adjacent barrier_region ops (#18541) by Quinn Dawkins · 6 months ago
  49. c9eca66 [Codegen][GPU] Allow iree_gpu.barrier_region to take multiple operands/results (#18490) by Quinn Dawkins · 6 months ago
  50. fa44a32 [LLVMGPU] Explicitly set configs for vector distribution pipeline lowering tests (#18553) by Kunwar Grover · 6 months ago
  51. 04144f6 [Codegen][GPU] Allow odd workgroup sizes when resolving non-warp foralls (#18549) by Quinn Dawkins · 6 months ago
  52. 0636abd Switch build_test_all_bazel to new dockerfile and runners. (#18533) by Scott Todd · 6 months ago
  53. 3a62d5c [GPU] Fix out of bounds access in setTileAndFuseLoweringConfig (#18537) by Max191 · 6 months ago
  54. 637190e [VectorDistribution] Add LICM to LLVMGPUVectorDistribute pipeline (#18510) by Kunwar Grover · 6 months ago
  55. f138e23 [Codegen] Add support for ParallelInsertSliceOp in DPS analysis (#18536) by Max191 · 6 months ago
  56. 337d49c [LinalgExt] Use f32 for accumulation for online_attention (#18456) by Kunwar Grover · 6 months ago
  57. 30b6374 [GPU][DT][NFC] Clenaup TODOs, styles and simplify logics. (#18544) by Han-Chung Wang · 6 months ago
  58. ad8f814 [LLVMGPU] Delete dead code in prefetch pass (#18543) by Jakub Kuderski · 6 months ago
  59. 6a44005 [GPU] Use alloca for private memory allocations (#18540) by Jakub Kuderski · 6 months ago
  60. 740e301 Preparation for data-tiled `multi_mma` codegen (#18532) by Benoit Jacob · 6 months ago
  61. 6fdc30f Start LLVM integrate integrates/llvm-20240917 (#18535) by Prashant Kumar · 6 months ago
  62. 7d823d2 [torch] Add dynamic support for `tm_tensor.attention` (#18527) by Rob Suderman · 6 months ago
  63. 898a95f Redirect links from GCP to Azure for RISCV/ARM files. (#18531) by Eliasj42 · 6 months ago
  64. f86a27d Bump torch-mlir to d6cf718 (#18530) by saienduri · 6 months ago
  65. a63717b Skip tests that have been failing on Windows nightly builds. (#18529) by Scott Todd · 6 months ago
  66. f2eaa2a [Codegen][LLVMGPU] Default to private memory space for scalar dispatches (#18523) by Quinn Dawkins · 6 months ago
  67. 5612307 Disable workflows still relying on GCP self-hosted runners. (#18526) by Scott Todd · 6 months ago
  68. 600cec4 Remove todo given upstream changes have landed (#18528) by Jacques Pienaar · 6 months ago
  69. cc891ba [Infra] Migrate rest of linux builder workflows off GCP runners. (#18511) by saienduri · 7 months ago
  70. 27b0829 Bump llvm/llvm-project@030c6da7af826b641db005be925b20f956c3a6bb (#18512) by Rob Suderman · 7 months ago
  71. b249aa7 [Flow] Add pass to fuse encoding ops into dispatch regions after hoisting (#18069) by Max191 · 7 months ago
  72. d1b4214 [EmitC] Cache values instead of operations (#18507) by Simon Camphausen · 7 months ago
  73. 70a5313 [GlobalOpt] Make extract_slice a non-leaf op (#18500) by Ian Wood · 7 months ago
  74. febe0ed Remove unused macro (#18504) by Kunwar Grover · 7 months ago
  75. 1dee3bd [IGEMM] Set lowering config on IGEMM matmuls (#18495) by Max191 · 7 months ago
  76. c7351e1 [PACKAGE] fix: make function really use the default value (#18486) by maxbartel · 7 months ago
  77. 861695b Integrates/llvm 20240910 (#18480) by Nirvedh Meshram · 7 months ago
  78. 5a6521c [Codegen][GPU] Add pass to resolve scf.forall ops (#18394) by Quinn Dawkins · 7 months ago
  79. 0dd358d [LLVMGPU] Tune tile sizes for large matmuls in TileAndFuse (#18493) by Max191 · 7 months ago
  80. 4395c11 GPU data tiling changes from `shared/gpu-data-tiling-materialize-encoding` (#18492) by Benoit Jacob · 7 months ago
  81. bb82e78 Adding HSA runtime headers submodule. (#18491) by Ben Vanik · 7 months ago
  82. 749ff34 [LLVMGPU] Add loop prefetching and use pipeline options in TileAndFuse (#18476) by Max191 · 7 months ago
  83. a0b56bd [DispatchCreation] Add flag for gather fusion (#18470) by Ian Wood · 7 months ago
  84. a9cbd54 [Bufferization] Fix alias mapping for ValueBarrierOpBufferizationInterface (#18475) by Max191 · 7 months ago
  85. e2464dd [Codegen] Add consumer fusion (#18427) by Prashant Kumar · 7 months ago
  86. 60843ec [VectorDistribution] Add support for multi-subgroup attention (#18188) by Kunwar Grover · 7 months ago
  87. c3cbfbd [Infra] Initial Migration off GCP Runners (#18381) by saienduri · 7 months ago
  88. 15d58e7 [NFC][GPU] Move LLVMGPUPipelineOptions to iree_gpu dialect (#18458) by Max191 · 7 months ago
  89. b197555 [EmitC] Remove `clearStruct` helper (#18464) by Simon Camphausen · 7 months ago
  90. 562e215 [Codegen][Common] Resolve `scf.forall` that are used for workgroup distribution (#18368) by MaheshRavishankar · 7 months ago
  91. 7212b48 Fixing leak when VM disassembly fails and making failure non-fatal. (#18399) by Ben Vanik · 7 months ago
  92. a730349 Enabling indirect command buffers and memoization by default. (#18467) by Ben Vanik · 7 months ago
  93. 7db3bdb Fixing issues found when enabling indirect command buffers. (#18382) by Ben Vanik · 7 months ago
  94. d5c4ef1 [LLVMGPU] Improve mfma tile sizes for TileAndFuse pipeline (#18459) by Max191 · 7 months ago
  95. 69ca7df [linalgext] Remove restriction on index type (#18437) by Rob Suderman · 7 months ago
  96. 97710b3 [Codegen] Add patterns for folding away no-op slices (#18419) by Quinn Dawkins · 7 months ago
  97. 6c9aad0 [Codegen][GPU] Improve forall hoisting pattern for single trip loops (#18418) by Quinn Dawkins · 7 months ago
  98. edc5d5e [Im2col] Add option to unroll decomposed im2col loops (#18342) by Max191 · 7 months ago
  99. b78def2 Update and pin dependencies (#18454) by Marius Brehler · 7 months ago
  100. 84d0789 Remove old `ireec` console script in favor of `iree-compile`. (#18435) by Scott Todd · 7 months ago