Google Git
Sign in
opensecura / 3p / openxla / iree / 82be925b7c392d936f0250259777ba9c0ce2229c / compiler
  1. 1922b49 [Codegen][LLVMGPU] Add emulateNarrowTypes to LLVMGPU Pipeline (#14816) by Kunwar Grover · 1 year, 8 months ago
  2. b24a6e2 [Flow] Raise special `linalg.generic` ops to `linalg.fill` ops (#14773) by Max191 · 1 year, 8 months ago
  3. 8314b5c Support default configs for BATCH_MATMUL_* in MaterializeEncoding pass (#14762) by Jerry Wu · 1 year, 8 months ago
  4. b0c77fa Improving util.align folding. (#14805) by Ben Vanik · 1 year, 8 months ago
  5. 8f62f9b Fix ir_tool argparse subcommand usage. (#14804) by Scott Todd · 1 year, 8 months ago
  6. ffde368 [Flow] Teach RaiseSpecialOps to raise tensor.extract and views (#14718) by Kunwar Grover · 1 year, 8 months ago
  7. fdc873c [StableHLO] Fix null deref in topk rewrite (#14794) by Diego Caballero · 1 year, 8 months ago
  8. c477184 Introduce GlobalOptimization layers to Pipeline. (#14767) by Han-Chung Wang · 1 year, 8 months ago
  9. cf2a4b7 Switching --iree-stream-resource-index-bits= to 64 by default. (#14787) by Ben Vanik · 1 year, 8 months ago
  10. 42e54ab Integrate llvm-project at 063c42e919c0 (#14725) by Han-Chung Wang · 1 year, 8 months ago
  11. 5c4f971 Make iree-vm-deduplicate-rodata O(ops) instead of O(replacements*ops). (#14782) by Ben Vanik · 1 year, 8 months ago
  12. 2a914c1 [StableHLO] Merge reshape(reshape(x)) by mariecwhite · 1 year, 8 months ago
  13. faa25f9 Add python bindings for IREE internal dialects. (#14770) by Stella Laurenzo · 1 year, 8 months ago
  14. e12b77c Fix TypePropagation pass to handle functions with control flow. (#14626) by MaheshRavishankar · 1 year, 8 months ago
  15. 5142828 [IREEInput] Add iree_input.buffer.subspan operation (#14756) by Eugene Zhulenev · 1 year, 8 months ago
  16. 56071dd Restrict the constexpr constant types to the subset we currently support (#14758) by Stella Laurenzo · 1 year, 8 months ago
  17. 836f6f1 [spirv] Fix element extraction order when breaking down vectors (#14755) by Lei Zhang · 1 year, 8 months ago
  18. 5bfc37a [spirv] Drop leading unit dim in f16 fused dequant matvec dispatch (#14752) by Lei Zhang · 1 year, 8 months ago
  19. f50d0d9 [spirv] Allow dynamic parallel dims in subgroup reduction pipeline (#14751) by Lei Zhang · 1 year, 8 months ago
  20. 839375f Make rematerialization trigger before tile and distribute only for CPU and CUDA backends. (#14750) by MaheshRavishankar · 1 year, 8 months ago
  21. 73ddcce [spirv] Add some tests for matvec pipeline configuration (#14746) by Lei Zhang · 1 year, 8 months ago
  22. 959a8a3 [Codegen] Add RematerializeParallelOps pass back to codegen pipeline (#14744) by Kunwar Grover · 1 year, 8 months ago
  23. c5e3f46 Add copy (cp) command to ir_tool. (#14743) by Scott Todd · 1 year, 8 months ago
  24. 14def8b [Flow] Add FuseDequantizationMatmulPass to flow passes (#14734) by Max191 · 1 year, 8 months ago
  25. 8885948 Rename Embed.cpp -> CompilerDriver.cpp. (#14733) by Stella Laurenzo · 1 year, 8 months ago
  26. ce0a612 Adds CAPI interop to export a module from an invocation. (#14722) by Stella Laurenzo · 1 year, 8 months ago
  27. 9b722a7 [Stream] Add timepoints and subranges propagation through cf.switch (#14728) by Eugene Zhulenev · 1 year, 8 months ago
  28. ff09cb2 [spirv] Improve CodeGen flow for transposed matmul (#14708) by Max191 · 1 year, 8 months ago
  29. 71d772b [Stream] Add structural transformation for cf.switch op (#14727) by Eugene Zhulenev · 1 year, 8 months ago
  30. 461ba76 [Flow] Add pass to fuse dequantization and matmul linalg.generic ops (#14638) by Max191 · 1 year, 8 months ago
  31. 858dcfe Add materialization of set/unset encoding for `batch_matmul` (#14705) by Jerry Wu · 1 year, 8 months ago
  32. b69a189 Fix dynamic offset handling of sub-byte element types in lowering to LLVM (#14664) by MaheshRavishankar · 1 year, 8 months ago
  33. 7aadf17 [spirv] Let matvec go down subgroup reduction pipeline (#14714) by Lei Zhang · 1 year, 8 months ago
  34. 7d4abf0 Register IREECodegenDialect in the transform dialect interpreter pass (#14720) by Quinn Dawkins · 1 year, 8 months ago
  35. 5320738 Fixing a bug in --iree-util-fold-globals when globals are uninitialized. (#14719) by Ben Vanik · 1 year, 8 months ago
  36. edb0610 Cherry-pick MLIR changes for multi reduction tiling and distribution (#14715) by Lei Zhang · 1 year, 8 months ago
  37. 3d5a12d [SPIRV] Do not multibuffer GEMMs with fused leading ops (#14712) by Quinn Dawkins · 1 year, 8 months ago
  38. e9b928f Adding --emit-mlir-bytecode flag to iree-compile. (#14710) by Ben Vanik · 1 year, 8 months ago
  39. 71081dd Move SCFToCFG into Util dialect and later in the pipeline (#14583) by Rob Suderman · 1 year, 8 months ago
  40. 8c88429 [CPU] Add memref aliasing ops folding right before sub-type emulation (#14569) by Han-Chung Wang · 1 year, 8 months ago
  41. 7e37aeb Integrate llvm-project at 030e315ee771 (#14678) by Han-Chung Wang · 1 year, 8 months ago
  42. 562f2ed NFC: Move RematerializeParallelOps to CodeGen/Common (#14693) by Lei Zhang · 1 year, 8 months ago
  43. 42b983c Removing initial_data from iree_hal_allocator_allocate_buffer. (#14674) by Ben Vanik · 1 year, 8 months ago
  44. f022d29 Reworking constant upload with a HAL file API. (#14665) by Ben Vanik · 1 year, 8 months ago
  45. d4e17ed Fixes stream.resource.subview use during iree-stream-schedule-allocation. (#14689) by Ben Vanik · 1 year, 8 months ago
  46. aef8d3c Add a StableHLO canonicalizer that rewrites transpose as reshape. (#14682) by NatashaKnk · 1 year, 8 months ago
  47. 4f77c30 [spirv] Fix SPIRVVectorizeLoadStore regarding sub-byte elements (#14687) by Lei Zhang · 1 year, 8 months ago
  48. 528fdd5 Integrate llvm-project at 913f21ae5c46 (#14659) by Han-Chung Wang · 1 year, 8 months ago
  49. df0a814 Revert "Fold `linalg.matmul(a, linalg.transpose(b))` into `linalg.matmul_transpose_b(a, b)` (#14645)" (#14663) by MaheshRavishankar · 1 year, 8 months ago
  50. dce0e41 Fix path location for packaged tools (shows up in releases). by Stella Laurenzo · 1 year, 8 months ago
  51. 4a34723 Fold `linalg.matmul(a, linalg.transpose(b))` into `linalg.matmul_transpose_b(a, b)` (#14645) by MaheshRavishankar · 1 year, 8 months ago
  52. 730c34a [API][Stream] Expose pipeline flag for fusing dispatch bindings (#14651) by Quinn Dawkins · 1 year, 8 months ago
  53. d619da5 Fix bounds check for dispatch break/trace pass (#14650) by Quinn Dawkins · 1 year, 8 months ago
  54. 6b0930c Fix calculation of byte offset to element offset during bufferization. (#14647) by MaheshRavishankar · 1 year, 8 months ago
  55. 1466742 [gpu] Consider operand element type in shared memory distribution (#14640) by Lei Zhang · 1 year, 8 months ago
  56. dca2f1f Add pass to decompose batch_mmt4d ops into mmt4d (#14628) by Jerry Wu · 1 year, 8 months ago
  57. e372c2d [gpu] Distribute fused producer elementwise ops in SIMT pipeline (#13908) by Lei Zhang · 1 year, 8 months ago
  58. faff7e0 [spirv] Enable vectorizing linalg.generic with sub-byte types (#14637) by Lei Zhang · 1 year, 8 months ago
  59. 8ab9e12 Fix python lib windows path issue exposed in postsubmit. by Stella Laurenzo · 1 year, 8 months ago
  60. ecc49f6 Add a CLI `iree-ir-tool` with a command to strip data. (#14636) by Stella Laurenzo · 1 year, 8 months ago
  61. 57b9239 Refine release build CMake options. (#14611) by Scott Todd · 1 year, 8 months ago
  62. e46fd20 [CPU] Enable masking for single pack/unpack dispatch if supported. (#14632) by Han-Chung Wang · 1 year, 8 months ago
  63. 91099ae Initialize results of buffer.load.* ops to zero. (#14589) by Simon Camphausen · 1 year, 8 months ago
  64. ca51caa Making `iree-util-strip-and-splat-constants` use byte_patterns. (#14627) by Ben Vanik · 1 year, 8 months ago
  65. 1ef5acf Integrate llvm-project@679c076ae446 (#14573) by Okwan Kwon · 1 year, 8 months ago
  66. a9ff376 Avoid constant folding of operations not in the same scope as the constant. (#14612) by MaheshRavishankar · 1 year, 8 months ago
  67. eeb6e80 Add a preprocessing pass to move entire function into a single dispatch. (#14578) by MaheshRavishankar · 1 year, 8 months ago
  68. 31738a8 Build and publish Windows Python releases. (#14575) by Scott Todd · 1 year, 8 months ago
  69. a62652e Fix bug with stablehlo canonicalizer which reordered operations that had more than one use. (#14599) by NatashaKnk · 1 year, 8 months ago
  70. d9b7f6d [HAL] Remove incorrect buffer_view.buffer folding (#14590) by Eugene Zhulenev · 1 year, 8 months ago
  71. 1a0b3fd Add error message for InlineExecutables with unsupported variants. (#14593) by Scott Todd · 1 year, 8 months ago
  72. b47ed8c [CPU] Add `contract` fast-math-flag to arith operations (#14551) by Diego Caballero · 1 year, 8 months ago
  73. 9e14c22 Only emplace dispatch results up to the first unemplacable one. (#14581) by Ben Vanik · 1 year, 8 months ago
  74. 9fa2998 Add canonicalizer that reorders unary elementwise ops and shape manipulation ops. (#14494) by NatashaKnk · 1 year, 8 months ago
  75. 3592c59 [SPIRV] Add transform dialect JIT mode (#14485) by Quinn Dawkins · 1 year, 8 months ago
  76. 1791958 Add simple pass to turn dense attributes into dense_resource attributes. (#14574) by Stella Laurenzo · 1 year, 8 months ago
  77. c35c88e Improving consteval support for non-ElementsAttrs attributes. (#14570) by Ben Vanik · 1 year, 8 months ago
  78. 3926ab7 Rework CFG detupling for nested regions (#14564) by Rob Suderman · 1 year, 8 months ago
  79. a5bfcd1 Add support for padding by complex values (#14367) by Rob Suderman · 1 year, 8 months ago
  80. 9d15f36 Adding `--module_mode=mmap` tooling flag. (#14563) by Ben Vanik · 1 year, 8 months ago
  81. ac0b103 Avoid folding unit extent dims within pre-formed dispatches. (#14394) by MaheshRavishankar · 1 year, 8 months ago
  82. e450c61 Register Arm SME passes in mlir::registerMlirPasses() (#14396) by Benjamin Maxwell · 1 year, 8 months ago
  83. 83284b2 Fix VMVX StrideBufferDescriptor initialization (#14541) by CindyLiu · 1 year, 8 months ago
  84. 54a9cca Fix forward mid-air API collision on #14556. (#14562) by Stella Laurenzo · 1 year, 8 months ago
  85. 509d27c Searches for tools relative to the current libIREECompiler.so. (#14559) by Stella Laurenzo · 1 year, 8 months ago
  86. 2e0657f Adding the `#util.byte_pattern` attribute. (#14556) by Ben Vanik · 1 year, 8 months ago
  87. 93f86d9 Integrate llvm-project@3eebf005ca07 (#14543) by Okwan Kwon · 1 year, 8 months ago
  88. a3d75cc Flip constexpr hoisting on globally. by Stella Laurenzo · 1 year, 8 months ago
  89. 0512707 Add SCF tuple flattening to StableHLO preprocessing (#14548) by Rob Suderman · 1 year, 8 months ago
  90. 45e9ff4 Running cleanup prior to iree-stream-outline-constants. by Ben Vanik · 1 year, 8 months ago
  91. 48f85e3 Rework consteval to use SerializableAttrInterface. by Stella Laurenzo · 1 year, 8 months ago
  92. 489d6bf Revert "Turn off GCC warnings from third_party libraries" (#14554) by Ben Vanik · 1 year, 8 months ago
  93. fb8263f [Transform] Add transform.iree.pack_shared_memory_alloc (#14503) by Kunwar Grover · 1 year, 8 months ago
  94. afedc36 Add i8 support to rng_bit_generator (#14483) by jvstokes · 1 year, 8 months ago
  95. af5b95b Turn off GCC warnings from third_party libraries (#14538) by CindyLiu · 1 year, 8 months ago
  96. cd3f195 Disable constexpr hoisting of types not explicitly known/allowed. (#14544) by Stella Laurenzo · 1 year, 8 months ago
  97. 8da7e9c Use default pipeline for dispatches with no compute ops. (#14514) by MaheshRavishankar · 1 year, 8 months ago
  98. 24e9133 Integrate llvm-project at 2dc1a27449a9 (#14532) by Han-Chung Wang · 1 year, 8 months ago
  99. c888ed5 Fix some bugs in consteval related to different type combinations. (#14534) by Stella Laurenzo · 1 year, 8 months ago
  100. 791f7a5 Fix StreamOps lambda function (#14522) by CindyLiu · 1 year, 8 months ago