Google Git
Sign in
opensecura / 3p / openxla / iree / 886ac3e269454180b149abd68e230ca079a3744a / compiler / src
  1. f1bbc57 [LLVMGPU] Add loop invariant code motion before software pipelining (#12540) by Thomas · 2 years, 1 month ago
  2. 1148f72 Remove memory side effect from hal.interface.binding.subspan (#12494) by Quinn Dawkins · 2 years, 1 month ago
  3. 9a135c2 [CUDA] Remove workaround running passes in old pass manager (#12516) by Thomas · 2 years, 1 month ago
  4. d5a9b1b Adding compilation reentrancy tests and new HAL pipeline phases. (#12503) by Ben Vanik · 2 years, 1 month ago
  5. 70e6e52 Add additional options to the ApplyPatternsOp (#12519) by Nicolas Vasilache · 2 years, 1 month ago
  6. 8007d74 Adding min/max VM ops and VM buffer allocation alignment. (#12522) by Ben Vanik · 2 years, 1 month ago
  7. e6425d6 Starting support for HAL dispatch specialization. (#12483) by Ben Vanik · 2 years, 1 month ago
  8. ff9bc6a Bumping bytecode version. by Ben Vanik · 2 years, 1 month ago
  9. 3e6130e Adding alignment to vm.buffer.alloc and vm.buffer.clone. by Ben Vanik · 2 years, 1 month ago
  10. 459f5ff Adding i32/i64/f32 min/max ops. by Ben Vanik · 2 years, 1 month ago
  11. 84b4802 Fully retire CanonicalizedSequenceOp (#12467) by Nicolas Vasilache · 2 years, 1 month ago
  12. 769ffda Evolve transform dialect usage towards non-blanket-canonicalized sequences (#12465) by Nicolas Vasilache · 2 years, 1 month ago
  13. d015bbc Integrate llvm-project at 33cf40122279342b50f92a3a53f5c185390b6018 (#12475) by Jakub Kuderski · 2 years, 1 month ago
  14. 08ae97f Refreshing VM performance by separating verification and tweaking buffer access. (#12426) by Ben Vanik · 2 years, 1 month ago
  15. 91e73d3 Adding requirement bits to bytecode modules and bumping version. by Ben Vanik · 2 years, 1 month ago
  16. 38abd13 Add bazel `iree_link_compiler_shared` flag. (#12508) by Stella Laurenzo · 2 years, 1 month ago
  17. aaa47cd Switching vm.buffer.fill to index with elements. by Ben Vanik · 2 years, 1 month ago
  18. fef4534 Build libIREECompiler.so in bazel. (#12507) by Stella Laurenzo · 2 years, 1 month ago
  19. 98a9105 Remove duplicate sources from Common (are in CommonPasses). (#12505) by Stella Laurenzo · 2 years, 1 month ago
  20. 79cbc19 Fail vectorize pad pattern when the pad value is not constant. (#12484) by MaheshRavishankar · 2 years, 1 month ago
  21. f43f8a3 Pad fusion bufferization workaround. (#12425) by MaheshRavishankar · 2 years, 1 month ago
  22. 787b518 Add `i1` type propagation for named Linalg ops. (#12474) by MaheshRavishankar · 2 years, 1 month ago
  23. 23ca7bb [LLVMGPU] Enhance TensorPad pass to handle tensor.unpack ops. (#12458) by Han-Chung Wang · 2 years, 1 month ago
  24. da22c84 Re-Enable reduction masking (#12472) by Murali Vijayaraghavan · 2 years, 1 month ago
  25. 98061e3 Fix incorrect API usage in RewritePatterns (round 1) (#12466) by Matthias Springer · 2 years, 1 month ago
  26. 4674a50 Reserving x18 for AArch64 platforms (#12460) by Murali Vijayaraghavan · 2 years, 1 month ago
  27. 7875b8e Revert the latest integrate to fix issues with `--iree-codegen-llvmgpu-enable-transform-dialect-jit` (#12451) by Jakub Kuderski · 2 years, 1 month ago
  28. 51bdf0a Work around test failures in `mobilenetv3_fake_weights` (#12445) by Jakub Kuderski · 2 years, 1 month ago
  29. 9c00793 fix k dimension for matmul ukernel (#12443) by bjacob · 2 years, 1 month ago
  30. 47872bc Passing constant uniform operands to dispatches and ignores dynamic ones. (#12432) by Ben Vanik · 2 years, 1 month ago
  31. 902aada Fixing symbol DCE stripping executable sources on round-tripping. (#12429) by Ben Vanik · 2 years, 1 month ago
  32. 259d709 [NFC] Add proper debug messages to LLVMGPUUtils to be better informed… (#12437) by Nicolas Vasilache · 2 years, 1 month ago
  33. 908c427 Integrate llvm-project at 523640f7a4c0 (#12419) by Jakub Kuderski · 2 years, 1 month ago
  34. 564ede9 Retire obsolete transforms and tests (#12436) by Nicolas Vasilache · 2 years, 1 month ago
  35. 21aeee8 [Common][TransformDialect] Fix the usage of num_threads for TileToFor… (#12438) by qcolombet · 2 years, 1 month ago
  36. 0312dd3 [GPU] Add a new pipeline optimized for nvidia ampere target (#12427) by Thomas · 2 years, 1 month ago
  37. e1c15ed [LLVMGPU] Add good tile size for large matmul in tensorcore pipeline (#12430) by Thomas · 2 years, 1 month ago
  38. 8f23c69 Allow 0-D interface MemRef in FlattenMemRefSubspan (#12414) by Lei Zhang · 2 years, 1 month ago
  39. 475d64d Adding FMA folding in the VM. (#12382) by Ben Vanik · 2 years, 1 month ago
  40. bc2a673 Enable hoisting uniform memref.load in reduction pipeline (#12421) by Lei Zhang · 2 years, 1 month ago
  41. 26fc399 [vulkan][spirv] Plumb through support for KHR Integer Dot Product (#12424) by Jakub Kuderski · 2 years, 1 month ago
  42. 17eafc9 [LLVMGPU] Enable tensor.pack op e2e execution on cuda. (#12370) by Han-Chung Wang · 2 years, 1 month ago
  43. 6ae5762 support unspecified channel_handle and use_global_device_ids (#12412) by Okwan Kwon · 2 years, 1 month ago
  44. 4a65a33 Retire most of LinalgExt::(Un)PackOp usages and transformations. (#12253) by Han-Chung Wang · 2 years, 1 month ago
  45. 4a8d063 Drop the tensor.pack/unpack -> LinalgExt lowering from transform dialect (#12401) by Han-Chung Wang · 2 years, 1 month ago
  46. f1c1799 Switching to use tensor.pack/unpack ops for data-tiling (#12247) by Han-Chung Wang · 2 years, 1 month ago
  47. 6a84459 Fixes tiling sizes for pulled producers in TileAndDistribute (#12399) by Han-Chung Wang · 2 years, 1 month ago
  48. 96b61ba [metal] Populate proper capability and limits for SPIR-V CodeGen (#12391) by Lei Zhang · 2 years, 1 month ago
  49. 69d4380 Promote bf16 to f32. (#12344) by Daniel Ellis · 2 years, 1 month ago
  50. c365243 Revert "Enable reduction masking (#12003)" (#12384) by Stella Laurenzo · 2 years, 1 month ago
  51. 7465d7e [LLVMGPU] Improve how we distribute small inner shapes (#12368) by Thomas · 2 years, 1 month ago
  52. b4490f2 [flow] NFC: Move code for cloning producers into RegionOpUtils (#12349) by Lei Zhang · 2 years, 1 month ago
  53. c35411b Strongly home flag registration statics in C++ files. (#12372) by Stella Laurenzo · 2 years, 1 month ago
  54. cf49d69 Moving bytecode files under iree/vm/bytecode/. (#12371) by Ben Vanik · 2 years, 1 month ago
  55. c319c2d Adding initial dispatch instrumention support. (#12357) by Ben Vanik · 2 years, 1 month ago
  56. 289b9a1 Add IREE_COMPILER_BUILD_SHARED_LIBS option. (#12310) by Stella Laurenzo · 2 years, 1 month ago
  57. 7a6b21b Minor fixes found while enabling micro-kernel usage e2e. (#12354) by MaheshRavishankar · 2 years, 1 month ago
  58. a23b52b Remove ListenerGreedyPatternRewriteDriver (#12358) by Matthias Springer · 2 years, 1 month ago
  59. f8a4aa2 Integrate llvm-project at 219ba2fb7b0a (#12342) by Han-Chung Wang · 2 years, 1 month ago
  60. f3ea9ee Blocking stream resource subview walks at timeline ops. (#12353) by Ben Vanik · 2 years, 1 month ago
  61. cd92019 Enable reduction masking (#12003) by Diego Caballero · 2 years, 1 month ago
  62. 689e0fa [spirv] Support multiple reductions fused in single dispatch cases (#11975) by Lei Zhang · 2 years, 1 month ago
  63. eba1769 Add explicit template instantiations that seem to fail with old compi… (#12340) by Nicolas Vasilache · 2 years, 1 month ago
  64. f20da41 Integrate llvm/llvm-project@c33fd3b47faa (#12320) by Han-Chung Wang · 2 years, 1 month ago
  65. 309dc5b Separate the compiler stable API into compiler/bindings/c. (#12332) by Stella Laurenzo · 2 years, 1 month ago
  66. 17e1111 Adding sym_visibility to vm.import and fixing parsing. (#12325) by Ben Vanik · 2 years, 1 month ago
  67. fee7018 Adding support for the `vm.import.module` attribute to force linking. (#12328) by Ben Vanik · 2 years, 1 month ago
  68. 023b0de Adding `--iree-hal-preprocess-executables-with=` option. (#12313) by Ben Vanik · 2 years, 1 month ago
  69. b27c48a Enable conversion of `depthwise_conv_2d_nhwc_hwcm_q` to its fixed-point counterpart (#12314) by Diego Caballero · 2 years, 1 month ago
  70. 5c78e87 Rework the logic that avoids fusions when that lead to stack allocatiions. (#12277) by MaheshRavishankar · 2 years, 1 month ago
  71. 49de0b6 Update `ResolveBufferDescriptors` to handle `memref.extract_strided_metadata` (#12205) by MaheshRavishankar · 2 years, 1 month ago
  72. c9ef78a Data-tiling encodings: rename RHS_TRANSPOSE to RHS (#12298) by bjacob · 2 years, 1 month ago
  73. 3b1a0ac Fix distribution to warps (#12306) by Thomas · 2 years, 1 month ago
  74. c7b2912 Update references from `iree-org` to `openxla`. (#12304) by Scott Todd · 2 years, 1 month ago
  75. 1f0b62b Fix mhlo.scatter with i64 indices by casting to i32 (#12302) by Rob Suderman · 2 years, 1 month ago
  76. 38ad88e Adding support for linking bitcode files in LLVM CPU/GPU targets. (#12265) by Ben Vanik · 2 years, 1 month ago
  77. c889a4c Adding `iree_bitcode_library` using our own clang build. (#12097) by Ben Vanik · 2 years, 1 month ago
  78. de17d3d Scoping the tensor.empty -> flow pass to top level ops. (#12303) by Ben Vanik · 2 years, 1 month ago
  79. 7f6c57c Generating iree/builtins/device using our own clang. by Ben Vanik · 2 years, 2 months ago
  80. 5c2172c Added ComplexToStandardPass to the LLVM compilation pipelines (#12273) by Rob Suderman · 2 years, 1 month ago
  81. 2277151 Add transform dialect op to allow hoisting bounded allocs in a function (#12284) by Nicolas Vasilache · 2 years, 1 month ago
  82. a4c74dc Extend MapNestedForallToGpuThreadsOp to support distrution to warpId (#12272) by Thomas · 2 years, 1 month ago
  83. 96d959e Add an option to disable use of bundled LLVM. (#12256) by Stella Laurenzo · 2 years, 1 month ago
  84. 1f3295a [rv32] Enable 32-bit mulsi_extended lowering for +zve32f (#12266) by Diego Caballero · 2 years, 1 month ago
  85. 2f40854 [spirv] Add Mali i8 matmul config (#12271) by Jakub Kuderski · 2 years, 1 month ago
  86. 03b368c [NFC] Centralizing dispatch region constant inlining. (#12235) by Ben Vanik · 2 years, 1 month ago
  87. f59bd6e Remove `hal.interface.binding.subspan` ops that are only used `memref.assume_alignment_ops`. (#12269) by MaheshRavishankar · 2 years, 1 month ago
  88. ad8782e Integrate llvm/llvm-project@eb141867 (#12264) by Lei Zhang · 2 years, 1 month ago
  89. 5b8757b Adding `--iree-hal-substitute-executable=` flag. (#12240) by Ben Vanik · 2 years, 1 month ago
  90. e752a83 [spirv] Enable vectorized codegen for i8 matmul (#12262) by Jakub Kuderski · 2 years, 1 month ago
  91. 14828ec Pass by reference in reproducer causes crash (#12257) by Rob Suderman · 2 years, 1 month ago
  92. e231b39 Making paths required on `#hal.executable.object`. by Ben Vanik · 2 years, 1 month ago
  93. 9955ff0 Adding `--iree-hal-substitute-executable-*=` flags. by Ben Vanik · 2 years, 1 month ago
  94. cd5347c Support mhlo collective ops (#11988) by Okwan Kwon · 2 years, 1 month ago
  95. 261471c Plumb vector to mma.sync through the transform dialect (#12244) by Nicolas Vasilache · 2 years, 1 month ago
  96. f560f42 [NFC] Switch to use upstream transform::TrivialPatternRewriter. (#12245) by Han-Chung Wang · 2 years, 1 month ago
  97. 6d4aadb [NFC] Switch to use upstream mlir::verifyCompatibleShape method. (#12243) by Han-Chung Wang · 2 years, 1 month ago
  98. 6be3f5a [rv32] Expand `arith.mulsi_extended` before going to LLVM (#12241) by Jakub Kuderski · 2 years, 1 month ago
  99. 1290401 Integrate llvm/llvm-project@4f15267d (#12226) by Lei Zhang · 2 years, 1 month ago
  100. a04c262 [spirv] Vectorize integer extend ops in lowering to subgroup_mma (#12202) by Quinn Dawkins · 2 years, 1 month ago