Commits

Branch selector

 main

User selector

All users

All time

Commit History

Commits on May 11, 2026

• [Codegen] Fix RAW in_bounds false for R&W (#24418)

  Show description for cd8e78a
  amd-eochoalo

  authored
  cd8e78a

  Copy full SHA for cd8e78a

• [LinalgExt] Refactor IGEMM detail computation and update layout decisions (#24245)

  Show description for 0cafdfa
  Max191

  authored
  0cafdfa

  Copy full SHA for 0cafdfa

• [VectorDistribute] Fix gfx1100 pipeline test (#24402)

  Show description for 17b796c
  sommerlukas

  authored
  17b796c

  Copy full SHA for 17b796c

Commits on May 9, 2026

• Revert "[Codegen][CPU] Replace transposed_intrinsic with explicit pre-transposed enums." (#24427)

  Show description for 0629d80
  kuhar

  authored
  0629d80

  Copy full SHA for 0629d80

• Revert "[Codegen][CPU] Add e2e matmul tests for the inner_tiled lowering path." (#24421)

  Show description for 919e956
  bjacob

  authored
  919e956

  Copy full SHA for 919e956

Commits on May 8, 2026

• [tools] Account both atol and rtol in floating point comparison. (#24394)

  Show description for f6e1872
  hanhanW

  authored
  f6e1872

  Copy full SHA for f6e1872

• [Codegen][CPU] Add e2e matmul tests for the inner_tiled lowering path. (#24405)

  Show description for fbe60d8
  bjacob

  authored
  fbe60d8

  Copy full SHA for fbe60d8

• [Codegen][CPU] Replace transposed_intrinsic with explicit pre-transposed enums. (#24404)

  Show description for a65bb23
  bjacob

  authored
  a65bb23

  Copy full SHA for a65bb23

• [INTEGRATION] Bump llvm to 180409dc6757 (#24409)

  Show description for 2bf97ad
  lialan

  authored
  2bf97ad

  Copy full SHA for 2bf97ad

• Revert "[LinalgExt] Fix attention NaN for fully-masked rows" (#24413)

  Show description for 83b0606
  hanhanW

  authored
  83b0606

  Copy full SHA for 83b0606

• [Codegen] Use divisibility analysis for im2col vectorization (#24318)

  Show description for b1ddcab
  Max191

  authored
  b1ddcab

  Copy full SHA for b1ddcab

• [VectorExt] Add generic lowering of iree_vector_ext.arg_compare (#24387)

  Show description for 23c1830
  bangtianliu

  authored
  23c1830

  Copy full SHA for 23c1830

• [VectorDistribute] Analysis to propagate promotion information (#24227)

  Show description for 1d87665
  sommerlukas

  authored
  1d87665

  Copy full SHA for 1d87665

• Enable building without google benchmark dependency (#24167)

  Show description for e156832
  

  AaronStGeorge

  and

  claude

  authored
  e156832

  Copy full SHA for e156832

• [Codegen][GPU] Remove the single-iteration workaround and distribute pad-fused copies (#24116)

  Show description for 8107036
  jerryyin

  authored
  8107036

  Copy full SHA for 8107036

• [Codegen] Limit async scope in pipelining (#24350)

  Show description for 597629e
  sommerlukas

  authored
  597629e

  Copy full SHA for 597629e

• [VectorExt] Change `to_layout` `shared_memory_conversion` (#24377)

  Show description for 98ddf0c
  sommerlukas

  authored
  98ddf0c

  Copy full SHA for 98ddf0c

• [Codegen] Move remaining pipelines to `iree_codegen` attrs (#24398)

  Show description for c0f5d4b
  kuhar

  authored
  c0f5d4b

  Copy full SHA for c0f5d4b

Commits on May 7, 2026

• [CPU] Improve tiling config for elementwise ops with dynamic shapes. (#24383)

  Show description for 8cc05f5
  hanhanW

  authored
  8cc05f5

  Copy full SHA for 8cc05f5

• Revert "[Codegen] Enable DMA by default for F16/BF16 Gemm on gfx950 (#24373)" (#24395)

  Show description for 16191ce
  Yu-Zhewen

  authored
  16191ce

  Copy full SHA for 16191ce

• [Torch][LinalgExt] Support GQA in torch.hop_flex_attention lowering (#24313)

  Show description for 0fe0ca8
  

  keshavvinayak01

  and

  codex

  authored
  0fe0ca8

  Copy full SHA for 0fe0ca8

• [Compiler] Use Repeated<T> for repeated value ranges. NFC. (#24392)

  Show description for d62d69b
  kuhar

  authored
  d62d69b

  Copy full SHA for d62d69b

• [Codegen][CPU] Add a type-polymorphic generic-scalar MMA fallback. (#24389)

  Show description for c4da71c
  bjacob

  authored
  c4da71c

  Copy full SHA for c4da71c

• [INTEGRATION] Bump llvm to 0f3ca6bb9 (#24390)

  Show description for fac9d3d
  lialan

  authored
  fac9d3d

  Copy full SHA for fac9d3d

• [ROCM] Workaround LLVM #194924 partial-unroll regression (#24379)

  Show description for f4fb944
  lialan

  authored
  f4fb944

  Copy full SHA for f4fb944

• [Codegen] Canonicalize transfer_{read,write} vector<1xT> (#24382)

  Show description for 1c14508
  amd-eochoalo

  authored
  1c14508

  Copy full SHA for 1c14508

Commits on May 6, 2026

• [INTEGRATION] Bump llvm to 3ed76d05a78d (#24376)

  Show description for 4030534
  lialan

  authored
  4030534

  Copy full SHA for 4030534

• [ROCm] Drop deprecated --iree-hip flag aliases (#24381)

  Show description for ac49ab6
  kuhar

  authored
  ac49ab6

  Copy full SHA for ac49ab6

• Reapply "[Codegen] Enable DMA by default for F16/BF16 Gemm on gfx950 (#24117)" (#24235) (#24373)

  Show description for 4f99043
  Yu-Zhewen

  authored
  4f99043

  Copy full SHA for 4f99043

• [LinalgExt] Fix attention NaN for fully-masked rows (#24178)

  Show description for 5fbfe29
  keshavvinayak01

  authored
  5fbfe29

  Copy full SHA for 5fbfe29

• Bump iree-org/torch-mlir@46cbd27f7c (#24380)

  Show description for d2f4f44
  keshavvinayak01

  authored
  d2f4f44

  Copy full SHA for d2f4f44

• [InputConversion] Lower AtenArgmax/AtenArgmin to iree_linalg_ext.arg_compare (#24291)

  Show description for 9092be0
  bangtianliu

  authored
  9092be0

  Copy full SHA for 9092be0

• Generalize FoldMaskedTransferRaw and add FoldTransferReadOfEmptyTensor (#24301)

  Show description for 6efc2ca
  amd-eochoalo

  authored
  6efc2ca

  Copy full SHA for 6efc2ca

• [VectorDistribute] Lower and distribute `async_dma` (#24299)

  Show description for ca7e063
  sommerlukas

  authored
  ca7e063

  Copy full SHA for ca7e063

• [Codegen] Duplicate operations in tile size analysis (#24246)

  Show description for c6525dd
  sommerlukas

  authored
  c6525dd

  Copy full SHA for c6525dd