Repository navigation

support more than 2 warp groups - #9

Open

AutumnKite wants to merge 191 commits into

silentCoder-dev:auto-schedulefrom

AutumnKite:auto-schedule

Open

support more than 2 warp groups#9
AutumnKite wants to merge 191 commits into
silentCoder-dev:auto-schedulefrom
AutumnKite:auto-schedule

Commits on Mar 30, 2026
Commits on Mar 31, 2026
Commits on Apr 2, 2026
- [Feature] Support T.annotate_compile_flags, T.annotate_pass_configs, and out_idx as PrimFunc attrs (#2006)
  
  kurisu6912
  and
  claude
  authored
  e45ecf7
- [BugFix] Fix CI failures: clean /tmp on self-hosted runners and skip CuTeDSL alloc_global tests (#2009)
  
  kurisu6912
  and
  claude
  authored
  6e6295f
Commits on Apr 3, 2026
- [Test] Add 1D TMA regression test for issue #1842 (#2005)
  
  kurisu6912
  and
  claude
  authored
  5f70374
Commits on Apr 4, 2026
- [BugFix] Fix auto vectorization for binary operations after wider copy instructions (#1986)
  
  authored
  6fc3afa
- fix: add cudaGetLastError check after cuLaunchKernel in TVM FFI backend (#2000)
  kurisu6912
  authored
  01c714d
Commits on Apr 5, 2026
- [CI] Remove legacy dequantize gemm test (#2013)
  LeiWang1999
  authored
  bb79425
Commits on Apr 6, 2026
- [CI] [pre-commit.ci] autoupdate (#2014)
  pre-commit-ci[bot]
  authored
  1ff58bc
Commits on Apr 7, 2026
Commits on Apr 8, 2026
- remove wg_wait in gemm_auto_tcgen5mma.py
  AutumnKite
  committed
  39adf6a
- support more than 2 warp groups
  AutumnKite
  committed
  0402120
- change `before` and `after` to map
  AutumnKite
  committed
  b021c2a
- Bump transformers from 4.53.0 to 5.0.0rc3 in /examples/bitnet-1.58b (#2021)
  dependabot[bot]
  authored
  1cb8fde
- pin apache-tvm-ffi<0.1.10 (derived_object regression) (#2020)
  oraluben
  authored
  469a847
- Fix serial loop phase dtype mismatch in LowerTileOp (#2022)
  LeiWang1999
  authored
  86e37b7
- Fix typo
  
  silentCoder-dev
  and
  coderabbitai[bot]
  authored
  9a15696
- Merge branch 'auto-schedule' of https://github.com/silentCoder-dev/tilelang into auto-schedule
  AutumnKite
  committed
  7668827
Commits on Apr 9, 2026
Commits on Apr 10, 2026
- fix barrier logic
  AutumnKite
  committed
  101a672
- run format
  AutumnKite
  committed
  35f13d2
- Re-enable deprecated `TL_DISABLE_TMA_LOWER` pass config for TMA store (#2024)
  
  LJC00118
  and
  wangxiangwen
  authored
  a5e3f19
- [Misc] Remove mistakenly introduced temp file (#2027)
  SiriusNEO
  authored
  b1a88bf
- Fix warpgroup partition
  Denverjin
  committed
  9b4e0a2
- Merge pull request #1 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  4db4433
- fix barrier logic
  AutumnKite
  committed
  b31aaa9
- Merge branch 'auto-schedule' of https://github.com/silentCoder-dev/tilelang into auto-schedule
  AutumnKite
  committed
  7be6332
Commits on Apr 11, 2026
Commits on Apr 12, 2026
- [AMD][gfx950] Add gfx950 support for DeepGeem example (#2028)
  
  zhangnju
  and
  LeiWang1999
  authored
  7a515b5
Commits on Apr 13, 2026
Commits on Apr 14, 2026
Commits on Apr 15, 2026
- Support local var fragment
  Denverjin
  committed
  380fb5e
- Merge pull request #2 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  39c3c06
- [Feature] autodd: add __freeze__ annotation to protect code regions from reduction (#2045)
  kurisu6912
  authored
  a16ff86
- [BugFix] Skip MMA shared buffer layout inference when layout already exists (#2008)
  
  authored
  d2e02e1
- support tcgen05_gemm
  AutumnKite
  committed
  c93778f
- [Refactor] Remove obsolete RewriteWgmmaSync pass (#2046)
  LeiWang1999
  authored
  7893678
- Add if node
  Denverjin
  committed
  e3d214d
- Merge pull request #3 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  011d880
- remove debug info
  AutumnKite
  committed
  3c7f4a0
- run format
  AutumnKite
  committed
  db3d859
- [Refactor] Move target gating into InjectFenceProxy pass entry (#2047)
  LeiWang1999
  authored
  f12fb47
- fix loop break detection
  Denverjin
  committed
  7ae8d98
- fix control node break & format
  Denverjin
  committed
  dbadd77
- Merge pull request #4 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  510d773
- Fix if task collection bug
  Denverjin
  committed
  7ed8266
- Add regression test for 1D TMA load compilation and execution (#1989)
  
  huyhoang171106
  and
  LeiWang1999
  authored
  235ad7e
Commits on Apr 16, 2026
- Merge pull request #5 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  9e78312
- [Transform] Add InjectTcgen05Fence pass (#2003)
  
  LeiWang1999
  and
  kurisu6912
  authored
  891109e
- fix loop break
  AutumnKite
  committed
  8f67446
- fix tma load detection
  Denverjin
  committed
  844d04e
- fix pro/epilogue identification
  Denverjin
  committed
  45f5da4
- remove debug output & format
  Denverjin
  committed
  9970ac9
- reimplement barrier logic
  AutumnKite
  committed
  1660c85
- Merge branch 'auto-schedule' of https://github.com/AutumnKite/tilelang into auto-schedule-myh
  Denverjin
  committed
  25b0eb8
- partly fix pro/epilogue logic for barrier
  Denverjin
  committed
  37f05ee
- Merge pull request #6 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  e84cee9
- refactor ir structure clone
  Denverjin
  committed
  a4aa5f3
- fix bug
  Denverjin
  committed
  33e5c36
- Merge branch 'auto-schedule' of https://github.com/AutumnKite/tilelang into auto-schedule-myh
  Denverjin
  committed
  cee084c
- Merge pull request #7 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  36f51af
- fix bug
  AutumnKite
  committed
  7d1e6e3
- [Enhancement] Use atomic directory rename for cache writes (#1982)
  
  LeiWang1999
  and
  kurisu6912
  authored
  d098f51
Commits on Apr 17, 2026
- refactor shared memory buffer merge
  Denverjin
  committed
  a70fa26
- Merge pull request #8 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  561adfe
- upload latency & ii
  Denverjin
  committed
  189c99f
- Merge pull request #9 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  68f4710
- Replace syntactic loop-var checks with invariance checks (#2050)
  LJC00118
  authored
  cb00a60
- [Feature][Example] Introduce CLC tile schedule and add example for sm100 GEMM (#2029)
  
  authored
  27f1f81
- [Feature] Introduce T.CUDASourceCodeKernel (#1970)
  SiriusNEO
  authored
  77cbe6d
- run format
  AutumnKite
  committed
  aa0700b
- fix II of IfNode
  AutumnKite
  committed
  72d7748
- fix barrier
  AutumnKite
  committed
  4bf8678
- [BugFix] Keep shared-prelude local vars in producer-consumer WS (#2055)
  Rachmanino
  authored
  9c5fe44
- [Bugfix] Fix stage-expanded annotated-layout aliases in LayoutInference (#2031)
  
  TerminusAkivili
  and
  LeiWang1999
  authored
  04468a3
- fix pro/epilogue let stmt copy
  Denverjin
  committed
  6364f5d
- fix z3 small n error
  Denverjin
  committed
  557bcc1
- fix double kernel issue
  Denverjin
  committed
  3bc1c01
- format
  Denverjin
  committed
  105b73b
- undo failed merge
  Denverjin
  committed
  b18c60a
- Merge pull request #10 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  6ca1415
- check dependency in prologue
  AutumnKite
  committed
  aa877ab
- Merge commit '891109ea879839bbfe2ad937db208f2d7c6f1ce0' into auto-schedule-myh
  Denverjin
  committed
  b2abf0a
- fix header missing
  Denverjin
  committed
  c96dd9e
Commits on Apr 18, 2026
- [Cache] Refactor cache namespace layout (#2057)
  LeiWang1999
  authored
  e1d6388
- [Bugfix] Use shared::cta instead of shared::cluster for non-cluster T… (#2052)
  
  qqq-tao
  and
  qutao
  authored
  0924dab
Commits on Apr 20, 2026
Commits on Apr 21, 2026
- Add frontmatter for the build skill (#2068)
  VitalyAnkh
  authored
  96c649f
- Refactor ptx_ldmatrix to use tl.access_ptr with simplified signature (#2072)
  
  LeiWang1999
  and
  claude
  authored
  b744da1
- [FFI] Remove upper version bound on apache-tvm-ffi (#2071)
  
  LeiWang1999
  and
  claude
  authored
  046b1bd
- [Refactor] Phaseout legacy util `map_torch_type` with `T.dtype.as_torch` (#2075)
  LeiWang1999
  authored
  9c95a42
- [Bugfix] Fix reduce layout (#2074)
  bucket-xv
  authored
  64bd742
- [Refactor] Disable unhelpful warning print (#2077)
  LeiWang1999
  authored
  948d38a
- [CUDA] Improve int4 GEMM lowering and packed codegen support (#2073)
  LeiWang1999
  authored
  15309f5
- Bump pytest --numprocesses from 4 to 8 across all platforms (#2076)
  
  LeiWang1999
  and
  claude
  authored
  4e7d126
Commits on Apr 22, 2026
- fix dead-lock bug
  AutumnKite
  committed
  38cac96
- [Enhancement] Enhance alloc_var function to handle _ptr_sentinel dtype (#2078)
  LeiWang1999
  authored
  4b3127a
- [Release] Bump version into 0.1.9 (#2060)
  LeiWang1999
  authored
  6fee850
- fix register fragment reuse
  Denverjin
  committed
  1f831f3
- Merge branch 'auto-schedule' of https://github.com/AutumnKite/tilelang into auto-schedule-myh
  Denverjin
  committed
  cac434c
- Merge pull request #11 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  3c3cf53
- [Refactor] Strip build machine paths from LOG messages in wheel releases (#2080)
  
  LeiWang1999
  and
  claude
  authored
  441c3b0
- [AMD][Radeon] Add the Support of RDNA3/RDNA3.5(gfx11) WMMA (#2044)
  jiawei-real
  authored
  a640a89
- fix let stmt clone bug
  Denverjin
  committed
  3aeb962
- add the innermost task to sync infos
  AutumnKite
  committed
  4eed399
- [codex] Remove dead transform pass leftovers (#2083)
  LeiWang1999
  authored
  55cf9c1
- [Bugfix] Enable `.shared::cta` in TMA copy paths only on CUDA 12.8+ (#2087)
  ColmaLiu
  authored
  9aba41f
Commits on Apr 23, 2026
- fix hopper neutral stage
  Denverjin
  committed
  b6e75b1
- change layout map & remove unused letstmt
  Denverjin
  committed
  6e2fb56
- disable auto scheduling when using thread vars
  Denverjin
  committed
  89e6812
- Merge pull request #12 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  be0df30
- find first/last tasks of a buffer access and reduce syncs by checking the type of dependency
  AutumnKite
  committed
  f3f6e74
- Merge branch 'auto-schedule' of https://github.com/AutumnKite/tilelang into auto-schedule
  AutumnKite
  committed
  59c9b05
- format
  AutumnKite
  committed
  bdb20c2
- fix attr warp partition
  Denverjin
  committed
  ff067b0
- [AMD][gfx950] Add ds_read_tr16_b64 / ds_read_tr8_b64 support for gfx950 LDS transpose reads (#2085)
  zhangnju
  authored
  b88bbcb
- fix let missing bug
  Denverjin
  committed
  95f1d29
- Merge pull request #13 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  65738cd
- [AMD][Gfx950] Add the support of 160K LDS and copy.async (#2058)
  
  zhangnju
  and
  LeiWang1999
  authored
  6d0bffb
- add double-thread constraint
  
  AutumnKite
  and
  Copilot
  committed
  10b7f1f
- Merge branch 'auto-schedule' of https://github.com/AutumnKite/tilelang into auto-schedule
  AutumnKite
  committed
  4cec4ba
- fix local var copy
  Denverjin
  committed
  ef6a431
- [BugFix] Relax loop wait and adjust trailing drain behavior in async pipeline tests (#2092)
  
  Rachmanino
  and
  wutong.1109
  authored
  0f29f9c
Commits on Apr 24, 2026
- fix naive ir structure bug
  Denverjin
  committed
  e99d35a
- Merge pull request #14 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  2ca5e06
- format
  AutumnKite
  committed
  86989d7
- fix: remove unused let
  AutumnKite
  committed
  a552916
- remove redundant letstmts
  Denverjin
  committed
  c18c623
- Merge branch 'auto-schedule' of https://github.com/AutumnKite/tilelang into auto-schedule-myh
  Denverjin
  committed
  55cbe0e
- move the rewrites forward
  
  AutumnKite
  and
  Copilot
  committed
  0edb76c
- add WAW dependence & avoid duplicated dependence when iter=1
  
  AutumnKite
  and
  Copilot
  committed
  09d9071
- fix barrier around let missing
  Denverjin
  committed
  9897fe5
- Merge branch 'auto-schedule' of https://github.com/AutumnKite/tilelang into auto-schedule-myh
  Denverjin
  committed
  3c5578d
- Merge pull request #15 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  b46d709
- remove cross-warpgroup dependency for register buffers
  AutumnKite
  committed
  01bf798
- fix reused buffer analysis
  Denverjin
  committed
  d255c0a
- check kernel using barrier & format
  Denverjin
  committed
  3df8b46
- Merge pull request #16 from Denverjin/auto-schedule-myh
  AutumnKite
  authored
  468d51b
- format
  AutumnKite
  committed
  e18b8e6
- [Feature] Block-scaled GEMM support for MXFP8 on Blackwell (#1945)
  
  Rachmanino
  and
  LeiWang1999
  authored
  264efe2
Commits on Apr 25, 2026
- [Host CodeGen][Refactor] Cleanup namespace and remove useless C templates (#2091)
  SiriusNEO
  authored
  057e5ba
- Add opt-out for prelower semantic checks for DeepSeek V4 Flash on ARM64 (#2094)
  
  foraxe
  and
  SiriusNEO
  authored
  3f16e50
- [Example] Add HISA: hierarchical sparse attention indexer (#2069)
  
  xuyufei-a
  and
  yufeixu
  authored
  0ee6345
- [Language] Small cleanup and notes for alloc global (#2100)
  SiriusNEO
  authored
  8f4a08f
Commits on Apr 26, 2026
- [Enhancement] Optimize hopper fp8 deepgemm tile size (#2103)
  Rachmanino
  authored
  8e12157
- [CUDA][SM100] Include cuda_fp6.h when emitting FP6 types (#2102)
  TerminusAkivili
  authored
  ffdf514
- feat: support cdna4 v_mfma_i32_16x16x64_i8 & v_mfma_i32_32x32x32_i8 (#2097)
  
  Paran0idy
  and
  Jiaxing Ding
  authored
  6a29c76
- [AMD] [gfx950]Fix multiple HIP codegen bugs to support TileKernel (#2099)
  zhangnju
  authored
  53a4c98
Commits on Apr 27, 2026
- Merge remote-tracking branch 'tilelang/main' into auto-schedule-myh
  Denverjin
  committed
  73a54d2
- fix T.make_tensor buffer missing
  Denverjin
  committed
  a2f6a45
Commits on May 6, 2026
Commits on May 7, 2026
- Fix let & barrier bugs
  Denverjin
  committed
  28aa5ad
- Merge pull request #18 from Denverjin/auto-schedule
  AutumnKite
  authored
  b29d135
- add top-level barriers
  AutumnKite
  committed
  10503ba
- format
  AutumnKite
  committed
  f62744a
Commits on May 8, 2026
- fix empty for bug
  AutumnKite
  committed
  4f8823b

Commits

Commits on Mar 30, 2026

Commits on Mar 31, 2026

Commits on Apr 2, 2026

Commits on Apr 3, 2026

Commits on Apr 4, 2026

Commits on Apr 5, 2026

Commits on Apr 6, 2026

Commits on Apr 7, 2026

Commits on Apr 8, 2026

Commits on Apr 9, 2026

Commits on Apr 10, 2026

Commits on Apr 11, 2026

Commits on Apr 12, 2026

Commits on Apr 13, 2026

Commits on Apr 14, 2026

Commits on Apr 15, 2026

Commits on Apr 16, 2026

Commits on Apr 17, 2026

Commits on Apr 18, 2026

Commits on Apr 20, 2026

Commits on Apr 21, 2026