All pull requests

Pull requests

• [KMCompiler][Ascend] Lane vectorize pass

  ascendCOREtriton_v3.5.x
  #1306·zhongsanming opened Sep 29, 2026Contributor·

  ·
  4

• [BACKEND] Fix packed rotary output offsets

  cambriconCOREtriton_v3.2.x
  #1305·Cross2pro opened Sep 29, 2026·

  ·
  1

• [TLE] Add inter-node all-gather tutorial

  COREmaintle
  #1304·git-flyer opened Sep 29, 2026Collaborator·

  ·

• [CI] Add logic to check the pid.txt

  CI/CDhcuiluvatarmainmetaxmthreadsnvidia
  #1303·ShawnLiuxinghong opened Sep 29, 2026Collaborator·

  ·

• [DOC] Document tle.dsa.ascend.raw custom op interface

  DOCtletriton_v3.5.x
  #1301·togepro opened Sep 28, 2026Collaborator·

  ·
  1

• [MThreads][TLE] Add persistent GEMM pipeline and SPMC synchronization support

  COREmainmthreadstle
  #1294·Zhang-kg opened Sep 27, 2026Contributor·

  ·

• [SPEC][ILUVATAR] Specialize C++ sources

  COREDOCiluvatarmain
  #1293·Stardep-lmc opened Sep 25, 2026Collaborator·

  ·

• [HCU] Fix volatile load handling during LLVM lowering

  hcumain
  #1288·ChunhaoSun opened Sep 24, 2026Collaborator·

  ·
  1

  1

• [TLE] Add reuse_default_warps and extend extract_tile across warp counts

  COREmainnvidiatle
  #1286·kuoihao opened Sep 24, 2026Collaborator·

  ·

• [TLE] Inline DSL regions in warp-specialize bodies

  COREmaintle
  #1278·Zhang-kg opened Sep 23, 2026Contributor·

  ·

• [TLE][Ascend][XCS] Add tle.pipe for ascend

  COREtletriton_v3.5.x
  #1275·finneyyan opened Sep 23, 2026Collaborator·

  ·
  1

• [TLE][Ascend][XCS] add reduce sum strength

  ascendCI/CDCOREtletriton_v3.5.x
  #1273·huanghaoXcore opened Sep 23, 2026Collaborator·

  ·

• [QC][TLE][MTHREADS] Skip shared scratch for static CTA-aligned extract_tile

  mainmthreadstle
  #1268·monellz opened Sep 22, 2026Contributor·

  ·
  3

• [QC][TLE][MTHREADS] Fix SQMMA lowering, layout inference, shared addressing and partition sync on PH1

  mainmthreadstle
  #1267·monellz opened Sep 22, 2026Contributor·

  ·

• Nvidia dev 202609

  COREmainnvidiatle
  #1262·Jacob-yen opened Sep 22, 2026Contributor·

  ·
  7

• [MTHREADS] Update libdevice on musa

  mainmthreads
  #1261·Kylin1207 opened Sep 21, 2026Contributor·

  ·
  1

• [FlagPrism][Enflame] Integrate profiler and debugger

  COREenflamemain
  #1239·ChijinZ opened Sep 20, 2026·

  ·

• [TLE][Ascend][XCS]: Rename from tle.dsa.ascend.raw to tle.raw

  COREDOCtletriton_v3.5.x
  #1238·finneyyan opened Sep 20, 2026Collaborator·

  ·
  2

• [AUTOTUNER][PPU] Support T-Head MM model package v1.1.0

  COREmainppu
  #1236·HenryRenYz opened Sep 20, 2026Contributor·

  ·

• [BUILD] Declare cann-shmem as an Ascend wheel dependency

  COREtriton_v3.5.x
  #1230·tengqm opened Sep 20, 2026Collaborator·

  ·

• [PPU][MTHREADS] Update integer dot and dot_scaled support via backend capability contract

  COREmainmthreadspputle
  #1189·zeroherolin opened Sep 16, 2026Collaborator·

  ·
  1

• [CI][hcu] Add directory matching logic for checking PID

  CI/CDhcumain
  #1171·ShawnLiuxinghong opened Sep 14, 2026Collaborator·

  ·

• [QC][Nvidia] Barrier & Sync Optimization

  COREhcumainnvidiatle
  #1166·monellz opened Sep 14, 2026Contributor·

  ·
  8

• Fea/tiled smem

  COREhcumainnvidiatle
  #1161·Kafka-Hatsune opened Sep 14, 2026Contributor·

  ·
  4

• [XPU] Sample wrap tiles in OffsetAnalysis to fix a stride-0 view copy

  mainxpu
  #1147·hulong02-klx opened Sep 10, 2026Contributor·

  ·
  1