All pull requests

Pull requests

• Unify the async host runtime and remove cuda_async::simt

  #1604·elibol opened Oct 21, 2026Collaborator·

  ·

• refactor(cuda-core): share host handles and allocations across Tile and SIMT

  #1603·elibol opened Oct 21, 2026Collaborator·

  ·

• fix(cutile): retain source origins for generated code and host checks

  #1602·elibol opened Oct 21, 2026Collaborator·

  ·

• feat(cuda-core): add CudaStream::borrow_raw for externally owned streams

  #1601·drbh opened Oct 21, 2026Contributor·

  ·
  1

• cutile: one CodegenMode for device debug, plus CUTILE_JIT_* run-time overrides

  #1600·almightychang opened Oct 21, 2026Contributor·

  ·
  2

• docs(cutile-rs): make the PointerTile examples in the DSL reference type-check

  #1599·ImanolGo opened Oct 21, 2026·1 of 5 tasks·

  ·
  1

  1

• Add the cute-rs preview: CuTe kernels in Rust for sm_100a and sm_120a

  codegencuda-featurecuda-oxidedocumentationenhancementintrinsics
  #1598·nihalpasham opened Oct 21, 2026Collaborator·

  ·
  1

• Accept cold-path hints in checked collective kernels

  #1593·tpn opened Oct 21, 2026·2 of 3 tasks·

  ·
  1

• Support checked payload collectives on warp prefixes

  #1592·tpn opened Oct 21, 2026·2 of 3 tasks·

  ·
  1

• fix(mir-lower): give an aggregate niche payload's leaves typed slots

  #1591·midagedev opened Oct 21, 2026Contributor·

  ·
  1

  1

• Support checked native collectives with local payloads

  codegencuda-oxideDependsdevice-apisenhancementinteropsafety
  #1589·tpn opened Oct 21, 2026·2 tasks done·

  ·
  1

• fix(cuda-macros): load kernels through a vector so cleanup stays linear

  #1588·sachinsharma3191 opened Oct 21, 2026Contributor·

  ·
  1

  2

• Element-generic cuTile kernels + compiler fix for callee-generic shadowing

  #1587·robertream opened Oct 21, 2026·

  ·
  1

• Plan independent scratch for native Warp collectives

  codegencuda-oxideDependsdevice-apisenhancementinteropsafety
  #1586·tpn opened Oct 21, 2026·

  ·
  1

• Add checked per-block scalar output witnesses

  cuda-oxideDependsdevice-apisenhancementhost-apissafety
  #1585·tpn opened Oct 21, 2026·

  ·
  1

• Support checked logical Warp collectives

  codegencuda-oxideDependsdevice-apisenhancementinteropsafety
  #1584·tpn opened Oct 21, 2026·

  ·
  1

• Index immutable collective call records by MIR site

  codegencuda-oxideDependsinteropperf
  #1583·tpn opened Oct 21, 2026·

  ·
  1

• Preserve inline source scopes in line-table debug info

  bugcodegencuda-oxidedebug infoDependsllvm-export
  #1582·tpn opened Oct 21, 2026·

  ·
  1

• Validate native collective access within physical warps

  codegencuda-oxideDependsdevice-apisenhancementinteropsafety
  #1581·tpn opened Oct 21, 2026·

  ·
  1

• ci: move ubuntu-latest jobs onto linux-amd64-cpu4

  #1580·roivanov opened Oct 20, 2026Collaborator·4 of 8 tasks·

  ·
  1

  3

• test(cuda-oxide): write fake tools through cp

  #1579·praveenperera opened Oct 20, 2026Contributor·

  ·
  1

• ci: add prek commit hooks

  #1578·assignUser opened Oct 20, 2026Contributor·2 tasks done·

  ·
  1

  1

• Initial pass at GPU testing for cutile-rs

  #1577·cryos opened Oct 20, 2026Collaborator·5 tasks·

  ·
  3

• feat(cuda-device): support u64 and i64 in cooperative_groups reductions and scans

  #1576·praveenperera opened Oct 20, 2026Contributor·

  ·
  1

  2

• Retain typed Rust callbacks for native device functors

  codegencuda-oxideDependsdevice-apisenhancementinterop
  #1575·tpn opened Oct 20, 2026·

  ·
  1