All pull requests

Pull requests

• feat(fuzzer): generate Static and StaticMut with Miri-checked provenance

  #1525·uurl opened Oct 16, 2026Contributor·

  ·
  1

• fix(cuda-async): require unsafe kernel argument setup

  #1524·efegokdemir opened Oct 16, 2026·

  ·
  1

• fix(cutile-rs): keep every launch-site specialization

  #1521·rdeaton opened Oct 15, 2026·4 tasks done·

  ·
  1

• fix(cutile-rs): reject a const_grid with a zero dimension

  #1520·rdeaton opened Oct 15, 2026·4 tasks done·

  ·
  1

• cuda-core: safe element casts require AnyBitPattern; bool is DeviceCopy

  cuda-oxidecutile-rsruntimesafety
  #1516·elibol opened Oct 13, 2026Collaborator·

  ·
  1

  3

• fix: admit scalar arithmetic on verified pre-Ampere targets

  cuda-oxide
  #1512·ssoj13 opened Oct 13, 2026·

  ·
  1

• feat(cuda-oxide): carry inline intent through the device pipeline

  cuda-oxide
  #1511·ssoj13 opened Oct 13, 2026·

  ·
  1

• fix(cargo-oxide): pin release-like codegen in Cargo profiles

  cuda-oxide
  #1510·ssoj13 opened Oct 13, 2026·

  ·
  1

  1

• fix(cuda-oxide): keep host-written constant memory through LTO

  cuda-oxide
  #1509·ssoj13 opened Oct 13, 2026·

  ·
  1

  1

• docs(examples): move atomics to the typed-slice sound form

  cuda-oxide
  #1500·navneet83 opened Oct 12, 2026Contributor·

  ·
  1

• fix(codegen): keep the artifact objects of rlibs out of the LTO input

  cuda-oxide
  #1496·fxeqxmulfx opened Oct 11, 2026·4 of 5 tasks·

  ·
  1

  1

• docs(cuda-oxide): state what sync_threads orders at compile time

  cuda-oxidedevice-apisdocumentationneeds-changes
  #1493·alejandro-soto-franco opened Oct 11, 2026Contributor·

  ·
  1

  3

• fix(cuda-core): return an error on interior NUL in module loading

  #1492·alejandro-soto-franco opened Oct 11, 2026Contributor·

  ·
  1

• fix: return jit errors for malformed cutile dimensions

  #1486·uurl opened Oct 11, 2026Contributor·

  ·
  1

  1

• fix(intrinsics): admit FP16 m16n8k8 MMA on sm_75

  cuda-oxide
  #1484·praveenperera opened Oct 11, 2026·9 tasks done·

  ·
  1

  1

• fix(cutile): attribute generated entry setup to the user signature

  #1480·elibol opened Oct 10, 2026Collaborator·

  ·
  1

• build: add CUDA 13.4 and 13.x versions to default candidates

  #1474·temotskipa opened Oct 10, 2026·2 of 4 tasks·

  ·
  1

• fix: standardize non-negative enum types for windows msvc

  #1473·temotskipa opened Oct 10, 2026·2 tasks done·

  ·
  1

  1

• fix(cuda-macros): obey #[inline] on generic device functions

  cuda-oxide
  #1472·fxeqxmulfx opened Oct 10, 2026·4 of 5 tasks·

  ·
  1

  1

• fix(cutile): preserve boolean and unsigned literal values

  #1470·filipkunc opened Oct 10, 2026·2 tasks done·

  ·
  1

  1

• fix: keep symbolic shape on reduce results

  #1456·Arthur031221 opened Oct 8, 2026·

  ·
  1

  1

• fix(cutile): validate full tensor shapes before allocation

  #1452·uurl opened Oct 8, 2026Contributor·

  ·
  1

  1

• fix: validate runtime tensor strides before kernel launch

  #1450·uurl opened Oct 8, 2026Contributor·

  ·
  1

  1

• fix(cutile-rs): reject overflowing section lengths and varints in bytecode decoder

  #1448·dawnop opened Oct 8, 2026·

  ·
  1

  1

• feat(cuda-core): add borrowed DeviceBuffer subviews

  runtime
  #1447·uurl opened Oct 8, 2026Contributor·4 tasks done·

  ·
  1