All pull requests

Pull requests

• Compile an actually valid module in the NVVM "bad option" test

  #388·brandon-b-miller opened Oct 14, 2026Collaborator·

  ·
  4

  

• Fix debug metadata for scalar values

  #387·brandon-b-miller opened Oct 14, 2026Collaborator·

  ·
  4

  

• Add experimental scalar Host JIT execution

  #386·atmnp opened Oct 14, 2026Collaborator·

  ·
  2

• Extract MLIR package

  #383·brandon-b-miller opened Oct 12, 2026Collaborator·

  ·
  1

• Use cuda.pathfinder to locate nvdisasm+cuobjdump

  #380·Jlisowskyy opened Oct 12, 2026Collaborator·

  ·
  11

• Preserve volatile loads in LLVM70 translation

  #379·1sgtpepper opened Oct 10, 2026·

  ·
  1

  2

• Fix Array.strides lowering: crash on every array, and wrong values for non-C layouts

  #378·r-u-human-xyz opened Oct 10, 2026·

  ·
  2

• Lower integer constants of 2**63 or more as their two's complement

  #374·adimalkar opened Oct 9, 2026Contributor·

  ·
  1

  3

• Fix record array addressing through offset and strided views

  #365·atmnp opened Oct 5, 2026Collaborator·

  ·
  1

  8

• Fix unary minus and plus on arrays in kernels

  #364·pylaterreur opened Oct 3, 2026Contributor·

  ·
  1

  2

• Lower np.left_shift and np.right_shift

  #363·pylaterreur opened Oct 3, 2026Contributor·

  ·
  1

  9

• Handle NaN in maximum, minimum, fmax and fmin as NumPy does

  #358·pylaterreur opened Oct 3, 2026Contributor·

  ·
  1

  2

• Lower NumPy ufuncs called on scalars

  #357·pylaterreur opened Oct 3, 2026Contributor·

  ·
  1

  8

• Plug launcher/dispatcher memory leaks

  #333·ccam80 opened Sep 24, 2026Contributor·

  ·
  7

• Math functions return the float type they compute

  #332·ccam80 opened Sep 24, 2026Contributor·

  ·
  5

• Bind keyword-only arguments in device function calls

  #331·ccam80 opened Sep 24, 2026Contributor·

  ·
  13

• [Py315] Basic Support for Python 3.15

  #315·atmnp opened Sep 14, 2026Collaborator·

  ·

  · numba-cuda-mlir 0.6.0
  9

  

• Keep dispatcher compilation and launch state specific to each CUDA context

  #314·tpn opened Sep 14, 2026Contributor·

  ·
  26

• Separate dynamic shared memory from static allocations

  #312·tpn opened Sep 14, 2026Contributor·

  ·
  7

• Add cuda.unroll and cuda.nounroll loop unroll hints

  #298·ccam80 opened Sep 3, 2026Contributor·

  ·
  7

• Apply AST transforms to inlined device functions

  #292·ccam80 opened Sep 3, 2026Contributor·

  ·
  13

• Bump the actions-monthly group across 1 directory with 10 updates

  dependenciesgithub_actions
  #288·dependabot[bot] opened Sep 1, 2026Bot·

  ·
  1

• Normalize slices and allow dynamic shared mem views

  #263·ccam80 opened Aug 12, 2026Contributor·

  ·
  1

  4

• Accepts individual fastmath flags for LLVM flags + ftz

  #255·ccam80 opened Aug 7, 2026Contributor·

  ·
  1

  7

• Support structured extension arguments and local arrays

  #254·tpn opened Aug 7, 2026Contributor·

  ·

  · numba-cuda-mlir 0.6.0
  2