All pull requests

Pull requests

• Check KernelAbstractions for Core.Boxes, and remove them

  #903·giordano opened Oct 16, 2026Collaborator·

  ·
  2

• Keep the argument-independent launch work out of per-kernel code

  #900·maleadt opened Oct 16, 2026Member·

  ·
  2

• Track KernelAbstractions coverage in the back-end jobs

  #899·maleadt opened Oct 16, 2026Member·

  ·
  2

• Give @kernel's device function the location of the kernel

  #898·maleadt opened Oct 16, 2026Member·

  ·
  1

  3

• POCL: fuse muladd into fma on devices with a fused multiply-add

  #892·vchuravy opened Oct 13, 2026Member·

  ·
  1

• Add an optional graph capture interface

  #886·vchuravy opened Oct 12, 2026Member·

  ·
  4

  

• Grid-stride launches: transparently past the grid limits, and on request with gridstride

  #884·vchuravy opened Oct 12, 2026Member·

  ·
  3

• Recognize a qualified @Const in a kernel's arguments

  #880·vchuravy opened Oct 12, 2026Member·

  ·
  1

  1

• POCL: enable SPV_KHR_float_controls2 when the device accepts it

  #876·vchuravy opened Oct 11, 2026Member·

  ·
  2

• DGSEM benchmarks: variants with tiles (#864)

  #874·vchuravy opened Oct 11, 2026Member·5/5·

  ·
  2

• Add tiles

  #864·maleadt opened Oct 10, 2026Member·3/3·

  ·
  9

• Run full workgroups of kernels without @synchronize unmasked on the CPU

  #862·vchuravy opened Oct 10, 2026Member·

  ·
  1

  2

• Time kernels on the device without synchronizing (prototype)

  #837·vchuravy opened Oct 4, 2026Member·2/2·

  ·
  2

• Add a low-cost tracing subsystem

  #836·vchuravy opened Oct 4, 2026Member·1/2·

  ·
  4

• Enzyme support for KernelAbstractions 0.10

  #826·vchuravy opened Oct 3, 2026Member·

  ·

  · 0.10.1
  4

• Support Active kernel arguments on GPU in the Enzyme extension

  #806·vchuravy opened Sep 28, 2026Member·

  ·
  1

• Document the contract of custom NDRange mappings

  documentation
  #781·giordano opened Sep 14, 2026Collaborator·

  ·
  5

• attempt to force fastmath at the kernel level

  #550·leios opened Dec 14, 2024Contributor·

  ·
  10

• Use MultiplicativeInverse to speedup Linear to Cartesian indexing operations

  #539·vchuravy opened Oct 18, 2024Member·

  ·
  5

• RFC: Support Adapt.AbstractGPUDevice

  #297·oschulz opened May 22, 2022Collaborator·

  ·
  1