All pull requests

Pull requests

• Handle unavailable logs in CI failure analysis

  #12348·jrhemstad opened Oct 30, 2026Collaborator·

  ·
  2

• [CUB] Unify Logical Warp/Lane Utilities

  #12347·fbusato opened Oct 30, 2026Contributor·

  ·
  9

  

• Require primitive types in cub::NumericTraits to be trivially copyable

  #12346·bernhardmgruber opened Oct 29, 2026Contributor·1 task·

  ·
  2

• Add agent reading links to documentation

  #12345·tpn opened Oct 29, 2026Contributor·2 tasks done·

  ·
  1

• [CUB] Fix CUB segmented reduce global id overflow

  backport branch/3.6.x
  #12344·wmaxey opened Oct 29, 2026Member·2 tasks·

  ·
  9

• [libcu++] Fix cuda::ilog10 stack memory

  backport branch/3.6.x
  #12338·fbusato opened Oct 29, 2026Contributor·

  ·
  1

  9

  

• Fully fix the host launch callback leak

  #12336·Jacobfaib opened Oct 29, 2026Contributor·2 tasks·

  ·
  5

  

• Enable __float128 in cuda::std::sqrt

  #12331·KassaSana opened Oct 29, 2026Contributor·

  ·
  1

  15

• [CUB] Refactor cub::DeviceSegmentedReduce two-phase overloads to take an environment

  #12330·bernhardmgruber opened Oct 29, 2026Contributor·

  ·
  1

  4

• [CUB] Refactor cub::DeviceSegmentedSort pairs two-phase overloads to take an environment

  #12329·bernhardmgruber opened Oct 29, 2026Contributor·

  ·
  1

  1

• [CUB] Refactor cub::DeviceSegmentedSort keys two-phase overloads to take an environment

  #12328·bernhardmgruber opened Oct 29, 2026Contributor·

  ·
  1

  1

• [CUB] Refactor BlockLoadDirectVectorized

  #12327·miscco opened Oct 29, 2026Contributor·

  ·
  9

• [CUB] Vectorize DeviceReduce loads for any contiguous iterator

  #12326·bernhardmgruber opened Oct 29, 2026Contributor·

  ·
  9

• [libcu++] Fix cuda::ipow

  backport branch/3.6.x
  #12324·fbusato opened Oct 29, 2026Contributor·

  ·
  1

  3

  

• [Backport branch/3.6.x] [libcu++] Fix cuda::make_tma_descriptor shape limit

  #12320·github-actions[bot] opened Oct 28, 2026Bot·

  ·
  4

• Use tested DeviceAdjacentDifference documentation examples

  #12319·1sgtpepper opened Oct 28, 2026·2 tasks done·

  ·
  1

  7

• [DO NOT MERGE] clang-tidy readability umbrella

  #12316·Jacobfaib opened Oct 28, 2026Contributor·2 tasks·

  ·
  12

  

• Add CUDA and libcu++ types to bugprone-unused-return-value types

  #12315·Jacobfaib opened Oct 28, 2026Contributor·2 tasks·

  ·
  17

  

• [CUB] Fix NVRTC < 13.2 compilation of WarpScan partial scans

  #12314·pylaterreur opened Oct 28, 2026·2 tasks done·

  ·
  1

  14

• Add standalone cuda.coop.cutlass package

  #12311·tpn opened Oct 28, 2026Contributor·

  ·
  1

• Prototype broadcast warp top-k for inference row selection

  #12310·tpn opened Oct 28, 2026Contributor·

  ·
  3

• Add experimental warp all-lane float32 sum

  #12309·tpn opened Oct 28, 2026Contributor·

  ·
  6

• Enable __float128 in cuda::std::rint, nearbyint, lrint and llrint

  #12308·KassaSana opened Oct 28, 2026Contributor·2 tasks done·

  ·
  4

  7

• [docs] Unpin breathe now that 5.1.0 fixes template specialization names

  #12307·KassaSana opened Oct 28, 2026Contributor·2 tasks done·

  ·
  1

  7

• Remove the existing cuda.coop implementation

  #12296·tpn opened Oct 27, 2026Contributor·

  ·
  11