All pull requests

Pull requests

• [CUB] Add load balancing and a total-item-count input to DeviceSegmentedScan

  #11938·EzraPalmers opened Oct 8, 2026·2 tasks done·

  ·
  1

  4

• Add __builtin_constant_p to more functionalities

  #11936·fbusato opened Oct 8, 2026Contributor·

  ·
  2

  

• Apply clang-tidy misc-include-cleaner to cuda/std math functionalities

  #11934·fbusato opened Oct 7, 2026Contributor·

  ·
  4

  

• [libcu++]Refactor no-throw driver APIs to be actually no-throw

  #11933·pciolkosz opened Oct 7, 2026Contributor·

  ·
  4

• Decouple CI implementation from CCCL source branches

  #11932·chuckatkins opened Oct 7, 2026·

  ·
  1

  2

• Demonstrate CI workflow reuse from the ci branch

  #11931·chuckatkins opened Oct 7, 2026·

  ·
  1

• [CUB] Add fixed-size overloads to cub::DeviceSegmentedScan

  #11929·saiyambharara opened Oct 7, 2026·2 tasks done·

  ·
  1

  1

• [cuda.compute] Fix issue with missing inline in CTK 12.4

  #11928·NaderAlAwar opened Oct 7, 2026Contributor·2 tasks·

  ·
  1

  5

• [cub] Fix scan reduction for AccumT without identity element

  #11923·davebayer opened Oct 7, 2026Contributor·

  ·
  5

• Add cuda::is_arithmetic

  #11922·Jacobfaib opened Oct 7, 2026Contributor·2 tasks·

  ·
  8

  

• Ensure all headers differ in content to avoid GCC #pragma once collisions

  #11920·robertmaynard opened Oct 7, 2026Contributor·

  ·
  3

• [CUB] Use one item per thread when cub::DeviceTransform fills values over 512 bytes

  #11919·SamMausberg opened Oct 7, 2026Contributor·2 tasks done·

  ·
  1

  2

• [cub] Use item terminology consistently in DeviceHistogram

  #11918·robobryce opened Oct 7, 2026·

  ·
  1

• [CUB] Skip transposed loads of synthesizing iterators

  #11912·miscco opened Oct 7, 2026Contributor·

  ·
  14

• [libcu++] Implement strongly typed fabric operations

  #11908·pciolkosz opened Oct 6, 2026Contributor·

  ·
  30

• Reorganize mgmn into cuda/experimental/mgmn

  #11905·Jacobfaib opened Oct 6, 2026Contributor·2 tasks·

  ·
  5

  

• Add zero-copy release, acquire, and byte reinterpretation to cuda::buffer

  #11904·bdice opened Oct 6, 2026Contributor·

  ·
  17

• [cub] Use fast division for privatized histogram scaling

  #11901·robobryce opened Oct 6, 2026·

  ·
  13

• Keep -0.0 and +0.0 in order in Thrust's CPU stable_sort

  #11897·UnknownHacker1 opened Oct 6, 2026Contributor·

  ·
  1

  4

• [CUB] Consider whether iterators are synthesizing data

  #11896·miscco opened Oct 6, 2026Contributor·

  ·
  2

• Add cub::detail::DeviceSetOps pairs APIs

  #11891·bernhardmgruber opened Oct 6, 2026Contributor·

  ·
  9

• [DO NOT MERGE] Compile the lookahead scan at full width and with nv_atomic builtins under Circle

  #11890·caugonnet opened Oct 6, 2026Contributor·1 of 2 tasks·

  ·
  4

• [STF] exception_policy: catch a declined exception once and inline the composite dispatch

  #11889·caugonnet opened Oct 6, 2026Contributor·2 tasks done·

  ·
  4

• [STF] exception_policy: declare the answer type of | and * so nested alternations compile

  #11888·caugonnet opened Oct 6, 2026Contributor·2 tasks done·

  ·
  6

• Rename shared_barrier to shared_memory_barrier

  #11884·pciolkosz opened Oct 6, 2026Contributor·

  ·
  2