Issues

• [clang-tidy] Fully qualify CUB and libcu++ function calls

  infra
  Enhancement

  NVIDIA/cccl#12624 · Filter by author Jacobfaib opened on Nov 12, 2026
  

• [FEA]: Support zip_iterator in the iterator unpacking facility

  NVIDIA/cccl#12620 · Filter by author bernhardmgruber opened on Nov 12, 2026
  

• [FEA]: Add representative benchmarks for fancy iterator unpacking in DeviceTransform

  NVIDIA/cccl#12619 · Filter by author bernhardmgruber opened on Nov 12, 2026
  

• [FEA]: Support iterator unpacking in ublkcp DeviceTransform

  NVIDIA/cccl#12618 · Filter by author bernhardmgruber opened on Nov 12, 2026
  

• [FEA]: Support iterator unpacking in LDGSTS DeviceTransform

  NVIDIA/cccl#12617 · Filter by author bernhardmgruber opened on Nov 12, 2026
  

• [FEA]: Support iterator unpacking in vectorized DeviceTransform

  NVIDIA/cccl#12616 · Filter by author bernhardmgruber opened on Nov 12, 2026
  

• [FEA]: Support iterator unpacking in prefetch DeviceTransform

  NVIDIA/cccl#12615 · Filter by author bernhardmgruber opened on Nov 12, 2026
  

• [BUG]: Correctness issues across the libcu++ random module

  Bug

  NVIDIA/cccl#12614 · Filter by author RAMitchell opened on Nov 12, 2026
  2

• [Task] Change API tests of segmened_scan algorithm to use cuda buffers

  NVIDIA/cccl#12581 · Filter by author oleksandr-pavlyk opened on Nov 10, 2026
  

• [Task] Change non-API segmented scan tests to use cuda buffers

  NVIDIA/cccl#12580 · Filter by author oleksandr-pavlyk opened on Nov 10, 2026
  

• Consider a version of memcpy_async that has no fallback to reduce register usage

  NVIDIA/cccl#12578 · Filter by author pciolkosz opened on Nov 10, 2026
  2

• Pin new CUB commit in FlashInfer that includes fix for +/-0.0 normalization and the commit recouping performance

  NVIDIA/cccl#12576 · Filter by author elstehle opened on Nov 10, 2026
  

• Task: Refactor two-phase launch implementation

  NVIDIA/cccl#12574 · Filter by author oleksandr-pavlyk opened on Nov 10, 2026
  

• [BUG]: fpmp2 math functions mishandle ±0, ±inf, NaN and out-of-domain inputs; spec tables echo the input

  needs triage
  NVIDIA/cccl#12550 · Filter by author rmlarsen opened on Nov 9, 2026
  3

  

• Align fpmp2_stat scalar copy-initialization with fpmp2

  NVIDIA/cccl#12542 · Filter by author coderabbitai opened on Nov 8, 2026

• [FEA]: Use LRB-based segmented sort as default untunable backend

  NVIDIA/cccl#12541 · Filter by author gevtushenko opened on Nov 8, 2026
  

• [FEA]: Microbenchmarks for CUB cooperative primitives

  NVIDIA/cccl#12540 · Filter by author gevtushenko opened on Nov 8, 2026
  1

  

• Spike on atomics, flags and barriers in inference

  NVIDIA/cccl#12539 · Filter by author gevtushenko opened on Nov 8, 2026
  

• Spike on device-resident problem sizes in inference

  NVIDIA/cccl#12538 · Filter by author gevtushenko opened on Nov 8, 2026
  

• Spike on TMA and bulk-copy abstractions

  NVIDIA/cccl#12537 · Filter by author gevtushenko opened on Nov 8, 2026
  

• [FEA]: Integrate fancy iterator unpacking facility in transform

  NVIDIA/cccl#12536 · Filter by author gevtushenko opened on Nov 8, 2026
  1

  

• Spike on compilation overhead and opportunities for improvement for device-side CCCL interfaces

  NVIDIA/cccl#12535 · Filter by author gevtushenko opened on Nov 8, 2026
  10

  

• [FEA]: Minimize group design for initial productization

  NVIDIA/cccl#12533 · Filter by author gevtushenko opened on Nov 8, 2026
  

• [FEA]: Make cuda.compute a default backend for CuPy

  NVIDIA/cccl#12520 · Filter by author gevtushenko opened on Nov 7, 2026
  

• [CUCO][FEA]: Unify detail::__index_type with size_type and drop __distance

  cucocudax
  NVIDIA/cccl#12511 · Filter by author PointKernel opened on Nov 6, 2026