Issues · NVIDIA/cccl · GitHub

Labels Milestones New issue

[FEA]: Provide mappping that selects first `N` units from a group

#8520

· davebayer opened

on Apr 18, 2026

[FEA]: Implement `cudax::group_synchronizer_instance` concept

#8519

· davebayer opened

on Apr 18, 2026

[FEA]: Expose `cudax::group_mapping_result` concept

#8518

· davebayer opened

on Apr 18, 2026

[FEA]: Implement `cudax::lane_synchronizer` synchronizer

#8517

· davebayer opened

on Apr 18, 2026

[FEA]: Implement `cudax::barrier_synchronizer` synchronizer

#8516

· davebayer opened

on Apr 18, 2026

[FEA]: Implement `cudax::group_as` mapping

#8515

· davebayer opened

on Apr 18, 2026

[FEA]: Make `cudax::group` constructible from other groups

#8514

· davebayer opened

on Apr 18, 2026

[FEA]: Make `cudax::group` generic for all levels

#8513

· davebayer opened

on Apr 18, 2026

[CUB] Expose tuning parameters for `WarpReduceBatched`

#8512

· pauleonix opened

on Apr 18, 2026

[CUB] Add `WarpReduceBatched` benchmarks

#8511

· pauleonix opened

on Apr 18, 2026

[CUB] Add `Min` and `Max` member functions to `WarpReduceBatched`

#8510

· pauleonix opened

on Apr 18, 2026

[CUB] Vectorize shuffles in `WarpReduceBatched` for <4B types and blocked output

#8509

· pauleonix opened

on Apr 18, 2026