NVIDIA / cub

[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Add policy parameter to allow tuning

gevtushenko opened this issue · comments

Not all CUB algorithm follow the scheme of accepting tuning policy at dispatch layer. This prevents these algorithms from being tuned.

Dispatch layers to be modified