[FEA] Pass `cudaStreamPerThread` to numba/CuPy kernels
Open
feature request
Python
- Dominant language
- C++
- Stars
- 9.8k
- Forks
- 1.1k
- Avg merge
- 3d 6m
- Merged PRs (30d)
- 278
Description
Memory allocations should already use PTDS since both numba and CuPy allocate memory using RMM. Kernels on the other hand, may explicitly need to be passed the `cudaStreamPerThread` stream handle.
cc: @jakirkham @kkraus14
Contributor guide
Assessment
This issue has not been assessed yet.