Open andronat opened 4 months ago
Any plans for supporting CUDA's Unified memory API?
Assuming you are referring to cuMemAllocManaged https://docs.nvidia.com/cuda/cuda-driver-api/group__CUDA__MEM.html#group__CUDA__MEM_1gb347ded34dc326af404aa02af5388a32?
cuMemAllocManaged
We can add definitely something for that
Any plans for supporting CUDA's Unified memory API?