Vyas Ramasubramani

icon indicating a website link www.vyasr.com icon indicating an email [email protected]

icon company @rapidsai icon location San Francisco, CA icon location @rapidsai software developer at @NVIDIA. PhD in Chemical Engineering and Scientific Computing from @glotzerlab at the University of Michigan

Results 914 comments of Vyas Ramasubramani

[FEA] Support stream semantics in rmm::mr::pool_memory_resource with rmm::mr::pinned_memory_resource as an upstream

rmm will be moving to use cudaMallocFromPoolAsync for different memory types, which would resolve this problem. I'm not sure if we have support for everything that we need to do...

[FEA] Support stream semantics in rmm::mr::pool_memory_resource with rmm::mr::pinned_memory_resource as an upstream

https://github.com/rapidsai/rmm/issues/2054#issuecomment-3339881396 describes a possible solution, it seems like we might have what we need already to do this in CUDA 12. UPDATE: [Not a viable solution for CUDA 12 in...

Document `cuda-libraries`; use of cudatoolkit in CUDA 12+ world

As Bradley said, I wrote those docs around CUDA 12.5 so that we would have a clear resource to point to given how many major changes were happening at the...

Migrate NVText Replacing APIs to pylibcudf

/ok to test