Version française
Home     About     Download     Resources     Contact us    
Browse thread
daml - a CUDA binding
[ Home ] [ Index: by date | by threads ]
[ Search: ]

[ Message by date: previous | next ] [ Message in thread: previous | next ] [ Thread: previous | next ]
Date: -- (:)
From: Sashan Govender <sashang@g...>
Subject: daml - a CUDA binding
Hi

I've uploaded some work on a CUDA binding to OCaml  that I've been
doing in my spare time over here
https://forge.ocamlcore.org/projects/daml/. Here's a brief summary of
what you can do with it at the moment:

- allocate memory on a gpu
- deallocate that resource
- copy and integer array from OCaml onto the gpu
- execute a kernel function to process the data on the gpu
- copy the result back into the OCaml array.

There are a few example programs in the repository as well.

Pretty basic but gotta start somewhere and my time for free software is limited.

Things to todo:
- Devise a strategy for minimizing copies from the host data structure
to the gpu.
- Expose more of the CUDA runtime API (this can for the most part be
done with camlidl)
- Expose the CUDA driver API.
- Write some more example progams.
- Tidy up the OMakefiles.

Any feedback appreciated.

There's a mailing list here daml-users@lists.forge.ocamlcore.org for questions.

-- 
sashan
http://sashang.orcon.net.nz/