- high level wrapper for cublas in lisp, for example
(multiply A B)
- automatic deallocation of tensors
- automatic casting casting between the gpu and cpu depending on operation.
You will require the following in order to use this package:
- CUDA, version 8 and 9 seem to work fine.
- SBCL, maybe other implementations will work.
To get started, clone this repository into your Quicklisp
local-projects directory. Ensure that your CUDA lib folder is
available in the
LD_LIBRARY_PATH. Run the demos script with
This is a very young project and started mostly for fun. Contributions are absolutely welcome, just send a pull request if you want to add something. :)