It seems that MlOpen has been supporting OpenCL for some time now, and there appear to be libraries that use it more or less successfully.
I know that supporting so many libraries is quite a burden for you, and you'd probably be cautious before adding the support for another one. OTOH, deep learning is very important right now, and having a cuDNN alternative in OpenCL for the AMD platform might give JOCL a much needed appreciation from Java community.
Of course, I would like to help with that as much as I can. Although the tools and the process to create the initial support are only available to you, the least I can do is create bindings to my Clojure libraries, test it, and dig out as many bugs as I can.
Is there a chance that you'd consider supporting MIOpen?
Hi Marco,
It seems that MlOpen has been supporting OpenCL for some time now, and there appear to be libraries that use it more or less successfully.
I know that supporting so many libraries is quite a burden for you, and you'd probably be cautious before adding the support for another one. OTOH, deep learning is very important right now, and having a cuDNN alternative in OpenCL for the AMD platform might give JOCL a much needed appreciation from Java community.
Of course, I would like to help with that as much as I can. Although the tools and the process to create the initial support are only available to you, the least I can do is create bindings to my Clojure libraries, test it, and dig out as many bugs as I can.
Is there a chance that you'd consider supporting MIOpen?