I know there is native linux support for CUDA. That is why I am hoping this will not be to difficult. Currently the Folding @ Home GPU client is windows only. We are trying to get it running in linux via Wine. We have gotten to the point where it is trying to initialize a CUDAstream but it fails. Now I am assuming that a simply wrapper needs to be written in order to get the cudart.dll calls translated to libcudart.so.2.0 (or libcudart.so).
My question is, is there any documentation out there listing the difference between the windows and linux version? I’m going through the documentation now but if Nvidia has already told me the difference then it makes it a little easier. I know that the base of the code has to be similar so I know some of the calls are already support natively but I also know some of the other calls must be translated into the appropriate command for the linux lib.