I have been reading over the multiGPU example and trying exparements with it. THe biggest problem I have not been able to solve is the data the actuall data I want to preform computations on has to be defined in the main program. When I tried to send a pointer to the data in the arguments for cudStartThread, it gave me an error message. How can I send more arguments to the CUT_THREADROUTINE method so that I can send the data?
CUDA Thread wrap around pthread. So if you want to use CUDA thread, you should understand pthread first. All mechanism that you use with Pthread, you can apply same thing with CUDA thread.
where would I find the documentation on pthread, I haven’t found it, or any documentation on CUDAthread either which is part of the problem.
Pthreads: google ‘posix threads’ or ‘posix threads programming’ or such. Also there are books on it that you should find in a decent technical bookstore or online. Stevens’ title “Advance Programming in the UNIX Environment” and his title “UNIX Network Programming, vol 1” both cover pthreads and are probably considered classics in the UNIX programming community.
CUDAThreads: no idea, first I heard of it.