Skip to Main Content
A grid-enabled programming toolkit called GridCuda is proposed in this paper. This programming toolkit provides a platform for users to write programs with the CUDA API, and exploit GPGPU resources available in computational grids to execute their programs. Whenever the CUDA functions in user programs are invoked, they will be transparently redirected to remote GPGPUs for execution by means of remote procedure calls. This programming toolkit also supports multithreaded programming so that the work of a CUDA program can be distributed over multiple GPGPUs for parallel execution.