How can i determine Registers per work-item and Shared Memory Per work-group? I know that is possible with CUDA, what about OpenCL?
Thanks