Is there any way to control the number of cores used concurrently for a multicore processor in OpenCL?
As we all know that CPU is a devcie and each core is a compute unit in OpenCL. I want to konw is there any way to controll the number of cores used concurrently in a kernel? In default, the current runtime implementation in SDK2.2 will use all the cores together.
I think this manner is hard for GPU, however, it is feasible for general purpose processor.