Specifies the grid and block dimensions for the device call to be executed similar to the execution configuration syntax. cudaConfigureCall() is stack based. Each call pushes data on top of an execution stack. This data contains the dimension for the grid and thread blocks, together with any arguments for the call.
|