Hi guys, i am new on OpenCL;
i have the following AMD GPU specifications:
7 compute units, each one with a wavefront of 64 work-items
max workgroup size = 256
max work-item size is (1024,1024,1024)
Local memory size = 32KB
Local memory size per CU = 64KB
In this scenario, can someone tell me how i can determine the maximum number of wavefronts that a single CU can manage contemporaneously?
is this value equal to
- Local memory size per CU/Local memory size = 2
or - max workgroup size/ 64 = 4
Thank you very much for the help