Hi, Is there some difference between clEnqueueSVMMap for coarse SVM and clEnqueueMapBuffer in theory? If there is, will this cause difference about performance?
Hi,
The performance of these should be similar for Mali GPUs most of the time provided the buffers don't use CL_MEM_USE_HOST_PTR. clEnqueueSVMMap may be a bit faster in some cases (especially with smaller allocations).
Regards,
Kevin