hi, I am a new OpenCL.
I am going to do a small test on my small GUP via OpenCL.
There are just 2 compute unit but 16 cores in my GUP. In other words, there are 8 cores in each compute unit.
In the small test, every cores would generate a random number at the same time so that there are 16 random in the end.
There is a problem. How to set the variants, global_work_offset, global_work_size and local_work_size, in the function, clEnquenueNDRangeKernel? Is the global_work_size mapping to the hardware compute unit? or Is the global_work_size mapping to the hardware single processor? Meanwhile, what is work-group here?