Hi! I was wondering if there is a way to sync two threads, one on the cpu and one on the gpu.
I am creating to cpu cores, one of the cores will offload to the GPU side and will have a lot of computation there. But at some point I need to sync them together. Is there a way to do that sync, through a barrier or something, cause I saw that volatile does not work.
This question seems to be more related to development than it does performance tuning. I'll forward this to our developers and see if anyone has some insight to your particular problem :)
Did you get your question answered? If not, I would also recommend that you post this question in the following forum: http://software.intel.com/en-us/forums/user-community-for-visual-computing
As of now Intel compiler provides support only synchronous calls to GPU. So any computation offloaded to GPU is a blocked call and the CPU thread will only resume after the computation on GPU side completes and returns.
Thanks and Regards