Multiple context synchronization

Discussion created by dittamo on Jul 5, 2008


Data sharing across contexts is possible by mapping the same resource into multiple contexts. Synchronization of multiple contexts is the client’s responsibility, any support is provided for shared-memory access: are there thread-safe CAL\IL instructions ? are there atomic CAL \ IL instructions such as try&set or atomic swap? 

In order to limit communication between local and remote memory and perform as much calculus as possible on the GPU, I think it should be very important to have such support.