1 Reply Latest reply on Jul 3, 2012 1:18 AM by Wenju

    CPU-GPU zero-copy shared-memory accesses while  GPU kernel is running

    silbmarks

      Is there any way  to have a write-shared (zero-copy) memory accessible from CPU and GPU at the same time, so that both  GPU and CPU would be able to issue memory fence DURING the GPU kernel execution?

       

      In other words, can I write an OpenCL code where a CPU  writes data and GPU  reads it consistently WITHOUT having to stop and start the GPU kernel?

      I didn't find any such functionality documented in OpenCL, but maybe on APUs it's possible with lower-level non-OpenCL APIs?

       

      Similarly, is there any way to make sure that  data transferred into the GPU memory via OpenCL memory transfer calls,  can be consistently read from the memory by a GPU kernel even if the transfer was done in parallel with the kernel execution? Again, on APUs it might work if the GPU snoops the memory bus. But does it?

       

      Any pointer/example/ would be very much appreciated!