16 Replies Latest reply on Dec 10, 2012 9:13 AM by lbin

    Is GPU_ASYNC_MEM_COPY=2 still available in SDK 2.7?

    levyfan

      Is GPU_ASYNC_MEM_COPY=2 still available in 2.7?

       

      I enqueue a kernel and a buffer-reading at the same time on different queues, but they are executed one after another. I am supposed that they could be done parallel.

       

      the code is like that:

      // queue: Q0, Q1

      // kernel: vnd, cnd

      for(;;) {

          clEnqueueNDRangeKernel(Q0, vnd, ... , event0); // vnd produce an event0

          clEnqueueNDRangeKernel(Q0, cnd, ...); // cnd run after vnd in the same queue

          clEnqueueReadBuffer(Q1, ..., 1, event0, NULL); // buffer-reading wait for vnd(event0) to complete

          // wait for buffer-reading to complete

          // ...

      }

       

      I am supposed that cnd and buffer-reading could be executed together on GPU, and the execution sequence is:

      ... -> vnd -> (cnd/buffer-reading) -> vnd -> ...

      but in fact they are done serially.

      QQ截图20120908150702.png

       

      my environment:

      AMD APP SDK 2.7 + Catalyst 12.6 Beta (8.98-120522a-139735E-ATI)

      HD 7970