2 Replies Latest reply on Oct 21, 2011 12:45 PM by evk8888


      using CL_MEM_COPY_HOST_PTR


      Hi, this way of creating a buffer for the kernel also loads the data from the host to the buffer and we dont have to enqueue the data using the command queue. 

      cl_mem input1 = clCreateBuffer(context,CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR ,sizeof(float)*count ,data ,NULL);

      is there a similar way where we can read/copy the data from the (buffer)accelerator to the host without using command queues (clEnqueueReadBuffer)?
      thanks guys..