Quantcast
Channel: AMD Developer Forums: Message List
Viewing all articles
Browse latest Browse all 4104

Re: Execution of OpenCL Kernel on more than one GPU

$
0
0

Hello maxdz8,

 

in my host program after enqueuing the kernel i use the following code:

 

err = clEnqueueReadBuffer(queue, output_buffer, CL_TRUE, 0,
      sizeof(output_data), output_data, 0, NULL, NULL);

 

    if (err < 0) {

   perror("Couldn't read the buffer");
   client_cleanup(EXIT_FAILURE);

    }

 

    clFlush(queue);

    clFinish(queue);

 

Its a blocking read (CL_TRUE). Should i avoid this?

The queue is for one card only.

I call the host program with variable parameterlist.

At the end of the host program i release all allocated cl objects.

 

greets


Viewing all articles
Browse latest Browse all 4104

Trending Articles