Search:

Type: Posts; User: tmp

Search: Search took 0.00 seconds.

  1. Re: No GPU devices found... But yesterday everything worked...

    Problem solved. I was accessing the machine (a mac) using ssh, and apparently the user also needs to own the desktop session to access the GPUs. The problem is explained in further details here:
    ...
  2. No GPU devices found... But yesterday everything worked...

    Yesterday everything worked perfectly. But today, no GPU devices can be found and I haven't altered the code! I use PyOpenCL and the following code to find GPU devices:


    pl = cl.get_platforms()...
  3. Replies
    3
    Views
    1,769

    Re: How to index local accrays for best performance?

    Opps, the title should of course say "arrays", not "accrays" (whatever that is)... :-)
  4. Replies
    3
    Views
    1,769

    How to index local accrays for best performance?

    I have a problem understanding the memory access patterns for kernels in OpenCL. Consider the below snippet of a kernel. It exists in a two-dimensional 16x16 work-group and basically the 16x16...
  5. Replies
    7
    Views
    3,031

    Re: Mysterious kernel behavior. Can anyone explain?

    Problem solved! It seems that Nvidia hardware that is also used for display has a hard limit of 5 seconds of kernel execution time. After that, the kernel is aborted and an error returned.

    I...
  6. Replies
    7
    Views
    3,031

    Re: Mysterious kernel behavior. Can anyone explain?

    Below is a complete boiled-down example that can be run directly.

    To see if the code works, verify that it outputs (prints) a number larger than zero.
    You can make it work in two ways:
    1)...
  7. Replies
    7
    Views
    3,031

    Re: Mysterious kernel behavior. Can anyone explain?

    I forgot: yes, I have double, triple and quadro checked that the group size is really BLOCK_SIZE x BLOCK_SIZE. :-)
  8. Replies
    7
    Views
    3,031

    Re: Mysterious kernel behavior. Can anyone explain?

    I access the GPU using the python interface pyopencl-0.91.4. I'm not sure where to find the opencl SDK version, but according to the release notes in the doc/folder, it is "OpenCL R190 Update...
  9. Replies
    7
    Views
    3,031

    Mysterious kernel behavior. Can anyone explain?

    The behavior of the test kernel below makes absolutely no sense to me. Any clues to what's going on will be much appreciated!

    The kernel basically initializes an output array, C, with 1's. It...
  10. Replies
    3
    Views
    2,126

    Re: Is this caused by memory contention?

    OK, very interesting! How do I specify that the arrays should be located in "constant" memory? Or in "local" or "private" memory?
  11. Replies
    3
    Views
    2,126

    Is this caused by memory contention?

    I suspect that I experience problems with memory contention in the below setup. Do you agree? If so: can I do anything about it?

    I send two large arrays to the GPU (in form of read-only buffers)...
  12. Replies
    5
    Views
    2,345

    Re: Low number of compute units?

    Can I somehow verify that all 16 compute units (cores) are used? It worries me, that opencl only returns "2" when I ask for MAX_COMPUTE_UNITS, and my running times also match suspiciously well to a...
  13. Replies
    5
    Views
    2,345

    Re: Low number of compute units?

    Interesting. But what does that mean from a parallelization point-of-view? Will my kernel only be executed in two cores simultaneously, or will it automatically be distributed to all 16 units?
  14. Replies
    5
    Views
    2,345

    Low number of compute units?

    According to the official specs of my graphic card (look in the "Specifications" tab) it should hold 16 computation units. However, when querying MAX_COMPUTE_UNITS on the device, it only returns "2"....
Results 1 to 14 of 14