Search:

Type: Posts; User: jasno

Search: Search took 0.00 seconds.

  1. Re: fatal error 1083: Cannot open include file 'Cl/cl.h' no

    Hi,
    It seems like you haven't installed any openCL implementation. You need to download
    one from some vendor depending on the hardware you wish to use. If you have an nvidia
    graphics card then...
  2. Re: Command queue goes invalid after kernel execution

    How big is global_wsize? If it is greater than 32 then isn't this going to run off the end of the array

    bucket[addr + 27*tx] = 0;

    since
    __local short bucket[864];

    and
    int tx =...
  3. Re: Command queue goes invalid after kernel execution

    Well, I'm not sure about your command queue error, you might need to post more of the code
    that is causing the problem. One thing I am curious about though is why you do this

    mem2 =...
  4. Re: Compute Units of GTX560 is wrong, showing only 7

    The card has 7, of what Nvidia call, Streaming Multiprocessors (SMs) which OpenCL
    refers to as Comput Units. Each of thse (on that card) contains 48 Streaming Processors
    (SPs, also called CUDA...
  5. Replies
    10
    Views
    2,365

    Re: Trouble with first OpenCL program

    I'm afraid that you still aren't going to have this running in parallel. If you think about what your "if (get_local_id(0) == 0)" statement is achieving, it is saying that for each work-item in the...
  6. Replies
    14
    Views
    2,850

    Re: minimal efficient workgroup size

    Can't the GPU run another workgroup in parallel on the same compute unit to hide latency?[/quote]

    I believe that is the case, but typically a workgroup will have many times more threads than would...
  7. Replies
    11
    Views
    3,037

    Re: opencl programming

    Yes it will be compatible, but intel also provides an OpenCL implementation
    that you might like to try.

    http://software.intel.com/en-us/article ... pencl-sdk/

    --
    jason
  8. Replies
    10
    Views
    2,365

    Re: Trouble with first OpenCL program

    You say "After the execution of the kernel sigma should contain 2 results, the sum of workgroup1 and workgroup2" and in your code you have

    sigma[(int)sum] = sum;

    But how do you know where the...
  9. Replies
    2
    Views
    1,691

    Re: How to get the warp/wavefront size in runtime?

    Hi,
    I don't think these concepts exist in OpenCL. While I accept that in some underlying hardware
    they will exist, they are not universal. If you desperately need them AND you know you will...
  10. Replies
    6
    Views
    1,709

    Re: Second chance exception after kernel execution

    The globalSize is not directly related to the size of the data you have but is a combination of the
    size of the problem and the amount of work each instance of the kernel performs. It is not...
  11. Replies
    10
    Views
    2,365

    Re: Trouble with first OpenCL program

    I'm not clear what you are trying to achieve but I can tell you why the first 64 values
    appear to work (assuming your workgroup size is 64)

    you have

    sum = sum + mem[group_id * local_size + i +...
  12. Re: CL_OUT_OF_RESOURCES with any more than 49152 byte local

    I assume you are using an NVIDIA Fermi card of some sort ? The max shared memory for
    this card is

    CL_DEVICE_LOCAL_MEM_SIZE: 48 KByte

    (which is 49152 bytes). You can query this value in your...
  13. Replies
    10
    Views
    2,365

    Re: Trouble with first OpenCL program

    Hmmm, where to begin . . .

    you say that you are expecting 2 work groups with 64 work-tems in each and that you
    pass in an array of 128 numbers. You then get the global id of each work-item
    int...
  14. Replies
    3
    Views
    1,085

    Re: Crash after trying recreate buffers

    Hi,
    I was just wondering, you say it crashes after the first time, do you mean it always runs first
    time but can crash anytime after that ?

    Also, can you explain what is happening here

    ...
  15. Re: how to debug opencl kernel program running on GPU?

    Just to check, when using local memory you can only use data from work-items
    in the same work-group, and also did you set a barrier (barrier(CLK_LOCAL_MEM_FENCE);)
    before you started this serial...
Results 1 to 15 of 18