Search:

Type: Posts; User: AlexAtOSU

Page 1 of 2 1 2

Search: Search took 0.00 seconds.

  1. Replies
    2
    Views
    1,826

    Re: clBuildProgram crashes

    Good 'ol "make a backup copy and start deleting" debugging...


    It looks like I have an issue with:



    for (...)
    {
    if ( ... )
  2. Replies
    2
    Views
    1,826

    clBuildProgram crashes

    99 err = clBuildProgram(clProgram[0], 0, NULL, NULL, NULL, NULL);
    (gdb)

    Program received signal EXC_BAD_ACCESS, Could not access memory.
    Reason: KERN_INVALID_ADDRESS at address:...
  3. Re: Working with single channel images? (Is it possible?)

    Good to know it works though :). I've been working with CL_RGBA recently and have had no time to try anything with CL_R
  4. Replies
    5
    Views
    2,312

    Re: About writing to Global Memory

    Oh, I see the difference now. I'm not very sure now that I see what you're asking.

    Of these kernels, one writes a literal to your global memory space, the other uses an initialized local variable...
  5. Replies
    5
    Views
    2,312

    Re: About writing to Global Memory

    It sounds like this is due to the time it takes to do a JIT, just-in-time, compile of your kernel. Does the 3rd run execute as fast as the 2nd?
  6. Re: Why is default interpolation done this way?

    :lol: :roll:

    Thats what I figured. Someone has to draw the line between CL and GL somewhere and I can find no good arguments for or against changing it from how OpenGL does it. At least, none...
  7. Re: Why is default interpolation done this way?

    I understand what is going on with ijk... it is just a C-style means to get floor() and ceil(), but why the a, b, and c components?
  8. Why is default interpolation done this way?

    My question (before I give the backstory):
    Why does the specification subtract 0.5 from coordinates when interpolating? The values I wish to use are stored at the vertices of my texture. This seems...
  9. Re: Function to get human readable error messages

    I have found that this helps.
    export CL_LOG_ERRORS="stdout"
    (or stderr etc...)

    Also, owning a copy of the error codes and their integer values has been immensely useful. Finally, I would agree...
  10. Re: Is CL_KERNEL_WORK_GROUP_SIZE value hardware dependent?

    Yes it is. I usually hate thread redirects, but this thread goes into depth regarding your question almost exactly:
    viewtopic.php?f=37&t=2991

    I believe it comes down to cores and available...
  11. Replies
    7
    Views
    2,826

    Re: OpenCL and encapsulation

    Thanks david, good to know this info.
    (I'm still learning. I should probably research more before contributing to technical questions!)
  12. Replies
    3
    Views
    5,674

    Re: Tutorial series for OpenCL with C++

    Sounds interesting. If you want any help just message me. I am sort of in the same process myself, except I'm only posting things with regards to topics for which I have not been able to find...
  13. Replies
    7
    Views
    2,826

    Re: OpenCL and encapsulation

    I tried to edit that quickly, but it wouldn't let me, sorry for the extra post. To amend what I said:

    I meant, I do not know if opencl has overloaded copy methods for cl_command_queue and...
  14. Replies
    7
    Views
    2,826

    Re: OpenCL and encapsulation

    If you do everything by reference this should work regardless of where in memory the objects exist. If you are trying to pass by variable I am not sure. I don't know if cl_mem is a class object with...
  15. Replies
    5
    Views
    4,102

    Re: Out of resources problem

    The fact it works for small workgroup sizes leads me to believe you're exceeding your device's capabilities with high values.

    On Page 107 of the the 1.0.48 Specification Khronos lists various...
  16. Replies
    5
    Views
    4,102

    Re: Out of resources problem

    "local_work_size can also be a NULL value in which case the OpenCL implementation will determine how to be break the global work-items into appropriate work-group instances." --...
  17. Replies
    5
    Views
    3,050

    Re: correctly deleting buffers

    I looked in a function I made to clear my memory buffers / images. I have never actually checked to see if the video card memory is freed, but maybe you can test it for me ;).



    bool...
  18. Replies
    5
    Views
    3,050

    Re: correctly deleting buffers

    In addition, is this your real code or you are you also capturing error codes when you call clFlush? Make sure it CL_SUCCESS (e.g. == 0).
  19. Replies
    5
    Views
    3,050

    Re: correctly deleting buffers

    Maybe try following this up with clFinish?

    http://www.khronos.org/registry/cl/sdk/ ... Flush.html indicates there is no completion guarantee on clFlush.
  20. Replies
    1
    Views
    1,549

    Re: initializing constant variable from file

    You should be able to read the variable in from the host, and pass into the kernel as a READ_ONLY buffer.
  21. Replies
    6
    Views
    3,229

    Re: OpenCL on Windows much slower than on Mac?

    The standard is young. It only guarantees that implementations will execute the requirements correctly. They generally have few, if any, performance mandates.

    That said, Apple knows very well how...
  22. Replies
    1
    Views
    1,472

    Re: Ternary operation for vector (select)

    // 4-tuple integer
    int4 myResult;

    // Variable to hold boolean condition
    int4 c;

    // Component wise compare --
    // According to the spec:
    // These functions shall return a 0 if the...
  23. Replies
    6
    Views
    3,229

    Re: OpenCL on Windows much slower than on Mac?

    I can think of no good reasons. It could be any number of things from memory management at the OS level, to queue management inside the library -- and several other things at that.

    You could...
  24. Replies
    6
    Views
    3,229

    Re: OpenCL on Windows much slower than on Mac?

    Are these run from the same laptop? (e.g. Virtual Machine / Bootcamp install of XP) If yes, whose API are you invoking on the XP side? NVIDIA? Are they run from different machines? Which graphics...
  25. Replies
    3
    Views
    2,517

    Re: 16bits per channel RGB Image

    Does section 5.2.4.1 of the 1.0 Specification Revision 48, "Image Format Descriptor," help you?

    Host Code:
    If you have an image, you should also have a image format descriptor.
    Set the...
Results 1 to 25 of 27
Page 1 of 2 1 2