Search:

Type: Posts; User: coleb

Page 1 of 4 1 2 3 4

Search: Search took 0.00 seconds.

  1. Re: cl.hpp: KernelFunctor gone,replaced with KernelFunctorGl

    The KernelFunctor was technically never part of the OpenCL 1.1 C++ bindings specification: http://www.khronos.org/registry/cl/spec ... us-1.1.pdf

    This was because the design of this feature is...
  2. Re: CommandQueue async methods don't release Events properly

    Ah yes, I see it now. For a temporary workaround you can use a temporary cl::Event object and then assign it back. operator= will do the proper thing with retain and release.


    cl::Event...
  3. Re: CommandQueue async methods don't release Events properly

    Hi Federico,

    Can you post a more complete example of the memory leak. Including what OpenCL implementation you are using. The following code is rock solid constant in terms of memory usage on...
  4. Replies
    7
    Views
    2,750

    Re: Can the GPUs be leaveraged differently

    Keep in mind that concurrent kernel execution on Fermi has to be from the same context. So this isn't quite multi-processing yet. Furthermore, it's still SIMT (single instruction multiple threads),...
  5. Re: Can OpenCL run on a single core machine without any GPU?

    I know the Apple implementation and the AMD implementation both support a CPU device.
  6. Replies
    99
    Views
    67,002

    Re: OpenCL C++ Bindings

    Whoa, cool. Now I need to dig through the other compilers (msvc7, xlc, solaris CC, etc). We live in a messy world. :-)
  7. Replies
    99
    Views
    67,002

    Re: OpenCL C++ Bindings

    Very likely there's something I'm not getting about shared_ptr as well, we're not allowed to use it here yet since we have to support compilers as old as GCC 3.2 so I don't have that much experience...
  8. Replies
    99
    Views
    67,002

    Re: OpenCL C++ Bindings

    That's essentially what detail::Wrapper is, an implementation of shared_ptr.

    Also note the design of OpenCL C++ layer does not preclude you from using shared_ptr. The following should work:

    ...
  9. Replies
    99
    Views
    67,002

    Re: OpenCL C++ Bindings

    All the objects handle the reference counting automatically for you the same as shared_ptr. So I'm not quite sure what the question is. Is it "why not just use shared_ptr?" I believe the answer to...
  10. Replies
    5
    Views
    2,048

    Re: Crash when accessing struct fields

    Everything looks kosher to me. Though it doesn't preclude an issue on the host side.

    AMD's implementation will run on Intel CPUs, so give it a try.

    -Brian
  11. Replies
    5
    Views
    2,048

    Re: Crash when accessing struct fields

    The code you posted is incomplete. Where is the "event" variable used in the inner loop declared? If it's declared at program scope it is implicitly a __constant variable and can't be assigned to.
    ...
  12. Replies
    3
    Views
    2,044

    Re: vec3, mat3x3,... help!!!

    OpenCL 1.1 will have built in support for vec3 types. However, it will not have a matrix type. So in short, yes, you are better off writing a simple math library that suites your needs.
  13. Replies
    9
    Views
    7,542

    Re: non-blocking call to clEnqueueNDRangeKernel?

    The next version of the spec (OpenCL 1.1), has a lot more detail in the thread safety section. Since we're a new member of the khronos group I'm not sure how much more I'm allowed to say.



    I'm...
  14. Re: Constant pointer dereference issue, possible compiler bug?

    Sounds like a bug in Apple's implementation. Apple's implementation (even on NVidia hardware) is quite different. I suggest you post a bug report with them. Apple also has their own OpenCL forum...
  15. Replies
    99
    Views
    67,002

    Re: OpenCL C++ Bindings

    Using the same kernel from multiple threads will always be flat out evil. There's not much that can be done about it since clSetKernelArg and the subsequent clEnqueue* commands will always race. So I...
  16. Replies
    9
    Views
    7,542

    Re: non-blocking call to clEnqueueNDRangeKernel?

    In my experience with the NVidia implementation the best way to achieve parallelism across devices is to launch a CPU thread for each device. This worked and scaled very well for me on an S1070 (1U 4...
  17. Replies
    99
    Views
    67,002

    Re: OpenCL C++ Bindings

    There has been some talk about merging the KernelFunctor convenience methods into the Kernel class. This is because the way the API is structured now is misleading. For example, it appears like the...
  18. Re: how to do task parallel and data parallel on the same device

    The next generation Fermi cards will be able to execute multiple kernels at the same time. However, current cards can only execute one kernel at a time, there is no way around this. Why not place...
  19. Replies
    2
    Views
    1,892

    Re: Cannot run OpenCL Device Query

    So what is it doing?
  20. Replies
    99
    Views
    67,002

    Re: OpenCL C++ Bindings

    I've done the following in my local copy:


    namespace detail {
    // Specialized for cl_context
    template <typename Func>
    struct GetInfoHelper<Func, cl::Context>
    {
    static cl_int
    get(Func...
  21. Thread: compile errors

    by coleb
    Replies
    2
    Views
    1,797

    Re: compile errors

    Need a bit more information. What is the program? What is the kernel? What is the exact error message? Are you using C or C++?
  22. What should arg_size be when passing NULL to clSetKernelArg?

    The standard allows NULL to be passed as the arg_value parameter to clSetKernelArg when you want a global pointer argument of a kernel to be NULL. However, it doesn't specify what value should be...
  23. Re: #include headers from memory, not from filesystem?

    Why about using the inverse of a traditional header gaurd to turn the header file on and off? The trick is making sure to pass the files in the correct order to clCreateProgramWithSource. Something...
  24. Replies
    5
    Views
    3,172

    Re: Unique device/implementation name

    I went nuts and used everything I could find:


    static std::string GetUniqueDeviceName(const cl::Device &device)
    {
    std::string uname;
    uname +=...
  25. Re: Max __constant variables defined in program source

    The standard seems to indicate a difference between constant kernel arguments and the __constant variables at program scope. Though it's not clear what's the safe way to initialize a program scope...
Results 1 to 25 of 94
Page 1 of 4 1 2 3 4