Search:

Type: Posts; User: ajs2

Page 1 of 2 1 2

Search: Search took 0.00 seconds.

  1. Replies
    3
    Views
    1,534

    Re: how to differentiate between two GPU

    clGetDeviceInfo
  2. Replies
    3
    Views
    1,640

    Re: about clSetEventCallBack()

    This is not defined by the spec- the only guarantee is that the callback will eventually be executed after the command completes.



    Best thing to do is to profile your code and find out!

    Note...
  3. Re: CL/GL Interop, OSX -- ever shared a Renderbuffer or Text

    Anything output to the console or when you run with CL_LOG_ERRORS=stderr?
  4. Replies
    3
    Views
    1,482

    Re: A weird problem with "clEnqueueNDRangeKernel"!

    Does that event leak every iteration?

    Try finishing the queue before recording the begin and end times.

    Might be interesting to run the test in a profiler and see where the main thread spends...
  5. Replies
    1
    Views
    1,024

    Re: Releasing cl_event s

    Yes
  6. Re: Making kernels aware of a large amount of buffers

    clSetKernelArg
  7. Replies
    1
    Views
    897

    Re: Why opencl use source, instead IR

    Source code is the most portable encoding.
  8. Thread: Buffer Strategy

    by ajs2
    Replies
    2
    Views
    966

    Re: Buffer Strategy

    The trade-off is that if the composite buffers are too large, the implementation might have trouble placing them in device memory, leading to stalls while other data is shuffled to the host. On the...
  9. Replies
    1
    Views
    1,135

    Re: Why failed to create kernel?

    Try passing an int* as the last argument to clCreateKernel to obtain the error code. Also, on Mac, you can set the CL_LOG_ERRORS=stderr environment variable and a more verbose message will be printed...
  10. Replies
    2
    Views
    1,412

    Re: clCreateProgramWithBinary on Mac OSX

    This is an extension which allows a developer to compile their CL source code into a bitcode file (using the openclc compiler, or automatically in Xcode), and then ship that with their app instead of...
  11. Replies
    14
    Views
    3,755

    Re: How to avoid double allocation on CPU

    The CL_MEM_USE_HOST_PTR is often used in this case- it will wrap an application created memory allocation with a CL mem object.
  12. Replies
    5
    Views
    1,470

    Re: Mapping Data: remap to change?

    You would either use map and unmap, or a read/write command, each time the data must be modified on the host. The latter will always create a copy of your data using the specified pointer. While map...
  13. Replies
    1
    Views
    874

    Re: Reset or release invalidated command queue

    In some implementations invalid command queue is a generic catastrophic error- it might be unrecoverable. The best thing to do is to determine why it happens in the first place, it is some type of...
  14. Replies
    2
    Views
    1,283

    Re: Killing the watchdog timer under MacOSX Lion

    I don't think you can. If you're hitting the watchdog timer your app is probably already interfering with GUI performance of other apps in the system- hopefully there is some way to increase kernel...
  15. Re: multiple OpenCL applications and global memory effects

    The operating system and/or device driver are responsible for device memory management and are free to page on and off data as needed.
  16. Replies
    9
    Views
    3,986

    Re: I can't get a basic opencl program to run

    On Mac the include directive should be:
    #include <OpenCL/opencl.h>

    This follows the general form Framework/framework.h.

    To link the executable, the OpenCL.framework must be added to the...
  17. Replies
    7
    Views
    1,533

    Re: tiny change hangs kernel - vectorizing problem

    That's right, the original expression should produce a compilation error:

    float4 T;
    ...
    T += ( T < 0.f ) * 360.f;

    (T < 0.f) is a relational containing a vector and a scalar; 0.f will be...
  18. Replies
    2
    Views
    2,216

    Re: problem with clFinish

    The enqueued command might not be submitted to the device for execution until either a flush or a subsequent blocking call on the queue.
  19. Replies
    1
    Views
    787

    Re: Tool for viewing threads' information

    You could use a profiler like Instruments.
  20. Replies
    4
    Views
    1,648

    Re: Double type problem

    The AMD extension cl_amd_fp64 is not supported by Apple's OpenCL framework. On that config, the Khronos extension, cl_khr_fp64 is available for the CPU.
  21. Replies
    8
    Views
    1,877

    Re: What's going to happen if not releasing kernels

    Not releasing CL objects will result in memory leaks. In this case, the program would leak a lot more than just the kernel-

    Even though the context was released by the main program, since the...
  22. Re: "random" kernel crash after running for minutes.... HEP!

    Is anything recorded on the System Console when the program hangs or crashes? Both types of failures should produce entries in the system log.

    It is certainly worth filing a radar with Apple to...
  23. Replies
    4
    Views
    1,617

    Re: Error when create image with clCreateImage2D

    When host_ptr is NULL, image_row_pitch must be zero (here it is 512*4), but no error code is specified for this case.
  24. Thread: clCreateUserEvent

    by ajs2
    Replies
    3
    Views
    1,287

    Re: clCreateUserEvent

    It is not possible for the application to set event profiling information. It is also not possible to query profiling info for user events, clGetEventProfilingInfo will return...
  25. Thread: OpenCL on Mac OS

    by ajs2
    Replies
    1
    Views
    1,190

    Re: OpenCL on Mac OS

    Yes, OSX 10.6.x SnowLeopard includes OpenCL
Results 1 to 25 of 29
Page 1 of 2 1 2