profiling with no output

Accelerated Computing CUDA CUDA Programming and Performance
1

I have OPENCL_PROFILE=1 and OPENCL_PROFILE_CONFIG=file.cfg both set

The opencl_profile_log is being created with the appropriate column headers but there is no actual profiling occuring. Can anyone give me an idea as to why this is happening?

Also, the code compiles fine, runs fine and verifies fine.

For initial thought was that maybe it is using the CPU but that doesn’t seem to be the case. Also, the code was originally developed for AMD GPU but it seems to run fine so that shoudn’t be a reason it’s not profiling.

Here’s my “device” code:

    cl_int status;

    // 1. Get a Platform
    cl_platform_id platform;
    status = clGetPlatformIDs(1, &platform, NULL);
    cout<<"clGetPlatformIDs :"<<status<<endl;

    // 2. Find a GPU device
    cl_device_id device;
    status = clGetDeviceIDs(platform, CL_DEVICE_TYPE_GPU, 1, &device, NULL);
    cout<<"clGetDeviceIDs :"<<status<<endl;
    cout<<"device :"<<device<<endl;

    cl_context_properties properties[] =
    {
            CL_CONTEXT_PLATFORM,
            reinterpret_cast<cl_context_properties>(platform),
            0
    };
    // 3. Create a context and command queue on device
    cl_context context = clCreateContext(properties, 1, &device, NULL, NULL, &status);
    cout<<"status: "<<status<<endl;
    getchar();

ANY THOUGHTS?

2

Seriuosly?

No one knows how to use computeprof in command line?

3

Seriuosly?

No one knows how to use computeprof in command line?

4

Might help to post in the OpenCL Section rather than the CUDA section.

5

Might help to post in the OpenCL Section rather than the CUDA section.

6

Yeah, sorry, I did that and got a response but for the record aren’t the CUDA and OpenCL profiler essentially one in the same?

7

Yeah, sorry, I did that and got a response but for the record aren’t the CUDA and OpenCL profiler essentially one in the same?