• State Accepted Answer
  • +1 person also asked this people also asked this
  • Locked Locked
  • Replies 1 reply
  • Subscribers 138 subscribers
  • Views 16613 views
  • Users 0 members are here
2025 survey

We are running a survey to help us improve the experience for all of our members. If you see the survey appear, please take the time to tell us about your experience if you can.

Options
Related
How was your experience today?
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

OpenCL: why there is so big difference between the time counter of cl_profiling_info ?

willhua

I run an OpenCL task on Mali G-57 GPU, and profile the performance of kernel by code below:

cl_int err = clEnqueueNDRangeKernel(pocl->getCommandQueue(), kernel, 2, NULL, globalsize, localsize, 0, NULL, &event_kernel );
clWaitForEvents(1, &event_kernel);
cl_ulong queue_time = 0, submit_time = 0, start_time = 0, end_time = 0;
    err = clGetEventProfilingInfo(event_kernel, CL_PROFILING_COMMAND_QUEUED, sizeof(queue_time), &queue_time, 0);
    checkErr(err, "CL_PROFILING_COMMAND_QUEUED", true);
    err = clGetEventProfilingInfo(event_kernel, CL_PROFILING_COMMAND_SUBMIT, sizeof(queue_time), &submit_time, 0);
    checkErr(err, "CL_PROFILING_COMMAND_SUBMIT", true);
    err = clGetEventProfilingInfo(event_kernel, CL_PROFILING_COMMAND_START, sizeof(start_time), &start_time, 0);
    checkErr(err, "CL_PROFILING_COMMAND_START", true);
    err = clGetEventProfilingInfo(event_kernel, CL_PROFILING_COMMAND_END, sizeof(end_time), &end_time, 0);
    checkErr(err, "CL_PROFILING_COMMAND_END", true);
    printf("time value=%llu %llu %llu %llu", queue_time, submit_time, start_time, end_time);
the log shows:
D/OLOG:cl_success:CL_PROFILING_COMMAND_QUEUED
D/OLOG:cl_success:CL_PROFILING_COMMAND_SUBMIT
D/OLOG:cl_success:CL_PROFILING_COMMAND_START
D/OLOG:cl_success:CL_PROFILING_COMMAND_END
D/OLOG:time value=71608492999037 71608493875422 8517593063107986718 8517593063119119891
We can see that the value of CL_PROFILING_COMMAND_START or CL_PROFILING_COMMAND_END is very bigger than CL_PROFILING_COMMAND_QUEUED or CL_PROFILING_COMMAND_SUBMIT. If we chang these to time, the time cost from CL_PROFILING_COMMAND_SUBMIT to CL_PROFILING_COMMAND_START will be 8517521506304ms. It is impossible.
Can anybody tell me what's the wrong which cause the time counter of the task status is so big? 
Parents
  • +1 Arm Employee Badge

    It seems that the timestamps are in a different time base. This could either be a driver bug or a platform integration issue. What platform/device was this running on? What is the version of the driver on your platform? The values returned for CL_DRIVER_VERSION, CL_DEVICE_VERSION, CL_DEVICE_VENDOR and CL_DEVICE_NAME would be useful.

Reply
  • +1 Arm Employee Badge

    It seems that the timestamps are in a different time base. This could either be a driver bug or a platform integration issue. What platform/device was this running on? What is the version of the driver on your platform? The values returned for CL_DRIVER_VERSION, CL_DEVICE_VERSION, CL_DEVICE_VENDOR and CL_DEVICE_NAME would be useful.

Children
No data