Intel® oneAPI Base Toolkit
Support for core tools and libraries to build and deploy high-performance data-centric applications
317 Discussions

Program startup overhead for small program oneAPI compared to ParallelStudio

AndiMb83
Beginner
410 Views

Hi,

 

we have a small program under Linux, where the runtime is factor 10 higher when running with oneAPI libraries compared to Parallel Studio libraries. This does not depend on the compiler, which was used.

When running the program with the Parallel Studio environment/libraries the runtime is arround 0.02s.

When running the program with the oneAPI environment/libraries the runtime is arround 0.6 to 0.7s.

An strace shows that "/proc/kallsyms" is read, when running under oneAPI and I can see lots of "clock_nanosleep(CLOCK_REALTIME, 0, {tv_sec=0, tv_nsec=125000}, NULL) = 0".

Is it somehow possible to reduce the overhead of oneAPI for small programs?

Regards,

Andreas

Labels (1)
0 Kudos
1 Reply
ShivaniK_Intel
Moderator
371 Views

Hi,


Thanks for reaching out to us.


As we have received the closing request from your end we are closing this thread. This thread will no longer be monitored by Intel. If you need further assistance, please post a new question.


Thanks & Regards

Shivani


Reply