Will dispatching computation tasks individually possibly improve overall performance?

Autonomous Machines Jetson & Embedded Systems Jetson Xavier NX
1

Hi all,

If I dispatch several tasks together to GPU, there seem to be a context switch overhead among tasks because GPU has a round-robin scheduler by default. So will it be better if I dispatch tasks individually? I suppose this may depend on many factors, such as memory requirements. However, in my simple experiments, I didn’t observe an improvement in overall finishing time. Does anyone have any ideas about what happened and how to avoid this overhead?

Any thoughts or ideas are appreciated!

Best,
Sen

3

Hi,
Not sure if this helps but for profiling system performance, you can execute sudo tegastats to get overall system status. The information should be helpful in debugging performance issue.

5

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.