NetBurner Community Forum

Just realized that the problem goes away if we stay with default UserMain priority of 10, instead of changing it to 50. So something is indeed preempting our code for 1 tick when main is running at lower priority. Any idea what that might be?

Created ticket 33347 and attached our test code.

Thanks!

I am, actually. I created a new workspace and a new project in NBEclipse 2.6.3, verified that correct paths are used by the compiler, and even moved the old 2.6.2 install to a new location just to be sure..

Unfortunately, switching to the latest NNDK-2.6.3 did not substantially change the stats. And the dependence on TICKS_PER_SECOND is still there. There is another source of latency lurking somewhere...

Is there anything else we can try?

NNDK 2.6.2

Hello,

We are running a latency-sensitive application on a NANO54415 module and are experiencing very strange periodic spikes in UDP receive latency. The problem can be demonstrated with a following test setup:

1. NANO54415 is running a trivial loop that receives UDP packets and measures the time ...