Harlan,
I don't think so, and there's a way to prove it. The issue is how the
adjtime() routine handles clock adjustments. That syscall should
amortize the adjustment in microseconds at a 500-PPM rate. I bet it
doesn't. And, should the precision time kernel modifications are active,
it will go absolutely bonkers unless some parameters are carefully
recalculated. There should be enough detail in the code commentary for a
kernelmonger to do that.
As all modern machinery I touch here has a processor cycle counter of
some form or another, for good timekeeping you don't need to up the Hz.
Only if you need better than 10-ms precision in order to produce an
external event at a given time.
Dave
Dave
Harlan Stenn wrote:
You are probably losing clock interrupts like crazy at HZ=2000.
H
_______________________________________________
questions mailing list
[email protected]
https://lists.ntp.isc.org/mailman/listinfo/questions
