[ntp:questions] Sub-millisecond NTP synchronization for local network

Hal Murray hal-usenet at ip-64-139-1-69.sjc.megapath.net
Sun Dec 7 09:31:46 UTC 2008


>OK, strange. YOu can look at www.theory.physics.ubc.ca/chrony.html to see
>the delays on my network-- it is local to a university building, but all
>over the building. And I see essentially no difference between computers on
>one side or the other. The delays on the 100Mbs parts of the network are on
>the .15ms level with offsets at the .02ms level. However, the gigabit
>machines are much much worse. I had not looked recently, and an a bit
>perturbed now with how bad they are in comparison with previous levels (
>from 6 months ago).

Some Ethernet adapters have a bug/feature similar to RS-232 chips.  The
idea is to batch interrupts to reduce overhead.  Ethernets do it by
only making one interrupt for several packets as compared to several
bytes for the RS-232 chips.

I'd expect gigabit cards/chips will be more likely to have that feature
than old "slow" 100 megabit chips.

I'd expect the switches to do the right thing.  Too many people would
scream if they didn't.

I just checked a couple of PCs here.  The delay slot (I assume that's
round trip time) varries from 0.128 ms to 0.266.  That's through a 100
megabit switch.  Faster PCs have lower delay.

-- 
These are my opinions, not necessarily my employer's.  I hate spam.




More information about the questions mailing list