[ntp:questions] Effect of Gigabit Interrupt coalescence on ntp timing

unruh unruh at invalid.ca
Tue Jun 12 14:30:29 UTC 2012


I have posted a web page
www.theory.physics.ubc.ca/scatter/rt.html
giving some measurements of the effect of interrupt coalescence on a
Intel Gigabit Ethernet on the synchronization of client computers via
ntp packets. The difference between running the card at 100Mbps with
coalescence and at Gb without drops the offset scatter by a factor of
over 5 (from about 30us, to about 5us). This was discussed in another
thread (on RADclock) where the results could well be hidden from people
interested in the topic, so I have put this into a new thread.



More information about the questions mailing list