After watching the time vary wildly (by as much as 20 seconds - and +)
over the past day, I've disabled Windows time service and installed NTP
(per previous suggestion, downloaded from
http://www.meinberg.de/english/sw/ntp.htm ) on machine 2. It's been
running for about 3 hours so far, and it's stayed within 100 ms. I'll
keep my fingers crossed. It has slowly drifted from -2 ms right after
starting NTP up to -69 ms now.
Machine 1 is still running Windows time service (Microsoft's
questionable implementation of SNTP), and is performing quite well right
now. Maximum error is about 60 ms over the past few hours, much of the
time near or below 10 ms.
Thanks for you folks' help.
David
David J Taylor wrote:
Richard B. Gilbert wrote:
[]
Windows is simply not very good about keeping time! It can lose clock
interrupts when busy. Starting and stopping the multimedia timers can
mess up the clock.
W32TIME is an implementation of SNTP rather than NTP. It's not even a
very good implementation of SNTP. You may get better results with NTP.
Then again, Windows being Windows, you may not.
For an example of what Windows can do when running a recent version of NTP
(which takes the multi-media clock behaviour into account) and syncing to
a local stratum one FreeBSD server (top graph), please see:
http://www.david-taylor.myby.co.uk/mrtg/daily_ntp.html
W32time, at least in its older incarnations was not good, but for many
purposes today time-keeping in Windows using NTP is quite adequate.
David
_______________________________________________
questions mailing list
[email protected]
https://lists.ntp.isc.org/mailman/listinfo/questions
_______________________________________________
questions mailing list
[email protected]
https://lists.ntp.isc.org/mailman/listinfo/questions
