[ntp:questions] Re: SUSE OSS 10 NTPD Clock Drift

David J Taylor david-taylor at blueyonder.co.not-this-bit.nor-this-part.uk.invalid
Fri Jan 20 06:22:03 UTC 2006


Richard B. Gilbert wrote:
[]
> WinDoze exhibits the same problem with lost clock interrupts.  The
> difference is that there is nothing you can do about WinDoze except
> run a better O/S!  The problem results from interrupts being masked or
> disabled for two or more consecutive clock "ticks".   Only the first
> "tick" registers, the rest are lost.  This typically happens during
> heavy disk activity.  Runing the multi-media timers can also gum
> things up royally but there is a fix or workaround for that.  If your
> clock gains rather than loses time, you needn't worry about lost
> interrupts. If the machine drifts badly when idle, the problem is
> probably not lost interrupts.

What is your evidence for saying this about lost interrupts with Windows? 
I ask because I have Windows running NTP on a number of systems ranging 
from an old 266MHz AMD running Windows NT4 up to a 2.8GHz Pentium 4 
running Windows 2000, and a 1.9GHz Pentium 4 running Windows XP, and I 
have not seen problems with lost interrupts.  Of course, disk DMA should 
be enabled, and I suppose that a badly written third-party device driver 
could cause problems.  What situations might cause lost interrupts, and 
how can they be avoided?  Do you think these problems are present in the 
basic OS?

You are, of course, correct that the multi-media timers do need to be 
permanently enabled (if you run software which might enable and disable 
them), and fortunately there is now a version of NTP which does that:

  http://www.meinberg.de/english/sw/ntp.htm

David 





More information about the questions mailing list