[ntp:questions] Re: calculating mean PPM value

David L. Mills mills at udel.edu
Sun Sep 21 15:58:16 UTC 2003


Drk,

Averaging signed offsets would be pretty stupid, no? Try averaging
absolute offsets. Much smarter.

Dave

Drk Ryan wrote:
> 
> I understand that, but to calculate the mean for, say, 2
> machines...one with +50ppm frequency offset and the other with -50ppm
> frequency offset.
> Is the mean (50+50)/2= 50 or (50+(-50))/2 = 0
> 
> "David L. Mills" <mills at udel.edu> wrote in message news:<3F6CAC12.CE894AF8 at udel.edu>...
> > Drk,
> >
> > What you see is what you get. The data represent the frequency offset as
> > measured by ntpd on the machine in test. Remember, the only way to see
> > these data remotely is via the ntpq program.
> >
> > Dave
> >
> > Drk Ryan wrote:
> > >
> > > In Dr. Mills 1997 survey of 20,000 Internet hosts, the mean frequency
> > > error was found to be 78 PPM. When calculating this value do you add
> > > the |absolute| frequency errors or is it necessary to look at the +,-
> > > signs involved?



More information about the questions mailing list