[ntp:questions] Tracking the drift of a GPS clock relative to a HW clock

Unruh unruh-spam at physics.ubc.ca
Wed Jan 21 17:19:19 UTC 2009


David Woolley <david at ex.djwhome.demon.co.uk.invalid> writes:

>ryad.bek at gmail.com wrote:

>> This algorithm is robust because it is based on
>> special techniques (it uses common packets in the traces to provide
>> synchronisation).

>In the end, any such algorithm is going to have to use techniques 
>similar to ntpd or chrony's.  The one advantage is that it is 
>retrospective, and therefore probably better matches chrony's linear 
>regression approach.  (NTP uses a low memory, incremental algorithm.)

That is why ntp has a VSZ of 4460 vs 2189 for chrony and RSS of 940 vs 748
for chrony:-)
 


>A better example of this sort of problem, for this newsgroup, is 
>probably that of deriving UTC from historical information on the 
>individual clocks that contribute to UTC.

But he also does not realise that the unconstrained computer clock would
soon be out by seconds (in a day) minutes, or even hours. 


>I've also done this sort of thing on data with rather larger jitters and 
>the main error due to sloppy eyeball and wristwatch time setting 
>(performed at not well controlled times) on specialist (although rather 
>common) equipment, that doesn't support NTP.




More information about the questions mailing list