[ntp:questions] Repost: Are there algorithm differences between 4.2.4 and 4.2.5?

David Mills mills at udel.edu
Fri Apr 10 20:15:58 UTC 2009


The core clock discipline (ntp_loopfilter.c) PLL, FLL and poll-adjust 
routines have not changed between the release and development versions. 
Details in the initial frequency computation, kernel interface, 
leapsecond processing and configuration interface have been updated. 
There were minor changes in the loop dynamics for very long poll 
intervals (36 h) to improve stability. These changes have no effect for 
poll intervals less than 4096 s. The loop dynamics should be unchanged

Some of the routines in the release version date from 2006 and and have 
been significantly updated in the development version. Other routines in 
the release version date from this year and are very likely incompatible 
with the old routines in one way or another. I have no idea whether this 
might or might not affect the behavior.


David J Taylor wrote:

>David J Taylor wrote:
>>Are there any algorithm of other differences which might cause
>>noticeable performance differences between 4.2.4 and 4.2.5?  For
>>- 4.2.5 being more reluctant to increasing the polling interval from
>>64s (with a mixture of local stratum-1 and Internet pool servers)?
>>- 4.2.5 showing an increased variability of offset when viewed on a
>>timescale of minutes?
>>Of course, these may just be different manifestations of the same
>>underlying difference.
>It would be good to hear from the algorithm experts about this topic.
>There are some comparative results here:
> http://www.satsignal.eu/ntp/V4.2.4.vs.V4.2.5.html
>questions mailing list
>questions at lists.ntp.org

More information about the questions mailing list