[ntp:questions] Polling time backoff
unruh at wormhole.physics.ubc.ca
Thu Dec 3 22:31:56 UTC 2009
On 2009-12-03, David Woolley <david at ex.djwhome.demon.invalid> wrote:
> Rob wrote:
>> And of course it is always "better" to have a short poll interval, for
>> any kind of error that you want to correct. You may need to add some
>> filtering to the samples before feeding the result in the loop, but 16
>> samples in 1024 seconds will always be providing better information than
>> a single sample in that interval.
>> I understand that it does not work like this in the ntp reference
>> implementation, but that is a characteristic of the implementation, not
>> of the sampling.
> My understanding is that ntpd significantly oversamples, relative to its
> loop filter bandwidth, so further oversampling is unlikely to make a lot
> of difference.
But what it does it throw away 85% of those samples. That does no good.
It really does not do much oversampling when you only consider the
retained data points. Also, the "bandwidth" depends on the sampling
period. Also oversampling can always be used to beat down statistical
white noise. It is less useful for pink noise, but still useful.
More information about the questions