[ntp:questions] 10 identical PCs, but different clock drifts.

Richard B. Gilbert rgilbert88 at comcast.net
Mon Mar 2 22:42:23 UTC 2009

ryad.bek at gmail.com wrote:
> On Mar 2, 7:44 pm, "Richard B. Gilbert" <rgilber... at comcast.net>
> wrote:
>> ryad.... at gmail.com wrote:
>>> Hi everybody,
>>> I've bought 10 PCs with identical hardware (motherboard,...).
>>> The PCs are located in the same room and running identical
>>> configurations for 4 days now.
>>> They are synchronized using garmin 18 GPSes.
>>> I've just checked the NTP's loopstats file and it seems that all drift
>>> are different.
>>> Here are the reported drifts (4th column of loopstats file) :
>>> PC1=15,6
>>> PC2=1
>>> PC3=-20,7
>>> PC4=14,1
>>> PC5=15,5
>>> PC6=23,7
>>> PC7=-7
>>> PC8=3,1
>>> PC9=3,6
>>> PC10=4,3
>>> As all hardware are the same and brand new, I was wondering if you
>>> find that normal or if there is a problem with my HW config?
>>> Regards
>>> Ryad
>> Absolutely normal.
>> The clock will {\italics usually} not gain or lose more than 43 seconds
>> per day.  That's about the best you can hope for without paying for an
>> add on clock OR using ntpd to discipline your clock.  The equivalent
>> clock frequency error is +/- 500 Parts Per Million (PPM).
>> Only random chance would give you two computers with the same clock
>> drift rate.  Most will be somewhere between +/- 50 PPM.
> For PC1, the drift rate has varied between 15.632 15.650 these two
> last days

The difference is probably due to changes in temperature!  If you keep 
the room at 68 degrees F while you're awake and drop it to, say, 62 
degrees F while you sleep, you will see some changes in the frequency.

Ntpd is quite capable of dealing with this sort of thing as long as the 
changes are neither too extreme nor too rapid.  I don't know what the 
limits are.

More information about the questions mailing list