[ntp:questions] NTP Performance on WINNT

Unruh unruh-spam at physics.ubc.ca
Mon Jul 20 22:30:18 UTC 2009


"David J Taylor" <david-taylor at blueyonder.not-this-part.nor-this.co.uk.invalid> writes:

>Richard B. Gilbert wrote:
>[]
>> That is what a "site survey" is.  The GPS determines its location a
>> few hundreds of times over the course of a day and does a "least
>> squares" calculation to get a reasonable approximation of your
>> latitude and longitude.
>>
>> Cellular phone base stations do a more extended site survey, thirty
>> days instead of one.  This gives them a MORE reasonable
>> approximation.  When I did my site survey (24 hours) the locations 
>> plotted a locus about
>> 300 feet long in an East-West direction and with a width of about
>> thirty feet.
>> If I had had the patience and if I had required sub microsecond
>> accuracy I could probably have trimmed that locus further.  . . . .
>>
>> It is enough, for me, that my little herd of computers agree as to
>> approximately what time it is.

>Thanks, Richard.  I took your "You do need to run a "site survey"" as 
>something rather more involved than letting your GPS just sit there.

>For me, a few milliseconds is good enough, but I always want to do 
>better....

a) Don't run Windows.
b) Don't run Windows.
c) I get roughly 2us accuracy. That is about 1000 times better that you desire.
(again a Garmin 18 LVC receiver with PPS running onto my parallel port with my own 
interrupt service routine on a Linux system.)


>Cheers,
>David 




More information about the questions mailing list