Im running an intel atom also (old one, d945gclf2 board), and its only doing ntp at the moment with a fairly minimal 64bit ubuntu 12.10 server.

after 24 hours its now looking like this:

remote refid st t when poll reach delay offset jitter ============================================================================== +203.0.178.191 43.128.117.84 2 u 38 64 377 8.960 -11.714 0.169 *27.54.95.12 218.100.43.70 2 u 31 64 377 61.301 -1.256 0.246 -202.6.248.7 130.102.128.23 3 u 46 64 377 9.529 -0.273 0.160 -116.66.162.4 130.234.255.83 2 u 52 64 377 20.386 15.747 1.988 +121.0.0.41 204.152.184.72 2 u 59 64 377 25.295 -4.202 3.760 o127.127.20.0 .GPS. 0 l 1 8 377 0.000 -0.001 0.001

I've just tuned the gps so that its only outputing a single nmea sentence (GPZDA) and increased the speed to 9600 and now its no longer needing the fudge factor. Ultimately though, i'll want to try get it running from a geode based machine (which has some seriously low specs, but it'll be amusing to see how it performs).

Something else i've noticed as well when im watching it, when it adjusts the system time it seems to impact the jitter calculation quite considerably and with the other ntp servers in there, it doesnt always seem to want to choose the gps as the absolute truth.

On 07/04/13 23:46, David Taylor wrote:
On 07/04/2013 11:14, Paul J R wrote:
Hi All,

I've got a SiRF star 3 GPS with a PPS line which im trying to get
running on a machine with ubuntu 12.10 and a serial port using the DCD
line for pps signalling.
[]
o127.127.20.0    .GPS.            0 l 4    8  377    0.000 -0.065
0.003
[]
Seems like a pretty decent little gps unit, assuming i've got it
functioning and configured correctly?

Just for comparison:

Intel Atom PC running FreeBSD 8.2 (Pixie):
o127.127.20.1    .PPS. offset/jitter -0.001   0.002

Two Raspberry Pis running Linux 3.2.27+
o127.127.22.0    .PPS. offset/jitter 0.002   0.002
o127.127.22.0    .PPS. offset/jitter 0.001   0.002

Raspberry Pi running Linux 3.36.11
o127.127.22.0    .PPS. offset/jitter 0.000   0.002

65 microseconds offset seems a little high to me, but the PCs I've just quoted are all (mostly) /only/ serving NTP. I have been doing a little installation work on Raspberry Pi #3 today, and it now has a steady CPU and network I/O load. Its PPS was only configured yesterday. Performance graphs are here:

  http://www.satsignal.eu/mrtg/performance_ntp.php

I hope that helps your comparisons.


_______________________________________________
questions mailing list
[email protected]
http://lists.ntp.org/listinfo/questions

Reply via email to