On Wed, Sep 14, 2011 at 15:51, Chris Albertson
<[email protected]> wrote:
> At least in the case of a GPS reference clock, I don't think priority
> has an effect of accuracy.  The PPS signal gets time stamped inside
> the interrupt handler.  At long at NTP can process each PPS before the
> next one comes in it should be fine.

Correct, with a PPS source, priority of ntpd matters much less.
Similarly, most systems provide the SO_TIMESTAMP socket option which
ntpd uses to request the IP stack timestamp the arrival of NTP
packets.  Where SO_TIMESTAMP is not available, ntpd takes a timestamp
when it processes the packet and treats it as the arrival timestamp,
so higher priority helps.  Windows does not offer SO_TIMESTAMP, and in
fact the port of ntpd to Windows ignores -N and always attempts to
raise its priority, though it will proceed with a warning logged if
not allowed.  -N is ignored because historically the ntpd Windows
port's priority-raising code failed to take it into account, and
always attempted.

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

Reply via email to