[ntp:questions] NTP on CubieBoard
unruh at invalid.ca
Mon Oct 13 19:19:39 UTC 2014
On 2014-10-13, Rob <nomail at example.com> wrote:
> David Woolley <david at ex.djwhome.demon.invalid> wrote:
>> In this case, he is quoting a low jitter, which means that the offset
>> should also be low, unless ntpd has not yet acquired lock. If it can
>> measure a stable jitter of around a microsecond, I think you can almost
>> certainly rule out a pure interrupt driven clock. If the jitter
>> occasionally spikes, it could be that things are being synchronised to
>> ticks, but you would then see unusually high delays.
> Remember it is PPS synchronized, not network synchronized.
> The action of taking an interrupt, reading the current time and
> storing it for later processing at kernel or user level can be performed
> in well under a microsecond on a modern PC.
Well, I am not so sure of that. When I measured it (drive a parallel
port output pin and time that, attach it to the interrupt pin, and time
the interrupt) gave me about 1-2us. But that the machine had a parallel
port tells you something about the age of the system.
> The question is if a Cubieboard can do that as well.
Sorry, I do not know the cubie.
> associd=0 status=011b leap_none, sync_pps, 1 event, leap_event,
> version="ntpd 4.2.7p448 at 1.2483-o Sun Aug 24 10:33:47 UTC 2014 (1)",
> processor="x86_64", system="Linux/3.2.0-4-amd64", leap=00, stratum=1,
> precision=-23, rootdelay=0.000, rootdisp=1.000, refid=PPS,
> reftime=d7e6840d.6b29e4dd Mon, Oct 13 2014 19:02:05.418,
> clock=d7e6840e.58db107a Mon, Oct 13 2014 19:02:06.347, peer=5626, tc=4,
> mintc=3, offset=0.000420, frequency=-17.882, sys_jitter=0.000709,
> clk_jitter=0.001, clk_wander=0.000
More information about the questions