[aklug] Re: ntp and freebsd?

On Fri, Mar 16, 2012 at 3:16 PM, Shane Spencer <shane@bogomip.com> wrote:
> On Thu, Mar 15, 2012 at 6:49 PM, Royce Williams <royce@tycho.org> wrote:
>> Dr. Anderson's claim is built on the idea that the FreeBSD kernel
>> supports nanosecond time precision. =A0I do not know enough about it to
>> verify whether or not Linux is truly lacking in nanosecond-level timing
>> for this purpose. =A0I found multiple people saying that online, but no
>> authoritative source.
>>
>> With Stratum 1, you're building an NTP server directly attached to a
>> high-precision time source, for which millisecond precision isn't
>> precise enough, so just running an NTP daemon isn't sufficient for what
>> he's trying to do.
>>
>> Part of the reason that FreeBSD is so well-equipped for high-precision
>> time is because a famous FreeBSD guy, Poul-Henning Kamp, is a huge time
>> geek, and he did a bunch of work to make FreeBSD recision-time-friendly.
>> He notices when *GPS satellites* have time that's off by less than a
>> millisecond.
>>
>> http://people.freebsd.org/~phk/
>>
>> Royce
>>
>
> I don't think I've ever seen a Stratum 1 box in my life...
>
> I've always used the ntp.org server (ntp in debian) along with a GPS
> for systems that couldn't retrieve time reliably in other ways. =A0I
> really like this approach.. I typically use said systems to serve out
> a local broadcast time as well.
>
> I can't even imagine what hardware it would take to get nanosecond
> precision from Boulder, CO.
>
> - Shane

That said... anything beyond a hundredth of a millisecond as fair as
gaining current time (not measuring deltas) becomes 'magical' in my
book :)
---------
To unsubscribe, send email to <aklug-request@aklug.org>
with 'unsubscribe' in the message body.
Received on Fri Mar 16 15:18:10 2012