Date: Sun, 31 Mar 2002 13:29:40 -0700 (MST) From: "M. Warner Losh" <imp@village.org> To: bicknell@ufp.org Cc: dp@penix.org, hackers@FreeBSD.ORG Subject: Re: GPS time. Message-ID: <20020331.132940.91795402.imp@village.org> In-Reply-To: <20020330142859.GA19243@ussenterprise.ufp.org> References: <Pine.LNX.4.43L0.0203291757560.4473-100000@saruman.xwin.net> <20020330142859.GA19243@ussenterprise.ufp.org>
next in thread | previous in thread | raw e-mail | index | archive | help
In message: <20020330142859.GA19243@ussenterprise.ufp.org>
Leo Bicknell <bicknell@ufp.org> writes:
: In a message written on Fri, Mar 29, 2002 at 06:04:11PM -0600, Paul Halliday wrote:
: > I just connected my gps (garmin gps III plus) to my serial port
: > and realized that simply cat'ing cua0 displays date/time/position of the
: > unit. (neato). Anyway, how accurate would it be to use the time from this
: > output for ntp as opposed to my current setup using ntp servers.
:
: Your NTP servers are better.
:
: I tested a III Plus, and without a 1 PPS source (which that model
: doesn't provide) it's accurate to about 100ms, give or take. Since
: real NTP servers are < 1ms, they really aren't that good. It's
: not that the time isn't accurate, it's that they were not designed
: to communicate with that accuracy to an external device.
Also, the reference clocks that ntp servers use generally are in the
+- 1us or better rather than 1ms sync that you can get over a wan.
Of course, don't look too closely under the covers of ntp. it has
well documented startup transient conditions that some applications
may be senentive to.
Warner
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020331.132940.91795402.imp>