[ntp:questions] refclock_gpsvme.c from True Time GPS-VME toBrandywine PCI-SyncClock32

Jason Rabel jason at extremeoverclocking.com
Sat Aug 11 20:24:20 UTC 2007

If you check the official NTP docs, driver 21 actually has no info...


The wiki link to "ConfiguringTrueTimeRefclocks" is actually for driver 5 (I
wrote a majority of that page), why driver 21 goes to that page too I don't
know. Probably because there was no other documentation and the name was

If you look at the ntp source for the gpsvme refclock, the first line has a
comment saying:

/* refclock_psc.c:  clock driver for Brandywine PCI-SyncClock32/HP-UX 11.X

It could of just been that the TrueTime GPS-VME became obsolete /
discontinued / no longer operational. There are quite a few old GPS products
that have suffered an untimely death only because they need a minor firmware
update, but the manufacturers are not willing to support something so old.
So rather than make a new driver number they just used one for an obsolete /
dead product?

Writing your own refclock is not an overly complex task, I've written a
couple custom ones myself. Examining the existing refclocks makes it almost
self explanatory. I wonder when the last reflock was actually added to the

Anyhow, I don't think anyone would object to creating a wiki page for the
Brandywine / driver 21. Though you might want to make some note that it
*used* to be the TrueTime GPS-VME driver so other's will know what is going



>From ntp-4.2.0 to ntp-4.2.2, more precisely with changeset 1.1152.9.9,
>the refclock_gpsvme.c changed from a reference clock driver for a True
>Time GPS-VME to a reference clock driver for a completely different
>and totally not compatible Brandywine Communications PCI-SyncClock32.
>Does anybody know the history about this? Is seems like the people at
>USNO did not have a need for the True Time GPS-VME anymore, but
>changed to the PCI-SyncClock32 and reused the refclock_gpsvme.c and
>the assigned driver number 21 to make a driver for it.
>Did nobody else used the TrueTime GPS-VME driver? Apparently not.
>Why not assigning a new driver number for PCI-SyncClock32 and add a
>For me, it is quite confusing. For example
>http://support.ntp.org/bin/view/Support/ConfiguringRefclocks at the
>section "by refclock driver number" has at number 21 a link to the
>True Time serial based refclock, which has nothing to do with either
>True Time GPS-VME or PCI-SyncClock32.
>If nobody objects, I'd like to add a page for the PCI-SyncClock32 at
>number 21.
>Rode Biet

More information about the questions mailing list