| *Really* horrible; lost 45 minutes over the course of 9 hours and ntpd
| seemed to be primarily responsible for that badness... The clock in that
| box is nowhere near as bad by itself.

Yup, that's horrible, and something bad is clearly happening.

What's been syslogged by ntpd?

What does sysctl kern.clockrate say?

Have you let the box simply free-run without ntpd?

Alternatively, have you booted up such that ntpd free-runs,
with a bad drift value?

I'm quite satisfied that my userland + kernel, at least, is
quite happy ntp-wise.

	Sean.