On Mon, 13 Mar 2000, Simon Baker wrote:

> At 11:33 13/03/00 -0800, Bill Studenmund wrote:
> >On Mon, 13 Mar 2000, Simon Baker wrote:
> >
> > > BTW, you should be aware that mac's are notoriously bad for keeping
> > > time.  See http://www.macbsd.com/macbsd/macbsd-docs/faq/faq-3.html#ss3.17
> > > for an explanation....
> >
> >Uhm, that FAQ is for m68k macs. Interrupt handling on powermacs is
> >different, and the clock should be able to run at a higher interrupt
> >level.
> >
> >If it's systematically sucking on a powermac, we should look at the
> >problem. :-)
> 
> Then I say we have a problem....
> 
> My mac is losing ~0.13 of a second per 20 mins, then I'd say I have fairly 
> serious issue.  And if everybody else has....

There's a bus speed multiplication routine bug at work, in all likelihood.
Clock drift should be exactly constant, as it is a scaling bug, not a
load-related or interrupt-related bug.  I'll let the people who are
familiar with NetBSD's clock code dig for the issue.  ;-)


David