> >    My "old" 16-bit SMC Elite (without the 790 NIC) gives these errors a
> > few times per day with no harmful side effects in NetBSD-0.9.  The newer
> > 16-bit SMC cards (with the 790 NIC) do not give me the errors in 
> > NetBSD-1.0_BETA (and they too perform flawlessly).  There are two
> > variables changing here, both of which are relevant to the ring buffer
> > overruns.  Perhaps you should wait and install NetBSD-1.0 when it is
> > released, and if you still get serious errors, switch to a later rev board.
> > (By "serious", I mean something other than the occasional log message about
> > overruns. :) )
> > 
> > Dave Deaven <deaven@ishmael.amslab.gov>

> Well, my cards were purchased about a month and a half ago, so they
> are quite new.  I don't know how to check about the 790 NIC however.
> The serious error is the speed that is lost.  I can not have 30 second
> delays on a production server that many people will use.  Does the 1.0
> release address this problem?  When will it be released?  Will the
> Ultra cards be supported?  Is there a patch for 0.9?

I have a 8013 (Elite) card in front of me, and I remember that it had
ring buffer complaints a few months ago (so I switched back to an 8003
in my netbsd-current-as-of-a-few-months-ago box)

As to the 8216 (Ultra); I've got two boxes (one netbsd-sep-5 and the
other a dos box), both with 8216's in them. Works like a charm, and
great performance too! So, yes, Netbsd-current supports the ultra
(tested in both a 486-40 and a 386-40 machine), and has done since at
least early august (when I first put an 8216 in; before then I assume it
was supported, just that I never had one to test.)

--
Luke Mewburn, <lm@cpr.itg.telecom.com.au>
`Think of it as Evolution in Action.' - "Oath of Fealty", Niven & Pournelle