> Which performance difference is this?  Was OS/2 supposed to faster
> than NT, or slower?  Which magazine did he read? :-)  Either way it's
> probably marketing hype, one direction or the other.

    Well, this friend has always been a vehement OS/2 supporter, and
claims that NT and OS/2 have common ancestry (if NT didn't actually COME
from OS/2).  But he admitted that NT is considerably faster, and most of
it is due to it using dynamic caches, as opposed to OS/2's fixed caches.
 Point taken, though.

> FreeBSD has a merged buffer cache.  Supposedly it works really well.
> But it took a lot of work for them to get it that way.  Various people
> have expressed interest in porting it over to NetBSD, but I haven't
> seen anything serious come of it.  Supposedly, it will be a lot of
> work.

    I'd be willing to help, I'd like to gain some experience in writing
code on the OS-level, and I'd sure like to be able to contribute
something to the NetBSD community, it's helped me so many times already.

> It depends on what you're doing if that will be a big win for you.
> You can also increase the percentage of memory used in NetBSD for
> buffer cache.  The difference is that it's fixed.  And the memory you
> give to buffer cache takes away from memory you can run programs in
> (the merged cache does this dynamically -- and there are many ways to
> get this formula wrong on a merged buffer cache system).

    Yeah, I can imagine.  It's just that when I had 32, or especially
16MB, it was clear that I pretty much always had less RAM than things
needed.  But now that I have 64 (80 if I take the time to compile a
kernel...  =), last I looked after I was running all of my normal stuff,
there was still around 25MB free (if I understand what systat's things
say).  I'd like a decent chunk of whatever isn't being used to be cache
just because it seems like a waste otherwise.  But if I ever have the
need for all my RAM, I'd like it to be able to meet that need as well...
 =)


ANYHOW...  to the actual subject...  =)

    So I tried the original RAM configuration as I said:  my original 2
8MB SIMMs in the first two SIMM sockets.  In the past 24 hours it's
crashed 3 times (all logging as "reboot after panic: vrele: ref cnt",
and all occuring while / immediately after logging out of an xdm
session).  And honestly, this was the first time I was *glad* to see it
crash, because at least it wouldn't seem to suggest that my new 32MB
SIMMs are bad (phew!  =)  The problem becomes what to test if it's not
RAM.  I haven't changed anything else hardware-wise in months.  Given
that I think 1.2 binaries for i386 are out now (????), I think I'll try
reinstalling my NetBSD binaries, and going with a GENERIC kernel for a
few days, and see if that helps matters.  I could set my clock back and
try an older X, but most of my machine is kerberized / AFSized and such,
so if my clock's as much as 5 minutes off CMU's Kerberos Master, I'll
have a whole slew of new problems to deal with (assuming I can even log
in!!!  =)  So before that, I'll make sure that a non-accelerated X
server works (like SVGA instead of S3), and maybe even drop back to
3.1.2 (release), where hopefully I can use SVGA (my video card isn't
supported under S3 until 3.1.2D, so there is no "known stable" version
of XFree86 that I can even use!!).  I don't know what to hope the
problem is now.

> -----------------------------------------------------------------------------
>   Michael L. VanLoon                           michaelv@MindBender.serv.net
>         --<  Free your mind and your machine -- NetBSD free un*x  >--
>     NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3,
>         Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32...
>     NetBSD ports in progress: PICA, others...
> -----------------------------------------------------------------------------

    Ya know, if the recent posts on source-changes mean anything, you
probably can tack PowerPC in that list somewhere...  =)
    Which reminds me, I see all of the posts on source-changes, yet very
few of those changes make it out to the CVS commits and the sup server. 
In fact, often even most of the things in the CVS commits don't actually
get sup'ed.  Yet I have allsrc, security, othersrc, and doc in my
supfile (all current).  Am I missing something, or do most of these
changes not make it to the user world?

    Thanks again!

                                                          Will Ferry

--------------------------------------------------------------------------
William O Ferry <woferry@CMU.EDU> | finger:  woferry@WarpDrive.RES.CMU.EDU
talk:  finger for online status   | http://warpdrive.res.cmu.edu/~woferry/
--------------------------------------------------------------------------