Re: MAJOR trouble with 32 MB SIMMs in UDB

Lee Burnside (
Tue, 05 Nov 1996 13:30:20 -0600 (CST)

On Tue, 5 Nov 1996, Joshua M. Thompson wrote:

> On Tue, 5 Nov 1996, Lee Burnside wrote:
> > We've had great luck using 32MB SIMMS in our UDB's. Admittedly, an error
> > in addressing (when crossing over the 64MB boundary) pops up after a warm
> > boot, but a cold boot clears this right up.
> What kind of error do you get? Is it a machine check error? I only cold
> boot these machines and have never tried warm booting.

Most of the time we got the old reliable "machine check", though not
always. Occasionally a bad address access would occur, followed by a seg
fault. But the norm was a couple of machine checks (no more than three)
followed by a hard lock right around the time init would begin loading

> > You might make sure the SIMMS have no more than 24 chips on them; I don't
> > know about the UDB's but I've seen many boxen that won't map anything
> > with > 24 chips.
> Twelve chips per side. Looks like eight 4 MB x 4 chips and four 4 MB x 1
> chips.

Sounds OK to me.

> > I tried running INN under RH 4.0 as ELF last night; no problems that I
> > could see. Unfortunately, I didn't have much of a news spool set up, so
> > I'm not sure how memory intensive the feed was; the memory load seemed
> > fairly light.
> Our server processes several incoming and outgoing feeds and typically has
> a 50 - 60 MB footprint. The problem may only be showing up under a heavy
> load.

Actually, not too much greater than the temporary spool I had set up.

> > I didn't notice in your posting if you were sure the 8x36 SIMMS were
> > "true" parity. Logic parity SIMMS will cause most (if not all) AXP's to
> > croak.
> They are true parity. Also there are no PALs or other logic chips evident
> on the SIMMs to make me suspect that they are in fact logical parity
> chips.

Since I've lost your original post, I'm not sure what version of MILO
you're using. Our resident code-dog has had problems with MILO 1.3.76
and anything over 64MB, but neither he nor I have had problems with <=

Wish I could be of more help.

Lee Burnside --SysAdmin-- TTU Dept. of Physics
Physics Homepage:
University Homepage:

To unsubscribe: send e-mail to with
'unsubscribe' as the subject.  Do not send it to

Feedback | Store | News | Support | Product Errata | About Us | Linux Info | Search | JumpWords
No Frames | Show Frames

Copyright © 1995-1997 Red Hat Software. Legal notices