Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 5 Jan 1999 09:18:51 -0700 (MST)
From:      "Kenneth D. Merry" <ken@plutotech.com>
To:        ghelmer@scl.ameslab.gov (Guy Helmer)
Cc:        freebsd-smp@FreeBSD.ORG
Subject:   Re: ASUS P65UP5 Dual PPro problems
Message-ID:  <199901051618.JAA64014@panzer.plutotech.com>
In-Reply-To: <Pine.SGI.3.96.990105100936.20729D-100000@demios.scl.ameslab.gov> from Guy Helmer at "Jan 5, 99 10:11:55 am"

next in thread | previous in thread | raw e-mail | index | archive | help
Guy Helmer wrote...
> On Fri, 1 Jan 1999, Kenneth D. Merry wrote:
> 
> > Guy Helmer wrote...
> > > We are having trouble with a bunch of ASUS P65UP5 machines with dual
> > > 200MHz PPro's; each machine is configured exactly the same, with 256MB
> > > RAM, a Tulip Fast Ethernet interface and IDE disk drive.  Symptoms are
> > > that the machine will either freeze solid without any console message, or
> > > (according to top(1)) a process is running on CPU1 (and never changes from
> > > CPU1) but is not getting any CPU time (WCPU and CPU are both 0%).  This
> > > seems to happen randomly, but usually when the processes are doing network
> > > communication.
> > > 
> > > The kernel is FreeBSD SMP built from sources dated Nov 19 1998.  The
> > > machine's BIOS is set to MP spec 1.4.  The machines work fine under
> > > uniprocessor Linux 2.0.3x, but exhibit similar behavior with SMP Linux
> > > 2.0.3x or 2.1.x.
> > 
> > Well, FWIW, I have the same sort of machine (same motherboard and
> > processors), but running with -current from early December.  I haven't
> > had any trouble.
> > 
> > I've got SCSI disks, though, not IDE.  I do have two DEC Tulip based SMC
> > cards, though.
> > 
> > One question I have, though, is what kind of RAM you have in the machine?
> > i.e., what configuration.  I tried putting 256MB in my machine, using 8
> > 32MB (parity) SIMMs, but I wasn't able to keep it like that.  I got random
> > NMIs with 8 SIMMs on board.  I reduced it to 6 SIMMs (192MB), and the NMIs
> > stopped.
> 
> These machines have 256MB RAM using 8 32MB ECC SIMMs, ECC enabled in the
> BIOS setup.

I've heard (on one of the FreeBSD lists, probably -hardware, I think) that
enabling ECC knocks about 10% off memory performance.  It might be better
to just enable parity checking.

> > I'm fairly certain they weren't parity errors, since I've had bad memory on
> > other machines and FreeBSD would actually panic with a "ram parity error"
> > NMI.  The NMI panic message I got with these errors didn't state a specific
> > problem.
> > 
> > The SIMMs I have all have 24 chips on board, so they're within ASUS' stated
> > specs, but my guess is that I exceeded the load that the memory subsystem
> > could take.  The NMIs generally only occurred under high memory load.
> 
> Are the NMI's reported via syslog?  We've not seen any NMIs reported to
> the console on these systems...

NMI == Non Maskable Interrupt

NMI's cause panics.  If you've got DDB enabled, you can just continue on
after an NMI, but you may have weird things happen.


Ken
-- 
Kenneth Merry
ken@plutotech.com

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-smp" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199901051618.JAA64014>