From owner-freebsd-current  Tue Sep 17 20:21:21 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id UAA08765
          for current-outgoing; Tue, 17 Sep 1996 20:21:21 -0700 (PDT)
Received: from GndRsh.aac.dev.com (GndRsh.aac.dev.com [198.145.92.241])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id UAA08720;
          Tue, 17 Sep 1996 20:21:11 -0700 (PDT)
Received: (from rgrimes@localhost) by GndRsh.aac.dev.com (8.7.5/8.7.3) id UAA01366; Tue, 17 Sep 1996 20:21:07 -0700 (PDT)
From: "Rodney W. Grimes" <rgrimes@GndRsh.aac.dev.com>
Message-Id: <199609180321.UAA01366@GndRsh.aac.dev.com>
Subject: Re: RAM parity error
In-Reply-To: <199609180222.TAA09684@silvia.HIP.Berkeley.EDU> from Satoshi Asami at "Sep 17, 96 07:22:06 pm"
To: asami@FreeBSD.org (Satoshi Asami)
Date: Tue, 17 Sep 1996 20:21:06 -0700 (PDT)
Cc: current@FreeBSD.org
X-Mailer: ELM [version 2.4ME+ PL25 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-current@FreeBSD.org
X-Loop: FreeBSD.org
Precedence: bulk

>  * If you have ``logic parity'' instead of true parity you have defeated
>  * 75 % of the purpose of even having parity on memory.  
> 
> 75%?  I thought that was 100% ;)  (Ok, maybe it can check the memory
> bus failure, but how often would that happen?)
> 
>  * 							 I would highly
>  * encorage you to try a swapout with some ``real parity'' memory and
>  * more than likely watch your problem go away...
> 
> I'd never buy a "logic parity" board myself, but unfortunately, these
> machines are donated to our project (they aren't even supposed to be
> running FreeBSD!) so I can't change anything in them. :(
> 
> So, do you think this is a memory problem? 

Yes, ``memory parity error'' == hardware fault in the memory subsystem.

> Is there some other test I
> can run, other than stressing the SCSI system? 

Yes, ``make world''.

> (It doesn't crash when
> I run the "fast bcopy" benchmark, and I think that thing stresses the
> memory system a lot....)

That is only stressing the memory system from the ``host'' port, not
from the PCI bus master port.  Interactions between the two is what
really streeses the memory system to the limits.

> 
> Satoshi
> 


-- 
Rod Grimes                                      rgrimes@gndrsh.aac.dev.com
Accurate Automation Company                 Reliable computers for FreeBSD