From owner-freebsd-hardware  Fri Feb 16 07:25:07 1996
Return-Path: owner-hardware
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id HAA09956
          for hardware-outgoing; Fri, 16 Feb 1996 07:25:07 -0800 (PST)
Received: from Sysiphos (Sysiphos.MI.Uni-Koeln.DE [134.95.212.10])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id HAA09747
          for <hardware@freebsd.org>; Fri, 16 Feb 1996 07:23:20 -0800 (PST)
Received: by Sysiphos id AA04068
  (5.67b/IDA-1.5 for hardware@freebsd.org); Fri, 16 Feb 1996 16:21:11 +0100
Message-Id: <199602161521.AA04068@Sysiphos>
From: se@zpr.uni-koeln.de (Stefan Esser)
Date: Fri, 16 Feb 1996 16:21:10 +0100
In-Reply-To: ejc@nasvr1.cb.att.com
       "wb L2 cache improves sp3g by 10%" (Feb 16,  8:39)
X-Mailer: Mail User's Shell (7.2.6 alpha(2) 7/9/95)
To: ejc@nasvr1.cb.att.com
Subject: Re: wb L2 cache improves sp3g by 10%
Cc: hardware@freebsd.org
Sender: owner-hardware@freebsd.org
Precedence: bulk

On Feb 16,  8:39, ejc@nasvr1.cb.att.com wrote:
} Subject: wb L2 cache improves sp3g by 10%
} Hello
}         Dirty tag sram improves sp3g performance by 10% - 11%, with a
} AMD 5x86 133 processor.  I know there are a few people on this list
} using this mainboard, so I thought I would pass the info on.  I have
} not done a make world yet.  All times are in seconds.

Thanks a lot. This information is much appreciated!

I've been thinking about buying a Tag RAM for some
days, but I still need to find a local source. What
is the exact chip type you bought (and what did you
pay ? Seems they are soem $20 over here ...)

Could you perhaps give "bytebench" a try ?

I've done several tests with WT and WB without dirty
tag RAM, and I'd be interested in numbers for WB with
that chip added. I found the AMD 5x86 to be 150% of
an DX2/66 (user and real time of a make world), and
the tag RAM would make that 165% !

Regards, STefan
-- 
 Stefan Esser, Zentrum fuer Paralleles Rechnen		Tel:	+49 221 4706021
 Universitaet zu Koeln, Weyertal 80, 50931 Koeln	FAX:	+49 221 4705160
 ==============================================================================
 http://www.zpr.uni-koeln.de/~se			  <se@ZPR.Uni-Koeln.DE>