From owner-freebsd-hardware  Tue Feb 25 21:46:14 1997
Return-Path: <owner-hardware>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id VAA16778
          for hardware-outgoing; Tue, 25 Feb 1997 21:46:14 -0800 (PST)
Received: from dfw-ix15.ix.netcom.com (dfw-ix15.ix.netcom.com [206.214.98.15])
          by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id VAA16773
          for <freebsd-hardware@freebsd.org>; Tue, 25 Feb 1997 21:46:11 -0800 (PST)
Received: (from smap@localhost)
          by dfw-ix15.ix.netcom.com (8.8.4/8.8.4)
	  id XAA09414; Tue, 25 Feb 1997 23:44:47 -0600 (CST)
Received: from wck-ca5-07.ix.netcom.com(199.35.213.167) by dfw-ix15.ix.netcom.com via smap (V1.3)
	id sma009357; Tue Feb 25 23:44:27 1997
Received: (from asami@localhost) by silvia.HIP.Berkeley.EDU (8.8.5/8.6.9) id VAA01458; Tue, 25 Feb 1997 21:44:19 -0800 (PST)
Date: Tue, 25 Feb 1997 21:44:19 -0800 (PST)
Message-Id: <199702260544.VAA01458@silvia.HIP.Berkeley.EDU>
To: bde@zeta.org.au
CC: bde@zeta.org.au, msmith@atrad.adelaide.edu.au,
        freebsd-hardware@freebsd.org, ken@stox.pr.mcs.net
In-reply-to: <199702250918.UAA03535@godzilla.zeta.org.au> (message from Bruce Evans on Tue, 25 Feb 1997 20:18:42 +1100)
Subject: Re: Memory speed of P6-200 (256k)
From: asami@vader.cs.berkeley.edu (Satoshi Asami)
Sender: owner-hardware@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

I finally found what was wrong.  It was pilot error (shame on me! ;).

When I bought the motherboard, I went to the BIOS setup and called the 
menu item to reset everything to the default (like any smart boy will
do).  Well, the only problem was that I wasn't paying much attention
and didn't notice that there were two "default"s -- the "BIOS
defaults" and "setup defaults".  According to the manual (that I
finally read from beginning to end today), the "real" (optimized)
defaults are the latter; the first one is a fail-safe, debugging-mode
type of thing.  Duh.

So, I invoked the "setup defaults" and voila! I got >80MB/s.  I even
played around with the memory timings and managed to add a few MB/s.
(As Chris said, changing the "RAS Precharge" from 3T to 4T helped.)
Now I see something like:

>> dd if=/dev/zero of=/dev/null bs=1m count=100
100+0 records in
100+0 records out
104857600 bytes transferred in 1.198974 secs (87456107 bytes/sec)

There is no difference at all whether it's in ECC mode, parity mode or 
they are both disabled.  Quite interesting.

Anyway, so I'm happy now.  People, thanks for all the encouragement. :)

 * Yes, that was the original motivation for speeding up copyout() and
 * copyin().  Satoshi had lots of disk bandwidth (40MB/sec?) from multiple

More like >60MB/s, we had four F-W SCSI strings at one point.

 * controllers but couldn't use it all because copying alone was limited
 * to 40MB/sec.  He speeded it up to 70+MB/sec on a P5-Triton system by
 * using the FPU, and I speeded it up a few more MB/sec by fine tuning.

Yes.  With the new 80MB/s copyin/out, the max throughput of reads
of a ccd through the filesystem jumped from 21MB/s to 27MB/s using 4
Seagate Barracuda 15150WC's.

Satoshi