From owner-freebsd-hackers  Sat Nov 30 04:07:37 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id EAA09562
          for hackers-outgoing; Sat, 30 Nov 1996 04:07:37 -0800 (PST)
Received: from cyclone.degnet.baynet.de (cyclone.degnet.baynet.de [194.95.214.129])
          by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id EAA09530;
          Sat, 30 Nov 1996 04:07:28 -0800 (PST)
Received: from nada (ppp5 [194.95.214.135]) by cyclone.degnet.baynet.de (8.6.12/8.6.9) with SMTP id OAA14555; Sat, 30 Nov 1996 14:05:46 +0100
Message-ID: <32A03D57.6274@degnet.baynet.de>
Date: Sat, 30 Nov 1996 12:57:43 -0100
From: Darius Moos <moos@degnet.baynet.de>
Reply-To: moos@degnet.baynet.de
X-Mailer: Mozilla 3.0Gold (Win95; I)
MIME-Version: 1.0
To: "John S. Dyson" <toor@dyson.iquest.net>
CC: mango@staff.communique.net, hackers@freefall.freebsd.org,
        hardware@freefall.freebsd.org
Subject: Re: Adaptec UW controller and Seagate Elite performance ?
References: <199611300444.XAA09675@dyson.iquest.net>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

One thing that stroke me some time ago was the termination of the
SCSI-bus. After changing from passive to active termination, the
speed of my disks doubled !!!

Darius Moos.


John S. Dyson wrote:
> 
> >
> > # bonnie -s 40        (akira, http running)
> >    -------Sequential Output-------- ---Sequential Input-- --Random--
> >    -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
> > MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
> > 40  3271 34.8  3085 13.0  1559  8.3 10330 97.1 44763 95.7 991.0 20.4
> >
> > and
> > IOZONE performance measurements:
> >         2039008 bytes/second for writing the file
> >         34861747 bytes/second for reading the file
> >
> > It bothered me that:
> > 1- reading is 20 times faster than writing
> 
> You are seeing the results of caching.  The system remembers what
> is written so that it doesn't have to unnecessarily re-read the
> data.
> 
> >
> > 2- 2Mbytes/s for writing seems slow, even if this was a scsi-2 and not
> > an
> > scsi fast and wide drive.
> >
> I don't know why your iozone write perf is so slow, unless you are writing
> 512 bytes at a time.  Note also, that you might not have the write
> behind caching enabled on your drive.
> 
> Try the following commands:
> 
> # iozone auto
> and
> # iozone 40 8192
> and
> # iozone <size-of-memory*2> 8192
> 
> and let me know the results.
> 
> John
> dyson@freebsd.org