From owner-freebsd-current  Sat Aug  9 01:24:45 1997
Return-Path: <owner-freebsd-current>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id BAA14123
          for current-outgoing; Sat, 9 Aug 1997 01:24:45 -0700 (PDT)
Received: from smtp1.xs4all.nl (smtp1.xs4all.nl [194.109.6.51])
          by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id BAA14118
          for <freebsd-current@freebsd.org>; Sat, 9 Aug 1997 01:24:42 -0700 (PDT)
Received: from asterix.xs4all.nl (root@asterix.xs4all.nl [194.109.6.11]) by smtp1.xs4all.nl (8.8.6/XS4ALL) with ESMTP id KAA10596 for <freebsd-current@freebsd.org>; Sat, 9 Aug 1997 10:24:34 +0200 (MET DST)
Received: from plm.xs4all.nl (uucp@localhost)
	by asterix.xs4all.nl (8.8.6/8.8.6) with UUCP id KAA28041
	for freebsd-current@freebsd.org; Sat, 9 Aug 1997 10:22:30 +0200 (MET DST)
Received: (from plm@localhost) by plm.xs4all.nl (8.8.6/8.7.3) id AAA16491; Sat, 9 Aug 1997 00:54:14 +0200 (MET DST)
To: freebsd-current@freebsd.org
Subject: Re: IDE vs SCSI was: flags 80ff works (like anybody doubted it)
References: <87k9hw723z.fsf@totally-fudged-out-message-id>
From: Peter Mutsaers <plm@xs4all.nl>
Date: 09 Aug 1997 00:54:13 +0200
In-Reply-To: "John S. Dyson"'s message of Thu, 7 Aug 1997 23:27:18 -0500 (EST)
Message-ID: <877mdw2smi.fsf@plm.xs4all.nl>
Lines: 81
X-Mailer: Gnus v5.4.25/Emacs 19.34
Sender: owner-freebsd-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>> On Thu, 7 Aug 1997 23:27:18 -0500 (EST), "John S. Dyson"
>> <toor@dyson.iquest.net> said:

    JSD> This is the result of my "slow" recent, but not the latest, greatest
    JSD> IDE drive (WD 4GB drive.):

    JSD> Command overhead is 88 usec (time_4096 = 348, time_8192 = 607)
    JSD> transfer speed is 1.57828e+07 bytes/sec

    JSD> dd if=/dev/rwd1 of=/dev/null count=1600 bs=64k
    JSD> 1600+0 records in
    JSD> 1600+0 records out
    JSD> 104857600 bytes transferred in 10.881267 secs (9636525 bytes/sec)

    JSD> This is the result of my Hawk, SCSI drive, with an NCR 815 interface: 

    JSD> Command overhead is 845 usec (time_4096 = 2071, time_8192 = 3297)
    JSD> transfer speed is 3.34201e+06 bytes/sec

    JSD> dd if=/dev/rsd0 of=/dev/null count=1600 bs=64k
    JSD> 1600+0 records in
    JSD> 1600+0 records out
    JSD> 104857600 bytes transferred in 27.336979 secs (3835742 bytes/sec)

Hmm, my NCR815 does:

~> dd if=/dev/rsd2 of=/dev/null count=1600 bs=64k
1600+0 records in
1600+0 records out
104857600 bytes transferred in 15.215823 secs (6891352 bytes/sec)

This is a seagate ST15150N.

But for Wide and/or Ultra SCSI you can expect better values.

More interesting is:

~> dd if=/dev/rsd0 of=/dev/null count=1600 bs=64k&;dd if=/dev/rsd0 of=/dev/null count=1600 bs=64k&
[2] 16410
[3] 16411

i.e. start two of them at the same time.
I get as result:

104857600 bytes transferred in 41.430049 secs (2530955 bytes/sec)

and 

104857600 bytes transferred in 41.437855 secs (2530478 bytes/sec)

So still >5MB/s 

What does IDE do in this case?

Another variant:

~> dd if=/dev/rsd0 of=/dev/null count=1600 bs=64k&; sleep 2; dd if=/dev/rsd0 of=/dev/null count=1600 bs=64k&
[2] 16415
[3] 16417

i.e. two of them, but with 2 seconds wait in between to make sure that
the two readers do not read blocks that are very close. Results are
much, much lower now:

104857600 bytes transferred in 113.343774 secs (925129 bytes/sec)

and 

104857600 bytes transferred in 113.277994 secs (925666 bytes/sec)

Less than 1 MB/s total throughput remains! Quite dissapointing. I
wonder how IDE behaves in this case.

Anyway, apart from performance, I mainly use SCSI because I have 3
disks, a CD-ROM and a tapestreamer all on 1 cable, all with good
drivers and very reliable.

-- 
 /\_/\
( o.o ) Peter Mutsaers  |  Abcoude (Utrecht), |  Trust me, I know
 ) ^ (  plm@xs4all.nl   |  the Netherlands    |  what I'm doing.