From owner-freebsd-current  Wed Sep 10 13:42:17 1997
Return-Path: <owner-freebsd-current>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id NAA25858
          for current-outgoing; Wed, 10 Sep 1997 13:42:17 -0700 (PDT)
Received: from Octopussy.MI.Uni-Koeln.DE (Octopussy.MI.Uni-Koeln.DE [134.95.166.20])
          by hub.freebsd.org (8.8.7/8.8.7) with SMTP id NAA25843
          for <current@freebsd.org>; Wed, 10 Sep 1997 13:42:13 -0700 (PDT)
Received: from x14.mi.uni-koeln.de ([134.95.219.124]) by Octopussy.MI.Uni-Koeln.DE with SMTP id AA20143
  (5.67b/IDA-1.5 for <current@FreeBSD.ORG>); Wed, 10 Sep 1997 22:42:08 +0200
Received: (from se@localhost) by x14.mi.uni-koeln.de (8.8.7/8.6.9) id WAA01004; Wed, 10 Sep 1997 22:25:41 +0200 (CEST)
X-Face: "<d]#=8pzx);RzeqSKI86OVa7=!0/(uRa.+B.9Z9\eNUn@UG?!`y7yt2dFNn%k4'.}](uE%
 yCO)$e&Y1%3EO~ifu6Q-#YUM&JZ't,}JkPnAz,8Dj33u%@GBi%[Y#LHz$]h7a<p4)-jKI7~sKjlP-^
 EvA[G;]v&0]W!EL%shs,{7x0|oqN4YVIs5,NI#,V{9"WF):5&RkOhyj*#-IAG}Tnu;YCF,d
Message-Id: <19970910222540.16996@mi.uni-koeln.de>
Date: Wed, 10 Sep 1997 22:25:40 +0200
From: Stefan Esser <se@FreeBSD.ORG>
To: Mats Lofkvist <mal@algonet.se>
Cc: current@FreeBSD.ORG
Subject: Re: lousy disk perf. under cpu load (was IDE vs SCSI)
References: <19970909231006.58788@mi.uni-koeln.de> <199709101259.OAA02564@kairos>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.74
In-Reply-To: <199709101259.OAA02564@kairos>; from Mats Lofkvist on Wed, Sep 10, 1997 at 02:59:38PM +0200
Sender: owner-freebsd-current@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

On Sep 10, Mats Lofkvist <mal@algonet.se> wrote:
> Reassigning interrupts doesn't seem to be as easy as I hoped :-(
> 
> I have an Intel Providence motherboard with the ahc and fxp
> builtin (on pci) and an separate ncr for the narrow devices.
> I haven't been able to figure out how to change the interrupts,
> can't find anything in the bios letting me set them explicitly.
> (Setting plug-n-play to "bios" mode only lets me reserve irq's
> used by non-pnp cards, not reassign anything on pci. With pnp in
> the "pnp os" mode nothing can be changed.)

Perhaps you can free IRQs that are currently reserved, 
with the effect that the BIOS uses them for PCI devices ?

Do you have access to any other system, on which you could 
repeat your tests ? 

I guess the problem is specific to your motherboard, and
even if you moved the controller cards over, you would not
see the high impact you report ...


BTW: I see on my system (486/133, NCR53c810, 2GB Atlas I):

# dd if=/dev/rsd0a of=/dev/null bs=4k
3767+0 records in
3767+0 records out
15429632 bytes transferred in 4.135790 secs (3730758 bytes/sec)

# loop & dd if=/dev/rsd0a of=/dev/null bs=4k
6144+0 records in
6144+0 records out
25165824 bytes transferred in 6.925719 secs (3633677 bytes/sec)

# loop & loop & loop & loop & dd if=/dev/rsd0a of=/dev/null bs=4k
6144+0 records in
6144+0 records out
25165824 bytes transferred in 7.237373 secs (3477204 bytes/sec)

There is a slight decrease of throughput, but even with 4
loop processes (just a "main() {while(1);}"), it is less
than 10%. The interrupt rate is some 920/s with no load,
and decreases to some 850/s under load ...

Regards, STefan