From owner-freebsd-current  Wed Jul 30 10:41:27 1997
Return-Path: <owner-freebsd-current>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id KAA19287
          for current-outgoing; Wed, 30 Jul 1997 10:41:27 -0700 (PDT)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA19275;
          Wed, 30 Jul 1997 10:41:20 -0700 (PDT)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.5/8.6.9) id DAA26172; Thu, 31 Jul 1997 03:35:52 +1000
Date: Thu, 31 Jul 1997 03:35:52 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199707301735.DAA26172@godzilla.zeta.org.au>
To: cgull@smoke.marlboro.vt.us, sos@sos.freebsd.dk
Subject: Re: code talks:  announcing EIDE bus master patches
Cc: freebsd-current@FreeBSD.ORG, se@FreeBSD.ORG
Sender: owner-freebsd-current@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

>Well, bonnie reports an increase in CPU load and some increase in I/O
>speed on the hack box I developed the code on, while a little
>spin-loop-at-idle-priority hack i coded up shows an improvement in
>available CPU that it can consume.  My conclusion: bonnie's CPU load
>numbers are useless, at least for IDE drives.

Bonnie doesn't report interrupt overhead.  FreeBSD does too good a job
of separating interrupt overhead from syscall overhead, and there is no
system call for reporting the interrupt overhead.  The overhead is easy
to see using `systat -vmstat' and easy to account for provided there
is only one process causing most of the interrupts (the time for a few
hundred normal interrupts is insignificant except on slow machines).
This is the same setup as is required for the spinloop method.

Bonnie's CPU load numbers are also fairly useless for SCSI drives :-).
They measure mainly the driver, filesystem and buffer cache overheads.
DMA overhead is hard to measure, but the spinloop method works well for
giving the total overhead.

Bruce