From owner-freebsd-hackers  Mon Oct  9 18:58:03 1995
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id SAA01782
          for hackers-outgoing; Mon, 9 Oct 1995 18:58:03 -0700
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.34])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id SAA01773
          for <hackers@FreeBSD.org>; Mon, 9 Oct 1995 18:57:56 -0700
Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id LAA32670; Tue, 10 Oct 1995 11:47:48 +1000
Date: Tue, 10 Oct 1995 11:47:48 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199510100147.LAA32670@godzilla.zeta.org.au>
To: chuckr@eng.umd.edu, se@zpr.uni-koeln.de
Subject: Re: VLB Disk Controllers
Cc: hackers@FreeBSD.org
Sender: owner-hackers@FreeBSD.org
Precedence: bulk

>OK, I've been watching the posted results, many have as little variance 
>as yours, but many (where multiple runs were shown) had the huge variance
>mine had.  Would like to know why my checks to /dev/sd0 are all over the 
>board, when checks to rsd0 are fairly stable at 1180-1190.

Testing /dev/sd0 tests the buffer cache and shouldn't go anywhere near
the driver or the drive.  There may be more variance because running
the program (not to mention other processes) disturbs the buffer cache.
/dev/sd0 is about twice as slow as /dev/zero here, but the "command"
overhead is about twice as small for /dev/sd0.

Bruce