From owner-freebsd-scsi Sat May 4 17:56:04 1996 Return-Path: owner-freebsd-scsi Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id RAA22922 for freebsd-scsi-outgoing; Sat, 4 May 1996 17:56:04 -0700 (PDT) Received: from hauki.clinet.fi (root@hauki.clinet.fi [194.100.0.1]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id RAA22863 for ; Sat, 4 May 1996 17:56:00 -0700 (PDT) Received: from cantina.clinet.fi (root@cantina.clinet.fi [194.100.0.15]) by hauki.clinet.fi (8.7.5/8.6.4) with ESMTP id DAA18313; Sun, 5 May 1996 03:55:53 +0300 (EET DST) Received: (hsu@localhost) by cantina.clinet.fi (8.7.3/8.6.4) id DAA13322; Sun, 5 May 1996 03:55:53 +0300 (EET DST) Date: Sun, 5 May 1996 03:55:53 +0300 (EET DST) Message-Id: <199605050055.DAA13322@cantina.clinet.fi> From: Heikki Suonsivu To: hdalog@zipnet.net Cc: freebsd-scsi@freebsd.org In-reply-to: Peter Dufault's message of 5 May 1996 00:52:25 +0300 Subject: Re: Buffered Writes? Organization: Clinet Ltd, Espoo, Finland References: <199605042133.RAA20783@hda.com> Sender: owner-freebsd-scsi@freebsd.org X-Loop: FreeBSD.org Precedence: bulk From: Peter Dufault > Does anybody know how to or have a utility to turn on the > write cache on a barracuda? I have a wide barracuda on a NCR 825 and You would use the "-e" option of the mode page editor to edit mode page 8 and set the "WCE" (write cache enable) field to 1 to enable write back cacheing. Do "man 8 scsi". This will be similar to the async update of your file system meta data - you will introduce the problem of deferred errors, which presently will give you an error on the wrong disk transfer and treat the failed transfer as successful, and you'll have the possibility of not writing out the buffer on the disk in the case of power failure (unless your drive can write during spindown). queue algorithm modifier usually is "write in issue order" so just setting WCE to 1 does not do cause any damage (and nor it usually improves the performance if the load is continous like news disk). But the more important issue is that I still haven't seen a disk which actually *does* have a write cache (pointers welcome :). Easy test is to write data to every second block on the disk. If there is a write cache, you should get more than block / ~70ms (one disk revolution) written, optimally you should get the disk platter speed / 2, but scsi handshake overhead usually makes this impossible. -- Heikki Suonsivu, T{ysikuu 10 C 83/02210 Espoo/FINLAND, hsu@clinet.fi mobile +358-40-5519679 work +358-0-4375360 fax -4555276 home -8031121