From owner-freebsd-current Wed Jan 26 16:31:15 2000 Delivered-To: freebsd-current@freebsd.org Received: from relativity.student.utwente.nl (wit389306.student.utwente.nl [130.89.234.166]) by hub.freebsd.org (Postfix) with ESMTP id 4BBA115479 for ; Wed, 26 Jan 2000 16:31:13 -0800 (PST) (envelope-from djb@wit389306.student.utwente.nl) Received: by relativity.student.utwente.nl (Postfix, from userid 1001) id 3E1131E32; Thu, 27 Jan 2000 01:31:25 +0100 (CET) Date: Thu, 27 Jan 2000 01:31:25 +0100 From: "Dave J. Boers" To: Devin Butterfield Cc: Donn Miller , current@FreeBSD.ORG Subject: Re: pcm - stutters Message-ID: <20000127013125.A49548@relativity.student.utwente.nl> Reply-To: djb@relativity.student.utwente.nl References: <388E1C6B.489B84D2@cvzoom.net> <00012520102300.02605@dbm.wireless.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 1.0i In-Reply-To: <00012520102300.02605@dbm.wireless.net>; from dbutter@wireless.net on Tue, Jan 25, 2000 at 08:05:48PM -0800 Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG It is rumoured that Devin Butterfield had the courage to say: > I too notice these problems of mpg123 skipping during disk activity or X > graphics ops but I have always had these problems, both with -STABLE and > -CURRENT. I notice this with xmms too. So this is nothing new. Isn't this simply a typical issue of IDE hardware? I too notice xmms skipping on heavy disk activity (typically the find command that runs from cron at 01:59). This happens even though I have two processors and a disk that can do 16 Mb/sec on UDMA66. One would expect such a system to be able to do a find and play mp3's simultaneously. However, AFAIK the IDE hardware typically handles only one request at a time and handles all requests sequentially, contrary to scsi. It may thus happen that the read requests from xmms, small as they may be, get delayed too long and unnecessary. I'm not an expert on this, but I believe scatter/gather is the way scsi handles this problem in hardware. Perhaps the ata driver could be made to do something similar on the driver level. I have experienced significantly better responsiveness during high disk usage of much slower systems that are running with scsi disks only. In fact, I have a 486 that has better response times (it has an EISA bus and adaptec 2740 scsi with a disk that can do -- and does! -- 8 Mb/sec disk->memory) than my Abit BP6 dual celeron UDMA66 system during the cron job at 1:59. Regards, Dave Boers. -- djb@relativity.student.utwente.nl To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message