From owner-freebsd-hackers Fri Aug 18 14:42:10 2000 Delivered-To: freebsd-hackers@freebsd.org Received: from feral.com (feral.com [192.67.166.1]) by hub.freebsd.org (Postfix) with ESMTP id 6055537B423 for ; Fri, 18 Aug 2000 14:42:08 -0700 (PDT) Received: from zeppo.feral.com (IDENT:mjacob@zeppo [192.67.166.71]) by feral.com (8.9.3/8.9.3) with ESMTP id OAA00888; Fri, 18 Aug 2000 14:37:45 -0700 Date: Fri, 18 Aug 2000 14:33:58 -0700 (PDT) From: Matthew Jacob Reply-To: mjacob@feral.com To: Soren Schmidt Cc: Parag Patel , Robert Sexton , Jonathan Laventhol , freebsd-hackers@FreeBSD.ORG Subject: Re: Cheap 1000Gbyte machine In-Reply-To: <200008182123.XAA80962@freebsd.dk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Fri, 18 Aug 2000, Soren Schmidt wrote: > It seems Parag Patel wrote: > > On Fri, 18 Aug 2000 15:57:48 EDT, Robert Sexton wrote: > > > > > >Can IDE drives release the bus during seeks? Historically thats been > > >the big advantage of SCSI: Two IDE drives are no faster than one IDE > > >drive, while SCSI scales in performance. > > > > How about a single IDE (master) drive per controller, that is, no slave > > drives at all? This would halve the number of available drives per card > > (or require twice the number of cards) but essentially would use the PCI > > bus as a multiplex bus between drives instead of SCSI. > > > > You're still pretty much stuck waiting for one command to finish before > > another can be sent to a drive though, unlike SCSI drives. Still a lot > > cheaper tho'. > > Not really, if the ATA drives are configured as single masters, they can > run simultaniously, but you will still have the PCI bus speed limit :) > Which BTW is around 120MB/s as some overhead goes to keep the system > running (yes I have measured that :) ) > > And if you have modern ATA disks, they support tagged queuing as well, > have had that running in my lab too... Yes. I've seen tag depths of 32. Still not as much as SCSI, but 32 will saturate most spindles still (I've been having to work on some non-FBSD ATA and S.M.A.R.T. stuff lately....) To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message