From owner-freebsd-current Sun Aug 25 23:09:59 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id XAA20197 for current-outgoing; Sun, 25 Aug 1996 23:09:59 -0700 (PDT) Received: from GndRsh.aac.dev.com (GndRsh.aac.dev.com [198.145.92.241]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id XAA20186 for ; Sun, 25 Aug 1996 23:09:55 -0700 (PDT) Received: (from rgrimes@localhost) by GndRsh.aac.dev.com (8.6.12/8.6.12) id XAA18275; Sun, 25 Aug 1996 23:08:11 -0700 From: "Rodney W. Grimes" Message-Id: <199608260608.XAA18275@GndRsh.aac.dev.com> Subject: Re: Speedingup the "worldstone" To: michaelv@MindBender.serv.net (Michael L. VanLoon) Date: Sun, 25 Aug 1996 23:08:11 -0700 (PDT) Cc: michaelh@cet.co.jp, imp@village.org, current@freebsd.org In-Reply-To: <199608260143.SAA12137@MindBender.serv.net> from "Michael L. VanLoon" at "Aug 25, 96 06:43:53 pm" X-Mailer: ELM [version 2.4ME+ PL11 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > > >I think you've already done things nicely by spreading out the disk i/o, > >compiles are also CPU intensive so the x486 looks like your bottle neck. > >I'm skepical that you can really get the disks singing in harmony to take > >advantage of Ultra Wide. Though it's generally a good thing to have > >advances in bus architectures so that it puts pressure on the peripherals > >to catch up. > > I can tell you, for a fact, that you won't get Ultra Wide bandwidth > out of a 486 on real filesystem data. The 486 design just doesn't > have enough power. The best you'll probably do, on real-world use, is > maxing out fast SCSI (10MB/s), and then only if you tune well. More correctly you won't get more than 6 to 8MB/s on ISA based 486 systems, or 12 to 17MB/s on EISA based 486 systems, and about the same on PCI based 486 systems. The ISA based is limited by how fast you can bus master on the ISA bus, and few boards will do about 8MB/s. The EISA/PCI systems are limited by the bcopy speed of the 486, the only 486 board that I could get to hit 17MB/s was the ASUS PCI/I-486SP3G, and it could do it because it uses 2 way interleaved memory, and a decent PCI chip set. These are real live test results from my work 2 years ago on a stripping disk drive that was finally tuned to the customers application, I couldn't break the 17MB/s barrier until we changed the design to PCI/Pentium. -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation Company Reliable computers for FreeBSD