From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from colo.lackof.org (colo.lackof.org [198.49.126.79]) by dsl2.external.hp.com (Postfix) with ESMTP id 2194E4840 for ; Thu, 27 Nov 2003 11:49:41 -0700 (MST) Date: Thu, 27 Nov 2003 11:49:39 -0700 From: Grant Grundler To: "M. Grabert" Cc: parisc-linux@lists.parisc-linux.org Subject: Re: [parisc-linux] C3000 and 2.6 - need config help Message-ID: <20031127184939.GC9746@colo.lackof.org> References: <20031127042103.GC1321@colo.lackof.org> <20031127061557.GE1321@colo.lackof.org> <20031127181531.GB9746@colo.lackof.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: Sender: parisc-linux-admin@lists.parisc-linux.org Errors-To: parisc-linux-admin@lists.parisc-linux.org List-Help: List-Post: List-Subscribe: , List-Id: parisc-linux developers list List-Unsubscribe: , List-Archive: On Thu, Nov 27, 2003 at 06:21:26PM +0000, M. Grabert wrote: > On Thu, 27 Nov 2003, Grant Grundler wrote: > > > On Thu, Nov 27, 2003 at 03:16:08PM +0000, M. Grabert wrote: > > > The max. transfer rate of any 100BT network is around 5MByte/s. > > > > 100BT is capable of 90-95Mb/s ~= 11MB/s. > > That's theoretical, not really practically possible/feasible :-/ it's possible with a switched network. It's not possible with regular hub where one has to deal with collisions. Full duplex switched networks obsolete token passing as a method to avoid packet collisions. > > There's a website someplace that has 100BT performance numbers > > for all sorts of card/OS combinations. > > > > I've measured ftp transfers > 10MB/s between ia64 and ia32 machines > > over 100BT link. > > ftp? uncompressed? Well I'm impressed ... > The fastest 100MBit link I ever had was 7MBytes/s. This is an excellent paper on routing: http://robur.slu.se/Linux/net-development/jamal/FF-html/ Peak throughput (in one port of a 4-port tulip card and out another) is about 50Mb/s because the single CPU is saturated. He does better with an SMP machine. Jamal later got 20% perf improvement (measured in packets per second) using MMIO instead of IO port space . But still saturating the CPU. Jamal was confident at the time he could route at link rate (95Mb/s) with 64byte packets on a faster SMP machine. grant