From owner-freebsd-hardware Mon Nov 3 01:27:58 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id BAA20697 for hardware-outgoing; Mon, 3 Nov 1997 01:27:58 -0800 (PST) (envelope-from owner-freebsd-hardware) Received: from bubble.didi.com (sjx-ca35-05.ix.netcom.com [204.31.236.133]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id BAA20687 for ; Mon, 3 Nov 1997 01:27:55 -0800 (PST) (envelope-from asami@vader.cs.berkeley.edu) Received: (from asami@localhost) by bubble.didi.com (8.8.7/8.8.7) id VAA07343; Sun, 2 Nov 1997 21:44:24 -0800 (PST) (envelope-from asami) Date: Sun, 2 Nov 1997 21:44:24 -0800 (PST) Message-Id: <199711030544.VAA07343@bubble.didi.com> To: pete@sms.fi CC: michaelv@MindBender.serv.net, mike@sentex.net, dg@root.com, hardware@FreeBSD.ORG In-reply-to: <199711011833.UAA00692@silver.sms.fi> (message from Petri Helenius on Sat, 1 Nov 1997 20:33:28 +0200 (EET)) Subject: Re: fxp0 and full duplex From: asami@cs.berkeley.edu (Satoshi Asami) Sender: owner-freebsd-hardware@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk * From: Petri Helenius * > Do you know, for a fact, that MS IP stacks (from Win95 thru NT Server) * > are significantly less efficient than the BSD variety? Or are you * > just slamming MS for the hell of it? * > * While FreeBSD with P166 can easily fill a 100Mbps pipe, same hardware * running NT or 95 comes up to around 30-45Mbps. There are multiple * independent studies available on the web to confirm this story, with * FreeBSD, Solaris x86, etc... We have observed the following on two P6-200 (96MB memory, 14-disk (IBM DCHS) striped array, Intel EE Pro 100/B) connected via crossover cables: FreeBSD server - Windows NT client: 4.4MB/s FreeBSD server - FreeBSD client: >10MB/s That is the speed of ftp transfer of large files from disk to disk. (Of course, this could be the filesystem and not the network driver, but it's just a single datapoint anyway.) Satoshi