From owner-freebsd-current@FreeBSD.ORG Fri May 14 11:48:23 2004 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 4F55816A4CE for ; Fri, 14 May 2004 11:48:23 -0700 (PDT) Received: from mail2.speakeasy.net (mail2.speakeasy.net [216.254.0.202]) by mx1.FreeBSD.org (Postfix) with ESMTP id AB58F43D53 for ; Fri, 14 May 2004 11:48:22 -0700 (PDT) (envelope-from jhb@FreeBSD.org) Received: (qmail 28325 invoked from network); 14 May 2004 18:44:34 -0000 Received: from dsl027-160-063.atl1.dsl.speakeasy.net (HELO server.baldwin.cx) ([216.27.160.63]) (envelope-sender ) encrypted SMTP for ; 14 May 2004 18:44:34 -0000 Received: from 10.50.40.205 (gw1.twc.weather.com [216.133.140.1]) by server.baldwin.cx (8.12.11/8.12.11) with ESMTP id i4EIiRW3028206; Fri, 14 May 2004 14:44:27 -0400 (EDT) (envelope-from jhb@FreeBSD.org) From: John Baldwin To: freebsd-current@FreeBSD.org Date: Fri, 14 May 2004 14:44:54 -0400 User-Agent: KMail/1.6 References: In-Reply-To: MIME-Version: 1.0 Content-Disposition: inline Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <200405141444.54385.jhb@FreeBSD.org> X-Spam-Checker-Version: SpamAssassin 2.63 (2004-01-11) on server.baldwin.cx cc: Gerrit Nagelhout Subject: Re: 4.7 vs 5.2.1 SMP/UP bridging performance X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 14 May 2004 18:48:23 -0000 On Tuesday 04 May 2004 03:55 pm, Gerrit Nagelhout wrote: > Hi, > > For one of our applications in our testlab, we are running bridge(4) > with several user land applications. I have found that the bridging > performance (64 byte packets, 2-port bridge) on 5.2.1 is > significantly lower than that of RELENG_4, especially when running in > SMP. The platform is a dual 2.8GHz xeon with a dual port em (100MHz > PCI-X). Invariants are disabled, and polling (with idle_polling > enabled) is used. Well, I do have a patch you can try but I am unsure if it will help or not. Basically, the new interrupt code in 5.2 has a bug that can really hurt interrupt latency when doing kernel-intensive tasks. Since you are using polling it may not help much, but it's still something you can try. The patch should apply to 5.2.1 just fine. http://www.FreeBSD.org/~jhb/patches/ithread_preempt.patch -- John Baldwin <>< http://www.FreeBSD.org/~jhb/ "Power Users Use the Power to Serve" = http://www.FreeBSD.org