Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261294AbVDQK3V (ORCPT ); Sun, 17 Apr 2005 06:29:21 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261298AbVDQK3U (ORCPT ); Sun, 17 Apr 2005 06:29:20 -0400 Received: from [194.90.79.130] ([194.90.79.130]:41490 "EHLO argo2k.argo.co.il") by vger.kernel.org with ESMTP id S261294AbVDQK3Q (ORCPT ); Sun, 17 Apr 2005 06:29:16 -0400 Subject: Re: More performance for the TCP stack by using additional hardware chip on NIC From: Avi Kivity To: Arjan van de Ven Cc: Andreas Hartmann , linux-kernel In-Reply-To: <1113728880.17394.16.camel@laptopd505.fenrus.org> References: <1113728880.17394.16.camel@laptopd505.fenrus.org> Content-Type: text/plain Message-Id: <1113733753.15803.26.camel@avik.scalemp> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Sun, 17 Apr 2005 13:29:14 +0300 Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 17 Apr 2005 10:29:15.0026 (UTC) FILETIME=[4DABC720:01C54338] Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1508 Lines: 38 On Sun, 2005-04-17 at 12:07, Arjan van de Ven wrote: > On Sun, 2005-04-17 at 10:17 +0200, Andreas Hartmann wrote: > > Hello! > > > > Alacritech developed a new chip for NIC's > > (http://www.alacritech.com/html/tech_review.html), which makes it possible > > to take away the TCP stack from the host CPU. Therefore, the host CPU has > > more performance for the applications according Alacritech. > > there are very many good reasons why this for linux is not the right > solution, including the fact that the linux tcp/ip stack already is > quite fast so the "gains" achieved aren't that stellar as the gains you > get when comparing to windows. > TOEs can remove the data copy on receive. In some applications (notably storage), where the application does not touch most of the data, this is a significant advantage that cannot be achieved in a software-only solution. > Also these types of solution always add quite a bit of overhead to > connection setup/teardown making it actually a *loss* for the "many > short connections" types of workloads. Now guess which things certain > benchmarks use, and guess what real world servers do :) > again, this depends on the application. a copyless solution is probably necessary to achieve 10Gb/s speeds. Avi - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/