From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933504Ab1JDTqm (ORCPT ); Tue, 4 Oct 2011 15:46:42 -0400 Received: from mx.binnacle.cx ([74.95.187.105]:39786 "EHLO mx.binnacle.cx" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933210Ab1JDTql (ORCPT ); Tue, 4 Oct 2011 15:46:41 -0400 Message-Id: <6.2.5.6.2.20111004154056.03993660@binnacle.cx> X-Mailer: QUALCOMM Windows Eudora Version 6.2.5.6 Date: Tue, 04 Oct 2011 15:45:21 -0400 To: Joe Perches , Christoph Lameter From: starlight@binnacle.cx Subject: Re: big picture UDP/IP performance question re 2.6.18 -> 2.6.32 Cc: Eric Dumazet , linux-kernel@vger.kernel.org, netdev , Willy Tarreau , Peter Zijlstra , Stephen Hemminger In-Reply-To: <1317757118.3580.24.camel@Joe-Laptop> References: <6.2.5.6.2.20111003112108.03a83a28@binnacle.cx> <1317757118.3580.24.camel@Joe-Laptop> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org At 12:38 PM 10/4/2011 -0700, Joe Perches wrote: >On Tue, 2011-10-04 at 14:16 -0500, Christoph Lameter wrote: >> On Mon, 3 Oct 2011, starlight@binnacle.cx wrote: >> > I've come to the conclusion that Eric is right >> > and the primary issue is an increase in the >> > cost of scheduler context switches. Have >> > been watching this number and it has held >> > pretty close to 200k/sec under all scenarios >> > and kernel versions, so it has to be >> > a longer code-path, bigger cache pressure >> > or both in the scheduler. Sadly this makes >> > newer kernels a no-go for us. >> We had similar experiences. Basically latency >> constantly gets screwed up by the new fancy >> features being added to the scheduler and network >> subsystem (most notorious is the new "fair" >> scheduler, 2.6.23 made a big >> step down). > >Idly curious, have you compared bfs performance? >http://ck.kolivas.org/patches/bfs/bfs-faq.txt No but it certainly does look very interesting. Looks like trying it out not much more work than patching and building a kernel and running the benchmark. Will take a look and report back if I do. Little busy at present but should have time in a week or so.