From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1030372AbVH0MiJ (ORCPT ); Sat, 27 Aug 2005 08:38:09 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1030373AbVH0MiJ (ORCPT ); Sat, 27 Aug 2005 08:38:09 -0400 Received: from amsfep12-int.chello.nl ([213.46.243.17]:7254 "EHLO amsfep20-int.chello.nl") by vger.kernel.org with ESMTP id S1030372AbVH0MiI (ORCPT ); Sat, 27 Aug 2005 08:38:08 -0400 Subject: Re: Linux 2.6 context switching and posix threads performance question From: Peter Zijlstra To: Mateusz Berezecki Cc: LKML In-Reply-To: References: <20050827121158.GA18406@oepkgtn.mshome.net> <1125145582.20161.62.camel@twins> Content-Type: text/plain Date: Sat, 27 Aug 2005 14:38:02 +0200 Message-Id: <1125146282.20161.69.camel@twins> Mime-Version: 1.0 X-Mailer: Evolution 2.2.3 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Group replies work best with mailing lists. On Sat, 2005-08-27 at 14:30 +0200, Mateusz Berezecki wrote: > On 8/27/05, Peter Zijlstra wrote: > > > Well the obvious question is: what kernel version and which thread > > library? > > Ah, sorry I just forgot to put this info > > Linux blade1 2.6.12 #1 SMP Tue Jul 26 08:43:57 GMT 2005 i686 Intel(R) > Xeon(TM) CPU 3.06GHz GenuineIntel GNU/Linux > > the cpu is with HT so Linux thinks there are 4 cpus on board > > > > > 2.4 with LinuxThreads might have severe problems. However 2.6 with NPTL > > should be able to handle it, IIRC Igno once did a million threads with _Ingo_ Molnar that is; I really should have gotten those 4 letters in the right order by now, humble appologies. > > that combination just to show that it worked ;-). > > Yes, but how about switching contexts and the performance impact? > Will it slowdown the whole system or just result in low or medium > overhead?. > These threads are meant to be running an intensive network connection > and possibly analysing the dataflow and maybe filtering some stuff. > Each thread doing the same task. I think it should work; however I'm not the most qualified to answer this. -- Peter Zijlstra