From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933697AbXHFW7Z (ORCPT ); Mon, 6 Aug 2007 18:59:25 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1764766AbXHFW7L (ORCPT ); Mon, 6 Aug 2007 18:59:11 -0400 Received: from smtp1.Stanford.EDU ([171.67.22.28]:53007 "EHLO smtp1.stanford.edu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1764593AbXHFW7K (ORCPT ); Mon, 6 Aug 2007 18:59:10 -0400 X-Greylist: delayed 1725 seconds by postgrey-1.27 at vger.kernel.org; Mon, 06 Aug 2007 18:59:10 EDT Date: Mon, 6 Aug 2007 17:26:33 -0500 From: Gautam Iyer To: Pavel Machek Cc: linux-kernel@vger.kernel.org Subject: Re: Increased CPU usage after upgrading to 2.6.21 Message-ID: <20070806222632.GI24652@math.stanford.edu> References: <20070718164347.GE9506@math.stanford.edu> <20070806183207.GB5761@ucw.cz> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20070806183207.GB5761@ucw.cz> User-Agent: Mutt/1.5.15 (2007-04-06) X-Spam-Score: -101.1 X-Scan-Signature: d3e02c565cd5ac634b2ff74b6ba3e13d Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Aug 06, 2007 at 06:32:08PM +0000, Pavel Machek wrote: > > PS: I'm using the kernel with Gentoo patchset. But I confirmed the above > > persists with the vanilla kernels. > > > > PPS: I also played around with the timer frequency, and premption model. > > Would those make a difference to CPU usage? > > Yes. Or rather, it will make accounting more accurate, so that you see > that 1% cpu usage, unlike previous kernels where something still ate > cpu but kernel did not know it. Hi, Thanks for your response. Well, I now recompiled all the kernels I used with all possible permutations of these options (yes, that did take the better part of last week). I can confirm that under identical option for the preemption model, and timer frequency, the following happen: 1. With 2.6.19 (gentoo-r5), the system claims to use 0% CPU when idle. 2. With 2.6.21 (gentoo-r4), 2.6.22 (gentoo-r1) and the unmodified 2.6.22.1 kernel, the system claims to use between 1% and 3% CPU when idle. When I run "top", I see that all non-idle processes running claim to take 0% CPU (except occasionally the terminal I'm running top in, and top itself). The CPU line in the top output consistently claims something like CPU(s): .3%us, 1.0%sy, ... I'm not sure if this is something I should be worried about or not, Thanks again, Gautam -- 'Budget' -- A method for going broke methodically.