Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Tue, 14 Nov 2000 10:04:32 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Tue, 14 Nov 2000 10:04:12 -0500 Received: from mx3.port.ru ([194.67.23.37]:48911 "EHLO mx3.port.ru") by vger.kernel.org with ESMTP id ; Tue, 14 Nov 2000 10:04:08 -0500 From: "Samium Gromoff" <_deepfire@mail.ru> To: linux-kernel@vger.kernel.org Subject: RE: /proc tweaking Mime-Version: 1.0 X-Mailer: mPOP Web-Mail 2.19 X-Originating-IP: [195.34.30.66] Reply-To: "Samium Gromoff" <_deepfire@mail.ru> Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7BIT Message-Id: Date: Tue, 14 Nov 2000 17:33:23 +0300 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org on Tue, 14 Nov 2000 Arjan van de Ven wrote: >In article you wrote: >> Hey people, i`ve got such a thought >> not long ago: >> all boxes are different, but the /proc/sys/vm >> defaults are equal for every people, so there >> is a good issue in getting more performance >> from linux, just by making a way to autoadjust >> these mysterious values according to amount of >> RAM/swap and speed of CPU! Or this can be >done >> in userspace with an utility which look also >> on the field of box` use (eg workstation, >>server etc...) >> But who can make this better than the >>people who hack the kernel? >> And i wonder why such a issue is not >>clearly covered? (maybe >>i`m making mistake?) >> This can also be done for >>proc/sys/net/*... > > >Take a look at powertweak. >http://powertweak.sourceforge.net >Made by kernel people, for non-kernel people. Maybe i were not enough exact, but i`ve meant addition of some intelligence to tweaking /proc e.g. something what automates tuning, not only providing interface to such actions. But after lookthru ptweaks source i realized what its ONLY interface (to proc), and MAYBE it does some PCI tuning (really intelligent choices to advance system`s performance). BTW powertweak is a port from unfamous MD, therefore when it was created NO proc tuning was in mind... There is another argument, telling what doing autotune is ugly by design.... but thats another issue. Sorry for poor english/lameness - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/