From owner-freebsd-hackers@FreeBSD.ORG Wed Apr 6 14:01:10 2005 Return-Path: Delivered-To: freebsd-hackers@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 64A4016A4D2 for ; Wed, 6 Apr 2005 14:01:10 +0000 (GMT) Received: from poup.poupinou.org (poup.poupinou.org [195.101.94.96]) by mx1.FreeBSD.org (Postfix) with ESMTP id 061F843D39 for ; Wed, 6 Apr 2005 14:01:10 +0000 (GMT) (envelope-from ducrot@poupinou.org) Received: from ducrot by poup.poupinou.org with local (Exim) id 1DJB5a-0001vd-00; Wed, 06 Apr 2005 16:01:02 +0200 Date: Wed, 6 Apr 2005 16:01:02 +0200 To: Frank Behrens Message-ID: <20050406140102.GY2298@poupinou.org> References: <200504041645.j34Gj2ow002999@pinky.frank-behrens.de> <200504060649.j366nGQg021228@pinky.frank-behrens.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200504060649.j366nGQg021228@pinky.frank-behrens.de> User-Agent: Mutt/1.5.6+20040907i From: Bruno Ducrot cc: freebsd-hackers@freebsd.org Subject: Re: My experience with cpufreq in -STABLE X-BeenThere: freebsd-hackers@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Technical Discussions relating to FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 06 Apr 2005 14:01:10 -0000 On Wed, Apr 06, 2005 at 08:49:15AM +0200, Frank Behrens wrote: > Bruno Ducrot wrote on 4 Apr 2005 19:17: > > You may start looking at src/usr.sbin/powerd in -current, and improve it > > a bit? The actual algorithm used in powerd may need some rework IMHO. > > Which problems do you see? powerd use an exponentional decrease of the frequency. This might be not stable for certain workload. > My comments: > 1. If the frequency is raised it should not go down immediately on > short idle phases. To raise the frequency the current idle value > should used but to lower the frequency we should calculate a weighted > idle average value. It might be a solution. It has been tested in that paper: http://www-mtl.mit.edu/research/icsystems/pubs/conferences/2001/sinha_vlsi2001.pdf and it seems it's maybe not the better one, though (search 'MAW', its exactly what you suggest). > 2. The default polling time of 500 ms seems to be very short. It can > increased to several seconds. Problem if you increase teh polling intervall is that you can't be sure that the system can detect in time when going up. -- Bruno Ducrot -- Which is worse: ignorance or apathy? -- Don't know. Don't care.