From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Thu, 18 Jan 2001 20:48:24 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Thu, 18 Jan 2001 20:48:15 -0500 Received: from ns.virtualhost.dk ([195.184.98.160]:7695 "EHLO virtualhost.dk") by vger.kernel.org with ESMTP id ; Thu, 18 Jan 2001 20:47:59 -0500 Date: Fri, 19 Jan 2001 02:47:45 +0100 From: Jens Axboe To: Andi Kleen Cc: Andrea Arcangeli , Marcelo Tosatti , linux-kernel@vger.kernel.org Subject: Re: 2.4.1pre8 slowdown on dbench tests Message-ID: <20010119024745.G18209@suse.de> In-Reply-To: <20010119011629.C32087@athlon.random> <20010119024023.B18209@suse.de> <20010119024610.A7573@gruyere.muc.suse.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20010119024610.A7573@gruyere.muc.suse.de>; from ak@suse.de on Fri, Jan 19, 2001 at 02:46:10AM +0100 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Fri, Jan 19 2001, Andi Kleen wrote: > > Yes I agree, that values should probably be tweaked a bit. I'll > > try and squeeze some testing in to generate the best possible > > values. > > Please also add a sysctl. I always feel uncomfortable with such hardcoded > heuristics. There tends to be always another workload where the heuristic > fails and manual tuning is useful. Sure, we can do that. But it should only really make a difference for low memory machines, otherwise the numbers wouldn't change so much. So the limits are not really that important, and only need to be in the ball park. -- * Jens Axboe * SuSE Labs - 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/