From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Sat, 8 Sep 2001 02:39:57 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Sat, 8 Sep 2001 02:39:38 -0400 Received: from smtp6.mindspring.com ([207.69.200.110]:58384 "EHLO smtp6.mindspring.com") by vger.kernel.org with ESMTP id convert rfc822-to-8bit; Sat, 8 Sep 2001 02:39:22 -0400 Subject: Re: Linux Preemptive patch success 2.4.10-pre4 + lots of other From: Robert Love To: Dieter =?ISO-8859-1?Q?N=FCtzel?= Cc: Linux Kernel List In-Reply-To: <200109080629.f886TrG13025@zero.tech9.net> In-Reply-To: <200109080629.f886TrG13025@zero.tech9.net> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Mailer: Evolution/0.13.99+cvs.2001.09.05.07.08 (Preview Release) Date: 08 Sep 2001 02:39:51 -0400 Message-Id: <999931193.838.25.camel@phantasy> Mime-Version: 1.0 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Sat, 2001-09-08 at 02:29, Dieter Nützel wrote: > I've tested your former patch against 2.4.7-acX and sent you some feedback > but you didn't answered my post. > > Is the module problem (missing preempt_xxx symbols) thing fixed? I am not sure what module problem you are talking about. There was a bug in my original patch that caused a link error in most situations where CONFIG_MODULE was set, but this had to do with atomic_dec_and_lock, not preempt_xxx -- what preempt_xxx symbols? Anyhow, I suggest you grab 2.4.9-ac10 and http://tech9.net/rml/linux/patch-rml-2.4.9-ac10-preempt-kernel-1 and enable CONFIG_PREEMPT and give that a whirl. You should have no problems on a UP system. > Aside from that I saw some nice speed increase (UP Athlon) and very snappy > system. Very glad. Especially to hear there are no problems on Athlon. > dbench-1.1 32 clients load drops to mostly below 16 (33 before) Great. Data points such as this are welcome. What about the actual throughput? Both I and Nigel (see http://kpreempt.sourceforge.net) benched the patch with `dbench -16' and we both got decent throughput increase. > Go on with your great work! Thanks. -- Robert M. Love rml at ufl.edu rml at tech9.net