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, 11 Jul 2002 17:41:27 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Thu, 11 Jul 2002 17:41:26 -0400 Received: from gateway-1237.mvista.com ([12.44.186.158]:20985 "EHLO av.mvista.com") by vger.kernel.org with ESMTP id ; Thu, 11 Jul 2002 17:41:25 -0400 Message-ID: <3D2DFC08.445B2723@mvista.com> Date: Thu, 11 Jul 2002 14:43:36 -0700 From: george anzinger Organization: Monta Vista Software X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.2.12-20b i686) X-Accept-Language: en MIME-Version: 1.0 To: Alan Cox CC: mbs , dank@kegel.com, "linux-kernel@vger.kernel.org" Subject: Re: Periodic clock tick considered harmful (was: Re: HZ, preferably as References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Alan Cox wrote: > > > First blush is HELL YES! The issue is accounting. When you > > ask how long a program ran, you are looking at the > > accounting that happens on a tick. This is where one of two > > Thats also an implementation issue. Note that the current code is also > wildly inaccurate. Mr Shannon says we are good to at best 50 run/sleep > changes a second. I've got "100% busy" workloads that are 99% asleep. > > Tracking cpu usage at task switch works a lot better for newer processors > which as well as having rdtsc also have performance counters. In fact you > can do much more interesting things on modern PC class platforms like > scheduling using pre-emption interrupts based on instructions executed, > memory accesses and more. > Oh, I agree. Hardware could make all this a lot easier. -- George Anzinger george@mvista.com High-res-timers: http://sourceforge.net/projects/high-res-timers/ Real time sched: http://sourceforge.net/projects/rtsched/ Preemption patch: http://www.kernel.org/pub/linux/kernel/people/rml