From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S263187AbVG3XaB (ORCPT ); Sat, 30 Jul 2005 19:30:01 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S263190AbVG3XaB (ORCPT ); Sat, 30 Jul 2005 19:30:01 -0400 Received: from liaag2ae.mx.compuserve.com ([149.174.40.156]:3819 "EHLO liaag2ae.mx.compuserve.com") by vger.kernel.org with ESMTP id S263186AbVG3X37 (ORCPT ); Sat, 30 Jul 2005 19:29:59 -0400 Date: Sat, 30 Jul 2005 19:26:40 -0400 From: Chuck Ebbert <76306.1226@compuserve.com> Subject: Re: [sched, patch] better wake-balancing, #2 To: Ingo Molnar Cc: "Chen, Kenneth W" , Andrew Morton , Nick Piggin , linux-kernel , linux-ia64 Message-ID: <200507301929_MC3-1-A601-D4C2@compuserve.com> MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Fri, 29 Jul 2005 at 17:02:07 +0200, Ingo Molnar wrote: > do wakeup-balancing only if the wakeup-CPU is idle. > > this prevents excessive wakeup-balancing while the system is highly > loaded, but helps spread out the workload on partly idle systems. I tested this with Volanomark on dual-processor PII Xeon -- the results were very bad: Before: 5863 messages per second 124169 schedule 64.1369 64663 _spin_unlock_irqrestore 4041.4375 7949 tcp_clean_rtx_queue 6.5370 6787 net_rx_action 24.9522 After: 5569 messages per second 139417 schedule 72.0129 82169 _spin_unlock_irqrestore 5135.5625 9949 tcp_clean_rtx_queue 8.1817 7917 net_rx_action 29.1066 __ Chuck