From owner-freebsd-hackers@FreeBSD.ORG Thu Nov 13 02:06:32 2003 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 425ED16A4D1 for ; Thu, 13 Nov 2003 02:06:32 -0800 (PST) Received: from purple.the-7.net (purple.the-7.net [207.158.28.23]) by mx1.FreeBSD.org (Postfix) with ESMTP id 1662F43FF3 for ; Thu, 13 Nov 2003 02:06:31 -0800 (PST) (envelope-from ab@astralblue.net) Received: from astralblue.net (adsl-68-123-46-151.dsl.pltn13.pacbell.net [68.123.46.151]) by purple.the-7.net (8.12.9p2/8.12.9) with ESMTP id hADA7gKF053384 for ; Thu, 13 Nov 2003 02:07:49 -0800 (PST) (envelope-from ab@astralblue.net) Message-ID: <3FB357A3.4030907@astralblue.net> Date: Thu, 13 Nov 2003 02:06:27 -0800 From: "Eugene M. Kim" User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.6a) Gecko/20031103 X-Accept-Language: en-us, en, ko MIME-Version: 1.0 To: freebsd-hackers@freebsd.org References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Spam-Status: No, hits=-4.9 required=5.0 tests=BAYES_00 autolearn=ham version=2.60 X-Spam-Checker-Version: SpamAssassin 2.60 (1.212-2003-09-23-exp) on purple.the-7.net Subject: cpu_idle_hlt (Re: Confused about HyperThreading and Performance) 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: Thu, 13 Nov 2003 10:06:32 -0000 John Baldwin wrote: > Also, as someone else mentioned, setting 'machdep.cpu_idle_hlt=1' can > be useful on some HTT systems. However, p4's have a problem with their > interrupt routing that can leave the second CPU halted for a long time > if you do that. I have a few quick questions... Searched on Google but couldn't get satisfactory answers: 1. Without cpu_idle_hlt, is the problem that the idle spin loop one logical CPU executes would eat up CPU time and prevent the other logical CPU from running? 2. If so, would it explain the unusually high percentage of system time and unusually low percentage of user time (shown on systat -vm 1) when processes should be mostly doing CPU work and some heavy disk I/O at the same time? 3. Is cpu_idle_hlt also potentially unsafe on P4 Xeon-based SMP systems? Thanks, Eugene P.S. It'd be greatly appreciated if someone could point to an in-depth discussion about Hyperthreading and cpu_idle_hlt... *yells at his poor Googling skill XD*