Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262182AbUCSDTF (ORCPT ); Thu, 18 Mar 2004 22:19:05 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262194AbUCSDTF (ORCPT ); Thu, 18 Mar 2004 22:19:05 -0500 Received: from fmr10.intel.com ([192.55.52.30]:30858 "EHLO fmsfmr003.fm.intel.com") by vger.kernel.org with ESMTP id S262182AbUCSDS7 (ORCPT ); Thu, 18 Mar 2004 22:18:59 -0500 Subject: how to disable HT From: Len Brown To: Kernel Mailing List , ACPI Developers Content-Type: text/plain Organization: Message-Id: <1079666329.3274.55.camel@dhcppc4> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.3 Date: 18 Mar 2004 22:18:49 -0500 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1421 Lines: 39 BIOS SETUP is the best method to disable HT. But sometimes that isn't practical, so disabling at Linux boot-time is needed. Most OEMs' BIOS enumerates primary logical processors before secondary logical processors -- so reducing NR_CPUS at build-time can start a processor per package before hitting the limit -- effectively disabling HT. 2.6.5-rc1-mm2 includes a patch that allows boot-time maxcpus=N to work the same as NR_CPUS=N: http://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.5-rc1/2.6.5-rc1-mm2/broken-out/bk-acpi.patch It moves maxcpus=N earlier to cpu enumeration time from CPU startup time. So if disabling HT at boot-time is important to you, please try it out and comment on the success/failure in the bug report: http://bugzilla.kernel.org/show_bug.cgi?id=2317 I expect that it will work on most systems, but not all, and so it is important to look in dmesg to verify that it disabled the secondary logical processors at intended. eg. system with 4 logical processors booted with maxcpus=2: Total of 2 processors activated (11091.96 BogoMIPS). WARNING: No sibling found for CPU 0. WARNING: No sibling found for CPU 1. thanks, -Len - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/