Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Thu, 2 Nov 2000 13:35:04 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Thu, 2 Nov 2000 13:34:54 -0500 Received: from delta.ds2.pg.gda.pl ([153.19.144.1]:36798 "EHLO delta.ds2.pg.gda.pl") by vger.kernel.org with ESMTP id ; Thu, 2 Nov 2000 13:34:38 -0500 Date: Thu, 2 Nov 2000 19:23:41 +0100 (MET) From: "Maciej W. Rozycki" Reply-To: "Maciej W. Rozycki" To: Ulrich Drepper cc: root@chaos.analogic.com, kernel@kvack.org, "Dr. David Gilbert" , linux-kernel@vger.kernel.org Subject: Re: Dual XEON - >>SLOW<< on SMP In-Reply-To: Message-ID: Organization: Technical University of Gdansk MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On 2 Nov 2000, Ulrich Drepper wrote: > I'm seeing this as well, but only with PIII Xeon systems, not PII > Xeon. Every single timer interrupt on any CPU is accompanied by a NMI > and LOC increment on every CPU. > > CPU0 CPU1 > 0: 146727 153389 IO-APIC-edge timer This is the legacy 8254 timer source, used for the system time, i.e. gettimeofday() and friends. > NMI: 300035 300035 This is the NMI watchdog at work. Every tick of the legacy timer all CPUs receive an NMI unless overridden by the "nmi_watchdog" command line argument. > LOC: 300028 300028 This is the internal local APIC timer used for scheduling. Every CPU is equipped with such a private timer. -- + Maciej W. Rozycki, Technical University of Gdansk, Poland + +--------------------------------------------------------------+ + e-mail: macro@ds2.pg.gda.pl, PGP key available + - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/