From owner-freebsd-ia64@FreeBSD.ORG Tue Jul 20 19:58:41 2004 Return-Path: Delivered-To: freebsd-ia64@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id C07CA16A4CE for ; Tue, 20 Jul 2004 19:58:41 +0000 (GMT) Received: from ns1.xcllnt.net (209-128-86-226.bayarea.net [209.128.86.226]) by mx1.FreeBSD.org (Postfix) with ESMTP id 93AA143D1F for ; Tue, 20 Jul 2004 19:58:41 +0000 (GMT) (envelope-from marcel@xcllnt.net) Received: from ns1.xcllnt.net (localhost [127.0.0.1]) by ns1.xcllnt.net (8.12.11/8.12.11) with ESMTP id i6KJwf7d028134; Tue, 20 Jul 2004 12:58:41 -0700 (PDT) (envelope-from marcel@ns1.xcllnt.net) Received: (from marcel@localhost) by ns1.xcllnt.net (8.12.11/8.12.11/Submit) id i6KJwf7j028133; Tue, 20 Jul 2004 12:58:41 -0700 (PDT) (envelope-from marcel) Date: Tue, 20 Jul 2004 12:58:41 -0700 From: Marcel Moolenaar To: Garry Forsgren Message-ID: <20040720195841.GB28030@ns1.xcllnt.net> References: <20040720183329.90874.qmail@web11306.mail.yahoo.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20040720183329.90874.qmail@web11306.mail.yahoo.com> User-Agent: Mutt/1.5.5.1i cc: freebsd-ia64@freebsd.org Subject: Re: Excessive SWI8: clock activity on Intel SDV (Bandeera 4X SMP) X-BeenThere: freebsd-ia64@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Porting FreeBSD to the IA-64 List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 20 Jul 2004 19:58:41 -0000 On Tue, Jul 20, 2004 at 11:33:29AM -0700, Garry Forsgren wrote: > After booting a SMP kernel from FreeBSD > 5.2-20040528-SNAP #1, on an Intel SDV (Bandeera 4X) > the system seems to be running very slow. With the > system idle, vmstat reports about 20% system cpu and > 10% interrupt cpu usage. Also systat indicates SWI8: > clock activity in addition to the normal idle > activity. > > The same SMP kernel appears to run fine on a different > Intel SDV (Lyon 4X). > > Are there any special setup steps required using the > Intel Bandeera SDV? In principle no, but I have no idea what kind of machine that is, so we may not have the necessary support. Can you send the output of the following commands (I interleaved sample output for reference): % sysctl debug | grep clock debug.clock_adjust_edges: 550 debug.clock_adjust_excess: 0 debug.clock_adjust_lost: 0 debug.clock_adjust_ticks: 1113 % sysctl debug | grep to_avg debug.to_avg_depth: 2500 debug.to_avg_gcalls: 1003 debug.to_avg_mpcalls: 1255 % vmstat -i interrupt total rate clock 633980 99 atapci0 47 0 mpt0 9154 1 mpt1 16 0 bge0 229712 36 uart3 362 0 puc0 4 0 Total 873275 137 NOTE: The above output has been made on an UP kernel, so numbers may differ. Also, try building a kernel with DIAGNOSTICS. This may provide some clues as well. HTH, -- Marcel Moolenaar USPA: A-39004 marcel@xcllnt.net