From owner-freebsd-questions@FreeBSD.ORG Fri Sep 23 09:35:25 2011 Return-Path: Delivered-To: freebsd-questions@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id BAE831065672 for ; Fri, 23 Sep 2011 09:35:25 +0000 (UTC) (envelope-from bonomi@mail.r-bonomi.com) Received: from mail.r-bonomi.com (mx-out.r-bonomi.com [204.87.227.120]) by mx1.freebsd.org (Postfix) with ESMTP id 898A08FC15 for ; Fri, 23 Sep 2011 09:35:25 +0000 (UTC) Received: (from bonomi@localhost) by mail.r-bonomi.com (8.14.4/rdb1) id p8N9YxRx088414 for freebsd-questions@freebsd.org; Fri, 23 Sep 2011 04:34:59 -0500 (CDT) Date: Fri, 23 Sep 2011 04:34:59 -0500 (CDT) From: Robert Bonomi Message-Id: <201109230934.p8N9YxRx088414@mail.r-bonomi.com> To: freebsd-questions@freebsd.org In-Reply-To: <4E7C3F83.2050805@my.gd> Subject: Re: load average with multi-core CPU's X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 23 Sep 2011 09:35:25 -0000 > From owner-freebsd-questions@freebsd.org Fri Sep 23 03:15:37 2011 > Date: Fri, 23 Sep 2011 10:12:51 +0200 > From: Damien Fleuriot > To: freebsd-questions@freebsd.org > Subject: Re: load average with multi-core CPU's > > On 9/22/11 10:59 PM, Rodrigo Gonzalez wrote: > > On 09/22/2011 04:29 PM, Mark Felder wrote: > >> On Thu, 22 Sep 2011 10:22:43 -0500, Henry M wrote: > >> > >>> Hi all, > >>> > >>> Can someone explain, or point me to correct documentation on what the > >>> load > >>> average on top/uptime is actually displaying? > >> > >> Load average is "average number of processes in the run queue" for the > >> 1, 5, and 15 minute intervals. If you have a quad core CPU a 4.00 load > >> average means you've been keeping the CPU busy at 100%. > > Not exactly as I understand it....IO (disk, network or whatever) affects > > it too... > > It is the number of task waiting in queue to be run....but IO is > > important...if 2 processes are waiting for IO and it is completely > > saturated they will be kept in queue so load will get higher > > I think there are other things that affect load average but are over my > > current knowledge... > > > > Regards > > > > Rodrigo Gonzalez > > Actually, I could be wrong but that is the number of tasks both in the > waiting *AND* the running queue. It is the average of the number of 'runnable' processes -- those that are actually running (which is -- obviously! -- limited to the number of logical cpu's present) and those that are -- in _all_ other respects -- 'ready' to be run. This list of processes -- 'running' and 'runnable -- is known as the 'run queue'. The cpu 'scheduler' allocated cpu time slots between the processes in the 'run queueu', _only_. Anything -not- in the 'run queue' is not eligible for a slice of cpu time -- because it "can't" use cpu time, if it were to be offered, because it is 'waiting' on something else.