From owner-freebsd-current@FreeBSD.ORG Tue Sep 16 13:40:08 2008 Return-Path: Delivered-To: freebsd-current@FreeBSD.ORG Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 1502C1065676 for ; Tue, 16 Sep 2008 13:40:08 +0000 (UTC) (envelope-from cpghost@cordula.ws) Received: from fw.farid-hajji.net (fw.farid-hajji.net [213.146.115.42]) by mx1.freebsd.org (Postfix) with ESMTP id 9F8F68FC19 for ; Tue, 16 Sep 2008 13:40:07 +0000 (UTC) (envelope-from cpghost@cordula.ws) Received: from phenom.cordula.ws (phenom [192.168.254.60]) by fw.farid-hajji.net (Postfix) with ESMTP id 63B86352D2; Tue, 16 Sep 2008 15:40:05 +0200 (CEST) Date: Tue, 16 Sep 2008 13:40:11 +0000 From: cpghost To: freebsd-current@FreeBSD.ORG Message-ID: <20080916134011.GA18208@phenom.cordula.ws> References: <20080915185446.GB69615@phenom.cordula.ws> <200809160906.m8G967H4020865@lurza.secnetix.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200809160906.m8G967H4020865@lurza.secnetix.de> User-Agent: Mutt/1.5.18 (2008-05-17) Cc: Subject: Re: Improved multiprocessor usage on amd64 X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 16 Sep 2008 13:40:08 -0000 On Tue, Sep 16, 2008 at 11:06:07AM +0200, Oliver Fromme wrote: > cpghost wrote: > > On Sun, Sep 14, 2008 at 08:38:00PM -0500, Stephen Montgomery-Smith wrote: > > > I have a dual core amd64 on which I run a processor intensive numerical > > > program. I had been frustrated because it seemed to run 3 or 4 times > > > faster under Linux. But with a recent upgrade of FreeBSD-CURRENT, it > > > now goes at about the same speed as Linux. > > > > > > The program takes about an hour. For the first minute, the program runs > > > rather slowly, but then it is as if the operating system finds its way, > > > and suddenly it speeds up. "top -H" suggests that for the first minute > > > that one thread is going really slowly, and is perhaps being starved or > > > something. > > > > > > My question is - why is this happening, and is this something I should > > > expect? Are there certain switches or sysctls I can set to make it go > > > fast from the get go? > > > > It looks like you're running powerd (see in /etc/rc.conf). It can take up > > to a minute for the load average of the machine to exceed a certain > > threshold where powerd would finally bump the cpu(s) to full speed. > > No. powerd(8) does not look at the load average at all, > it looks at the CPU usage. It detects within 0.5 seconds > (the default polling interval) when the CPU usage went up > and starts adjusting the performance. It certainly doesn't > take a minute. Oh, yes, you're right: I stand corrected. powerd looks at the kern.cp_time sysctl and not at the load average. > Best regards > Oliver Thanks, -cpghost. -- Cordula's Web. http://www.cordula.ws/