From mboxrd@z Thu Jan 1 00:00:00 1970 From: Dave Jones Subject: Re: cpufreq with ondemand/p4-clockmod crashes kernel on Celeron-D 325 Date: Mon, 5 Dec 2005 13:37:05 -0500 Message-ID: <20051205183705.GG12664@redhat.com> References: <12aa0420511181325m9878a72ub51f35e11bbee262@mail.gmail.com> <20051204171827.GB6017@isilmar.linta.de> <12aa0420512041411q1aab21d3o3bbddd38cc8140b4@mail.gmail.com> Mime-Version: 1.0 Return-path: Content-Disposition: inline In-Reply-To: <12aa0420512041411q1aab21d3o3bbddd38cc8140b4@mail.gmail.com> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: cpufreq-bounces@lists.linux.org.uk Errors-To: cpufreq-bounces+glkc-cpufreq=gmane.org+glkc-cpufreq=gmane.org@lists.linux.org.uk Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Rick Nas Cc: cpufreq@lists.linux.org.uk On Sun, Dec 04, 2005 at 11:11:10PM +0100, Rick Nas wrote: > Yeah, I gathered that from recent discussions on this list. But it seems > strange to me, if clock modulation has a thermal effect, shouldn't it have a > power effect as well? I mean, the heat is generated by the current right, > less heat => less current?? > > Anyway, I also read that the issue has been explained on this list a couple > of times already, so I guess you're right. With clock modulation, you're still running the CPU at the same amount of current, you're just not doing useful work every clock cycle. If you're only doing something once every four clocks, it'll take 4 times as long to finish that workload. So you don't save any battery because you're increasing the runtime. You get thermal wins because of the idle slots you're introducing, where the CPU isn't doing anything. Pretty pictures should explain it if the above still seems odd.. no modulation: ____ _| |_________________ with modulation: _ _ _ _ _| |___| |___| |___| |___ The 'level' part of the diagrams indicate the CPU idling. What an edge occurs, it's doing useful work, and generating heat. Dave