[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
On Wed, 31 Oct 2007 11:33:57 +0000 Simon Williams <simon@xxxxxxxxxxxxxxxxxxx> wrote: > Neil Williams wrote: > > Expect gradual improvement, sometimes held back by applications with a > > slower update cycle. > > This sounds very interesting. How much life do you think you could get > out of it, if all the software was written to be power-aware? Expect a 250-400% increase in battery life. > Taking a > rough figure of 2 hours in the current situation you would need to at > least triple the life. Surely you aren't going to manage to get that > much out of it? If not more. > I guess this also begs the question- are we wasting that > much power on our desktop systems? Yep. Why do you think server rooms get so darned hot? Why doesn't the temperature change that much when the servers are all under heavy load? All this activity on the embedded side is matched by activity in the kernel and core libraries. There is a lot going on in this area and improvements are being made. Power == heat. Power costs money to generate and heat costs MORE money to remove. That's an added incentive to get power management working on the server so there are people coming at this problem from all sides. OSX and Windows can achieve these levels of power saving using their proprietary drivers but only if you don't use non-vendor applications. GNU/Linux can implement these changes across all packages (via systems like Debian Policy / distro QA) so there's no reason why we shouldn't get better results, in time. By feeding the patches upstream, all distributions will eventually benefit. My own laptop easily did 6hrs running OSX and I have every expectation that GNU/Linux will match that. Currently, it's less than 2. It's not that GNU/Linux cannot do power management properly - until quite recently it simply hasn't been necessary. Now it's gone from "a good idea" to "we'd better get this sorted out sharpish!". It's big problem so it'll take a little time but it is a SMoP. [1] [1] Simple Matter of Programming. -- Neil Williams ============= http://www.data-freedom.org/ http://www.nosoftwarepatents.com/ http://www.linux.codehelp.co.uk/
Attachment:
pgpciGfu3kS6N.pgp
Description: PGP signature
-- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/linux_adm/list-faq.html