Even though the statement that you can optimize your linux to use less power is true, the truth is that most users won't do that (lack of knowledge or time). So it must be able to optimize itself. We shouldn't try to use the argument that windows also eats alot without the user installing specialized drivers because users are already trained to do that while a user switching to linux most likely doesn't know how to install drivers. Secondly, linux comes with tons of drivers so it should be able to optimize itself. And thirdly, after all wasn't linux supposed to be better? Since when are we emulating windows behavior?
Try using tuned. It's a power management daemon that has a number of presets (like power savings mode, performance mode, and etc), the ability to specify custom configurations as well as the pretty new powertop2tuned utility which you use by specifying a configuration name of your choosing, do whatever you do on the pc for a few minutes, and then it outputs the configuration to the tuned directory. Then you can tell tuned to run at boot with that special configuration through the tuned utility. I think the old kde powermizer used it as its back end, the but I'm not positive.
If you want to run ANY operating system on hardware that isn't supported by the manufacturer, run it in a virtual machine instance
Otherwise you're just asking for trouble
If you want to run Linux on bare hardware, buy supported hardware!
This is true of ANY operating system, not just Linux
I run Linux on Supermicro hardware. It says "Linux" right on the box and in the manual. They have specific Linux settings in the BIOS. They test their hardware with Linux. They list the Linux versions that they tested with, on their web site. I never have any trouble running Linux on Supermicro hardware.
VERY good point indeed.
I will try to very specifically buy Linux supported hardware when I next shop for a computer or components. Overall I've had a lot of success with the hardware I've always just had, which is a credit to Linux, but it would make sense to target the proper supported hardware for the best experience. Screw Crapple and M$.
Oh really? How about a "designed for linux" machine instead?
How about testing power consumption in a hardware tested for linux instead? How about: Running in a ZaReason laptop, Ubuntu vs MacOSX? See the problem?
Its OBVIOUS Apple would support and fine tune the hardware they have chosen. This is valid even for windows, where there have been at least some more testing by the manufacturer.
The open community has to work up hill, even reverse engineer stuff to get things barely working, and its a great triumph when they finally master the device or when the manufacturer released the specs and developers spent time fine tuning their software with it.
Or do you think there is any magic in NDA ridden Android world? No, it's simply better manufacturer support.
It's true that Apple controls their hardware but they can't forsee precisely what hardware they'll use in the future so massive optimization for specific hardware isn't something they do. They want OS X to be optimized for whatever hardware they use in the future for any version of OS X. The huge discrepancy in numbers is NOT accounted for (entirely or even mostly) by Apple fine tuning OS X for optimizations specific to today or yesterday's hardware.
I would like to see a better linux vs other OSes on power consumption article using fixed hardware that runs a single GPU that has drivers for these various OSes that are pretty much on parity. E.g. compare power consumption on a system running a single dedicated AMD GPU using the latest Catalyst on linux, Windows and OS X. But DON'T use the lousy open source drivers on linux!