Is there any statistics or/and test that proves which os that's most power effecient?

I've found older tests, but not any new.

Kind regards /Peter

In theory, if both are idle, and set the hardware to same power states (any CPU and GPU power saving states for example) they should be equal.

Differences come from things like:

- either OS fails to set a component in your hardware to some low power state when not in use

- either OS has some heavy cpu or heavy IO background processes

The first one improves all the time with better drivers, the second one you can configure on your own system at will.

I don't think there is much difference there. But for linux you don't need so powerful hardware as for windows to have a snappy and responsive computer, and you need to upgrade it less often.

There aren't any reliable tests, as it is all hardware dependent. As lm said, if something isn't put in a low power state it throws everything off.

My own tests on my desktop resulted in these power consumptions at idle:
Vista > XP Pro > Ubuntu

The difference was only a couple of watts though.

Overall it is a moot point. Five watts, eight hours a day, for a year is ~52.6 MJ, which is a tiny fraction of the energy that went into producing the computer (which is also very debatable and not well understood). If you are interested in the environmentally friendly way of doing things, linux wins hands down for its ability to outlast hardware. This would be as opposed to the other way around for Windows.

few watts make a difference. The point is to save as much energy as possible, even if only some watts.

So, Ubuntu is the most frugal OS...

Depends on your hardware. My laptop runs much hotter (and therefore sucks more power) on Ubuntu than it does on XP...

My notion is that it depends on which Linux distribution you choose out of the hundreds available. It seems logical that those which have less than, equal to or more bells and whistles than Windows will use, respectively, less than/equal to/more power... especially when you have it running on full bore with multiple power consuming programs.

If you're interested in lighter, minimalistic, probably lower power consuming versions of Linux, I would recommend checking out:
DamnSmallLinux (50mb) http://en.wikipedia.org/wiki/Damnsmalllinux
Feather Linux (RAM-based) http://en.wikipedia.org/wiki/Feather_Linux
Puppy Linux (RAM-based) http://en.wikipedia.org/wiki/Puppy_Linux
CrunchBang Linux http://en.wikipedia.org/wiki/CrunchBang_Linux
I also figure that if you can boot off the memory and not have to spin up your HDD or start your flash drive for simple OS start-up and usage, then that should save some watts...

or, if extreme lightness is not your cup of tea, any Linux distro using the lightweight Xfce environment should do, such as
Xubuntu (Ubuntu-like) http://en.wikipedia.org/wiki/Xubuntu
Fluxbuntu also looks interesting: http://en.wikipedia.org/wiki/Fluxbuntu

further reading:
Xfce info- http://en.wikipedia.org/wiki/Xfce
resource objectives for different Linux desktop environments- http://en.wikipedia.org/wiki/Comparison ... nformation

Granted, I have not tested the power consumption of any of these systems; it just seems logical: smaller + less fanciness = less resource (RAM, SSD or HDD) usage = less power consumption

If this notion is incorrect, why support monopolies in any case?

Sounds reasonable - the less bells and whistles, the less power used. The other thing to look at is whether/how much the OS will allow you to control the undervolting/underclocking of your hardware, integrate with speedstep etc.

I think that's one are where my laptop lets me down - Ubuntu doesn't seem to be able to throttle back my HW when it's idle.

It also doesn't cope well with sleep or hibernate modes on my hardware - it goes OK initially, but after a few months it refuses to wake up. I've had the same problem on 4 installs of 2 distros on 2 laptops...

Many untested notions are born out of religious conviction. Please test your conviction with a watt-meter and get back to us.

Being an apatheist I lack the time for/interest in religious convictions. These notions are the result of my reasoning (reason being the classic philosophical-historical enemy of religon).
However, I agree with you that untested notions are risky business which can lead to all sorts of unwanted behavior. So, at the moment let's call it my hypothesis. I'll try to tack down the necessary tools, spend some time in my lab/man-cave and see if I can prove myself wrong or right and get back to the community with the results.

Although, in the case of the tiny RAM-based Linux systems I really can't imagine how HDD-based, gigantic Windows systems can even come close to competing in resource-watt consupmtion reduction. I could be wrong...

for me the my own tests on my desktop resulted on this power consumptions at Vista goes to XP Pro

I don't get it. We can take electricity and convert it to heat. Has anyone been able to get heat back into electricity (without screwing around by boiling water and making steam and running a turbine to run a generator) directly? I mean then we could take these big beasts that put out like over 100W of heat and convert some of that back into power...so we lose less energy...

Is that possible?

This is the closest I can think of, a bit gimmicky and not a major power regenerator:
http://www.reghardware.co.uk/2008/02/29 ... g_cooling/

You could use a peltier or phase change system to draw heat out of the system and dump it into water for household use. Don't think anyone has done this but it sounds very complex.

Linux has far better potential to go minimalistic and cause less load on cpu. But it can lack in other fields like using power saving features of your hardware, using standby/hibernate or even spinning down hard drives!
These features work 100% only on few distros if any, and often you need to manually tweak them to fully enable.

I've got hard disk spin down working on Arch with no trouble. I think mileage can vary between hardware and software and for a beginner it's probably not as obvious as just installing the drivers that come with your hardware as most non-enthusiast Windows users will do.

You do however have better control over power saving with Linux. Parameters for pretty much anything can be set and as this is all available on the command line you can have a custom setup running transparently through daemons.

In terms of less CPU usage, I think not having the anti-virus, firewall, annoying little applications that insist on running in the system tray and other nagware that most Windows systems run it's going to help a lot.

Powering a peltier would add significantly to the electric load used by the system, wouldn't help more efficient use of the waste heat.

One could use a peltier as an electric generator (see for instance
http://en.wikipedia.org/wiki/Thermogenerator
http://en.wikipedia.org/wiki/Thermoelectric_effect ), but the thermal resistance of the peltier device would tend to make cooling less efficient, and you would need to make sure that it had a reasonable temperature difference across the device.

Would it generate enough energy to pay for itself, to make up for the energy involved in making, shipping, marketing, disposing of the peltier element?

Rather like the SPCR approach to noise - better to try to make it produce less noise to start with than to try to handle noise once produced. Likewise better to try to use less energy to start with.
For one thing - much of the energy is lost upstream from the "consumer" - generation losses, transmission losses, etc.
Using the heat directly (for climate control, etc.) might be a better bet.

There are lots of generators of waste heat - unless dealing with server farms, the heat generated by one personal computer is unlikely to be a significant part of your energy budget. (i.e. better payback by focus on efficiency and recovery in other areas - insulation, better control, more efficient climate control, transportation, etc.)