Ubuntu 10.04 Is More Power Hungry Than Windows 7

[Kano]

Why does nobody try to change the backlight? Use

/proc/acpi/video/PEGP/LCD/brightness

to control it and test again.

[Craig73]

So where to from here???

I appreciate the testing Michael is putting out there, we definitely need comparisons to be done amongst hardware, linux distributions, and operating systems themselves. I see so much commentary (pro and anti linux) that seems baseless, and without numbers there is no way to track improvement, or even to discuss what improvement really means.

But I am left feeling like I have no idea if the default in Windows/Ubuntu are similar or reasonable, and how they differ between power profiles. I have seen people comment that they can tweak extra power savings from their machine, but they haven't stated the trade-off's they are making, and the impact on performance/responsiveness, so that we can assess if the if costs associated with those power savings are acceptable. And it is really unclear what the true power savings are once normal usage is in place (if MS has indexing off by default and Ubuntu on, or antivirus software isn't installed, then the comparison doesn't have real world application)

But perhaps this will push the community to start trying to answer such questions, either by making the default power setting assumptions (and trade-offs) more obvious, or providing better defaults. I would also hope that Michael would start taking the tests to the next level of trying to see what the impact on battery life is, especially under normal usage scenarios (indexing on, reasonable level of app usage, normal utilities installed and running such as antivirus, etc.)

[BlackStar]

Talk about terminal overkill

Why does nobody try to change the backlight? Use

/proc/acpi/video/PEGP/LCD/brightness

to control it and test again.

Or simply use the brightness buttons on your laptop.

[BlackStar]

I appreciate the testing Michael is putting out there, we definitely need comparisons to be done amongst hardware, linux distributions, and operating systems themselves. I see so much commentary (pro and anti linux) that seems baseless, and without numbers there is no way to track improvement, or even to discuss what improvement really means.

But I am left feeling like I have no idea if the default in Windows/Ubuntu are similar or reasonable, and how they differ between power profiles. I have seen people comment that they can tweak extra power savings from their machine, but they haven't stated the trade-off's they are making, and the impact on performance/responsiveness, so that we can assess if the if costs associated with those power savings are acceptable. And it is really unclear what the true power savings are once normal usage is in place (if MS has indexing off by default and Ubuntu on, or antivirus software isn't installed, then the comparison doesn't have real world application)

But perhaps this will push the community to start trying to answer such questions, either by making the default power setting assumptions (and trade-offs) more obvious, or providing better defaults. I would also hope that Michael would start taking the tests to the next level of trying to see what the impact on battery life is, especially under normal usage scenarios (indexing on, reasonable level of app usage, normal utilities installed and running such as antivirus, etc.)

I think the idea is to develop PTS to the point where it becomes capable of such comparisons. I don't really think wide, cross-distro and cross-hardware testing should be Michael's responsibility.

In fact, distro/kernel/whatever-developers should take advantage of it to improve what needs improvement. The current tests are mostly a demonstration of the new PTS capabilities.

[V!NCENT]

Michael, in what performance mode was Windows 7 in?

Windows 7 reduces the speed of the amount of instructions per time based on the settings. Yeah I am not kidding... This could mean nothing...

[Kano]

@BlackStar

Those do not always work also you can test specific values better.

[yankeeDDL]

And this is the point why such articles are complete rubbish. People only see the numbers but never really read the article. And they also never read the complete discussion following the article.

Phoronix only tested the power consumption on AC-power, not the consumption when running on battery!

I'm no expert, and as much as I despise Windows, I can't say that this article is rubbish.
It proves that, at least in two cases, Windows has a better power management than Ubuntu and it also gives a fairly reasonable explanation: drivers.
It's fair to expect that Linux's drivers are less performing than Windows and therefore there isn't really much that can be done, other than pointing this out, accepting it, and trying to push ATI/Nvidia/Intel to make better open-source drivers.

Fact is, Linux's weakest point, today, is probably the gaming/graphic part. If you're a hard-core gamer, you're probably better off with "Win-tendo". Otherwise ...

[monraaf]

I'm no expert, and as much as I despise Windows, I can't say that this article is rubbish.
It proves that, at least in two cases, Windows has a better power management than Ubuntu and it also gives a fairly reasonable explanation: drivers.

The problem is that's just speculation. There are too many factors involved. Maybe Windows has indeed better power management for some notebooks. But then this should be investigated, and things like default brightness and ac versus battery power differences should be taken into account. So for me the article was not very useful.

[yankeeDDL]

The problem is that's just speculation. There are too many factors involved. Maybe Windows has indeed better power management for some notebooks. But then this should be investigated, and things like default brightness and ac versus battery power differences should be taken into account. So for me the article was not very useful.

Why do you say speculation?
They measured power consumption on the same machine with two different OS. You can say that there are many reasons that explain the difference, but the measurements are not speculation: that's data.
And sure, if they used different brightness settings that'll be pretty stupid, then again, I bet that if the results were reversed, everybody would be chanting how good Ubuntu is over Windows instead of trying to split hair.
It's certainly a limited test case, and the data collected is also somewhat limited, but the outcome is crystal clear, and especially for the Eee, the gap is so wide that there isn't much "tweaking" that can fix it": there's clearly something that is burning power and it is not managed in Ubuntu.

[monraaf]

Why do you say speculation?
They measured power consumption on the same machine with two different OS. You can say that there are many reasons that explain the difference,

That's the whole point. The interesting thing to me is not that Windows 'won this contest'. But why did it win, and that's where facts matter, and speculation won't do. But hey if all you care about is a pissing contest between two operating systems to declare a winner then I guess this article will do just fine for you.