Ubuntu: Faster, But More Power Hungry Than Mac OS X?

[phoronix]

Phoronix: Ubuntu: Faster, But More Power Hungry Than Mac OS X?

Earlier this week I noted there's new Apple hardware in our labs being used to tighten up our Mac OS X support within the Phoronix Test Suite, OpenBenchmarking.org, Phoromatic, etc. However, in the middle of working on Iveland, I have been carrying out a few Mac OS X benchmarks comparing its performance under the 2010 Apple Mac Book Pro to other operating systems. With the Core i5 notebook being much faster than the past Apple Mac Minis used in comparisons like looking at their enhanced OpenGL stack and benchmarking Mac OS X against Linux and Windows 7, the results are more interesting and there's also a greater variety of testing possibilities now with the recent Phoronix Test Suite advancements. Next week there are some very interesting Apple-related benchmarks to be published, but before the weekend here are a few tests from this Apple Mac Book Pro looking at its power consumption under Mac OS X 10.6.5 and Ubuntu 10.10.

http://www.phoronix.com/vr.php?view=15521

[jsantala]

[b]Phoronix: Ubuntu: Faster, But More Power Hungry Than Mac OS X?

You should however take into consideration the "rush to idle" factor - even though Linux uses much more peak power it also finishes in a fraction of the time OS X used to complete the task, which means that completing task might have actually used much less power overall in Linux than it did on OS X. The Linux box could idle (or even sleep/shutdown) when OS X would still be hard at work, thus using very little power.

[Peterix]

Exactly. Integrate over time and I think you'll get significantly less power used for the same job done.

The real question is what the power usage is for some real tasks you'd be running on a desktop, where the computer is mostly idle. Most people don't run benchmarks or CPU/GPU intensive tasks all day after all.

[bug77]

You should however take into consideration the "rush to idle" factor - even though Linux uses much more peak power it also finishes in a fraction of the time OS X used to complete the task, which means that completing task might have actually used much less power overall in Linux than it did on OS X. The Linux box could idle (or even sleep/shutdown) when OS X would still be hard at work, thus using very little power.

QFT.

In other words, which system is drawing more power: one that uses 40W for 10s to get the job done or the one that only uses 30W, but for 40s?

[crispy]

Exactly. Integrate over time and I think you'll get significantly less power used for the same job done.

The real question is what the power usage is for some real tasks you'd be running on a desktop, where the computer is mostly idle. Most people don't run benchmarks or CPU/GPU intensive tasks all day after all.

Yeah, when benchmarking power usage you should also include regular use scenarios. Like 30 minutes of browsing the tubes or 30 minutes of typing in a text editor and writing emails.

[iro_]

As others have already said, to make a comparison you would have to integrate the instantaneous power draw over the time that it is needed to complete the task. However that's impossible to do since the CPU can change power state several times in a second and thus can power draw. Since you must have a sampling rate that's higher than the frequency of the phenomenon you are trying to measure, you can't really do that.

If you really want to show something valuable and not those useless graphs you would have to query the initial battery charge and the final one. That's far simpler and more precise.

[Petteri]

Those graphs also show that Linux and Ubuntu are well optimised and utilises hardware really well, while OS X leaves parts of hardware idling.

[babbling]
"Perfect" binaries or 3D engines would cause imminent meltdown because they would use power like Intel burn test or Furmark.
[/babbling]

[zouboulou]

it would be interesting to measure the power used by a real watt meter instead of the values reported by the kernel. it's difficult to know if the values reported are correct at all on both systems, they're just estimations

[vstrien]

When looking at how OS X throttles the CPU & GPU, it looks like a long-term version of the CPU-governor we know under Linux as "conservative". Would be interesting to see if the "conservative" CPU-governor makes much difference

But again, Linux here uses the (Intel-recommended) "rush to idle" - hereby using more power for the short run.. When linux for example would use 10 Watts when being in the "idle" state, Linux uses on average 9 Watts on the OpenArena benchmark ((5.1 * 10 + 10 * 40) / 50) - where Mac uses on average 39 Watts. Maybe it's a good idea to include the power usage of Linux being idle too (while OS X completes his benchmark)? Just a suggestion..

[vstrien]

When looking at how OS X throttles the CPU & GPU, it looks like a long-term version of the CPU-governor we know under Linux as "conservative". Would be interesting to see if the "conservative" CPU-governor makes much difference

But again, Linux here uses the (Intel-recommended) "rush to idle" - hereby using more power for the short run.. When linux for example would use 10 Watts when being in the "idle" state, Linux uses on average 9 Watts on the OpenArena benchmark ((5.1 * 10 + 10 * 40) / 50) - where Mac uses on average 39 Watts. Maybe it's a good idea to include the power usage of Linux being idle too (while OS X completes his benchmark)? Just a suggestion..

That should be 18 Watts of course, and the correct equation is ((51 * 10 + 10 * 40) / 50).