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

[hechacker1]

OS-X cap

What your benchmarks suggest to me is that maybe Apple is purposely capping the maximum wattage for battery life/thermal reasons.

[Wheels]

Hi Michael, there's some small problems in the graphs generated by PTS:
On the OpenArena, Nexuiz, John The Ripper graphs it's writen Milliwatts, Higher Is Better;
The last graph doesn't say which one is better;
The C-Ray graph doesn't show the end of Ubuntu's timeline.

And I think you should use something like Higher-Lower instead of Higher-Less, but that is just an opinion and I'm not the best with english anyway.

Keep up the good work

I'll add to this that I'd like to see time markers in the x axis. Much easier than eyeballing relative values or counting up all the tick-marks from the left.

[garytr24]

Scheduler arguments are a lot like religious ones.

[squirrl]

BSD and Linux

The BSD/Mach memory management backend on MacOSX throttles more of the major processes when compared to Linux's allot and swap scenarios.

The schedulers are vastly different. BSD/Mach seems to handle heavy loads where for a long time Linux's standard practice was to cease up until the task was completed.

My facts are only seen by the individual who is willing to install both and witness for themselves.

Copy a Nine-Gigabyte file from one folder to another. Applications will still open on BSD/Mach. Within Thirty seconds you'll be unable to do anything on Linux.

File-systems can also request a large amount of processor power. But it's up to the memory management and scheduling to throttle that request down the queue.

I for one will be glad when Torvalds, Moln?r, and crew go to Best Buy and buy a consumer model PC and witness for themselves what everyone has been putting up with for 10 years.

[mirv]

especially when they are on battery power.



Whereby samplig actually is a technique to measure signals that are faster than the measuring instrument.
So a sampling faster than the signal to be measured is not quite sampling anymore.

Your sampling frequency must be at least twice the frequency of that which you are sampling. It's called the Nyquist frequency.

[Ragas]

Most people don't run benchmarks or CPU/GPU intensive tasks all day after all.

especially when they are on battery power.

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.

Whereby samplig actually is a technique to measure signals that are faster than the measuring instrument.
So a sampling faster than the signal to be measured is not quite sampling anymore.

[sal-e]

Do you have a reference that shows that this effect is significant when it comes to Lithium Ion batteries. I was under the impression that the discharge efficiency of Lithium Ion batteries is relatively flat in the temperatures and loads usually exerted on them.

Yes, the key word is "relatively". Compared to NiCd and NiMH batteries the Lithium batteries are flat, but you still need to do optimization based on the battery type and construction. Apple engineers have access to all data directly from battery manufacture. FOSS programmers don't and only very small numbers of programmers understand the limitations imposed by battery chemistry and construction.

http://www.mpoweruk.com/performance.htm

scroll down to "Discharge Rates" illustrates the effect of capacity offset
If you look carefully the Phoronix results it is visible that Ubuntu don't have any power budget cap and OS X has. The power usage under Ubuntu is jumping all over, but OS X is much flatter. Apple has implemented power cap on kernel level. And I am sure this is because Apple had set battery life as priority. Other manufactures have other priorities like better performance. I think this opens opportunity for other manufactures to distinguish their offerings from the competition.

[mirv]

Do you have a reference that shows that this effect is significant when it comes to Lithium Ion batteries. I was under the impression that the discharge efficiency of Lithium Ion batteries is relatively flat in the temperatures and loads usually exerted on them.

You'd be surprised at the differences that are taken into consideration. I don't have the data sheets here, but I do use them at work.

[Shining Arcanine]

This is very true when you have unlimited power source with static efficiency. But here we are talking about battery. The battery capacity (Ah or mAh) is specified at optimal discharge current (Id). If you discharge the battery with greater current the battery heats up and it capacity drops rather quickly. So I suspect that Apple engineers has reduced the max/avg load by slowing down the "OS X" in order to optimize the battery capacity. Ubuntu and other distribution engineers don't have the needed data in order to do the same. They just optimize for speed. I wish I had the knowledge of the kernel and slowly over time to build library of profiles that every Linux user can use to have maximum battery life if they want.

Do you have a reference that shows that this effect is significant when it comes to Lithium Ion batteries. I was under the impression that the discharge efficiency of Lithium Ion batteries is relatively flat in the temperatures and loads usually exerted on them.

[Shining Arcanine]

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.

I agree. Those graphs indicate that Ubuntu is using less energy than Mac OS X for the same task. That means better energy efficiency.