The Unexpected: Battery Life in OS X vs. Windows Vista

A while ago I started testing the Lenovo X300 with hopes of comparing it to Apple's MacBook Air. The review never saw the light of day but the testing was mostly complete. Obviously the X300 doesn't run OS X, but the MacBook Air can run Windows so I compared battery life between the two under Vista, which yielded some unexpected results.

There are no scripted battery life tests under OS X, while under Vista we have things like MobileMark - luckily, the tests I've put together for OS X can easily be reproduced under Vista. Let's take a look at my tests under OS X:

The wireless web browsing test uses the 802.11n connection to browse a series of 20 web pages varying in size, spending 20 seconds on each page (I timed how long it takes me to read a page on Digg and came up with 36 seconds; I standardized on 20 seconds for the test to make things a little more stressful). The test continues to loop all while playing MP3s in iTunes.

The DVD playback test is simple: I play Blood Diamond in a loop from an image on the hard drive until the battery runs out.

The final test is the multitasking workload. For this benchmark I'm downloading 10GB worth of files from the net (constant writes to the drive), browsing the web (same test as the first one) and watching the first two episodes of Firefly encoded in a 480p XviD format (Quicktime is set to loop the content until the system dies).

The system was set to never shut off the display and never go to sleep, although the hard drive was allowed to spin down when possible. The display brightness was set at 9 blocks (just over 50%), which I felt was comfortable for both day and night viewing.

The first test was easy to duplicate under Vista; web browsing is ubiquitous across platforms and iTunes has been cross platform for a while now. The same goes for the DVD playback test; instead of using Apple's DVD app I just used Windows Media Player 11 under Vista. And the final test also translates flawlessly to Vista.

Below are the results I got when I first ran the MacBook Air under both OS X and Vista:

  Wireless Internet Browsing DVD Playback Heavy Usage
MacBook Air (OS X) 4.98 hours 3.93 hours 2.7 hours
MacBook Air (Vista) 2.55 hours 2.05 hours 1.75 hours

 

Note that this is the same hardware and with the same brightness settings under both OSes. Vista's power management was set to Balanced and the display was set to never turn off under both OSes; the hard drives were free to spin down if possible.

The results are pretty staggering. The same usage model under both OSes results in a significant advantage for OS X. I basically got twice the battery life under OS X as I did under Vista. Now it is possible that Apple's power management is simply more sophisticated under OS X and not optimized for Vista, but what inspired me to include this in today's review was actually something AnandTech's own Jarred Walton brought up in a meeting earlier this week.

In testing the first batch of Centrino 2 notebooks that Jarred received he noted that he can't seem to find a mainstream notebook with a 50 - 60WHr battery that can come close to offering the sort of battery life you get out of the Macs. Even in his idle tests (just leave the computer at the desktop without doing anything or putting it to sleep) Jarred has been finding many mainstream notebooks only seem to last 3 or 4 hours at best. He asked me to run a simple test on the latest MacBook Pro just to confirm his findings: play a DVD under OS X then Windows Vista and see how battery life is impacted by the OS change.

Now I'd already ran this test on the MacBook Air earlier this year, but I had assumed that there was something wrong with my data. Repeating the DVD test (this time using an actual DVD of Sin City in the drive) I measured battery life for looping the entire movie (minus credits) on the new MacBook Pro:

  DVD Playback
New MacBook Pro (OS X) 3.07 hours
New MacBook Pro (Vista) 1.5 hours

 

Under OS X the new MacBook Pro lasted just over 3 hours while playing the DVD, but under Windows Vista I got a total of 1.5 hours. All of the bootcamp drivers were installed and the OS was as clean as could be with no additional background tasks other than what loads by default with a standard Vista Ultimate 32-bit install.

Now it's possible that Apple's notebooks may be at a battery life disadvantage under Vista vs. OS X. To find out here are the results for the Lenovo X300 compared to the MacBook Air under Vista:

  Wireless Internet Browsing DVD Playback Heavy Usage
MacBook Air (OS X) 4.98 hours 3.93 hours 2.7 hours
MacBook Air (Vista) 2.55 hours 2.05 hours 1.75 hours
Lenovo X300 (Vista) 2.82 hours 2.18 hours 1.68 hours

 

The Lenovo X300 actually offers similar battery life to the MBA under Vista, despite shipping with a 27WHr battery vs. the 37WHr unit in the MacBook Air. The comparison isn't that cut and dry however; the X300 uses a Core 2 Duo L7100 with a 12W TDP compared to the 1.6/1.8GHz 20W TDP processor in the MBA. Overall platform power consumption is lower on the X300 than on the MBA and thus the numbers here seem to support my point. The X300 manages to last a bit over 2 hours during the DVD playback test under Vista, while the MacBook Air can pull nearly 4 hours under OS X (despite also staying alive for ~2 hours under Vista).

Figuring out why OS X seems to be better for battery life is nearly impossible, at least without the aid of both Apple and Microsoft. I've brought up this topic with a handful of PC OEMs in the past and they haven't been able to shed any more light on things, other than to confirm that Vista is a strange beast. It's quite possible that Vista's constant performance optimizations are preventing CPU and platform power management techniques from being effective, but that seems a little too simplistic of a view.

All I can do for now is report the numbers as is. An unexpected benefit of OS X appears to be better battery life. Go figure.

Battery Life: Take Two Steps Forward, and Two Steps Back Final Words
POST A COMMENT

65 Comments

View All Comments

  • MacMatte - Sunday, June 21, 2009 - link

    For those of you who insist that Apple brings back the matte screen option, please leave a comment at http://macmatte.wordpress.com">http://macmatte.wordpress.com - it's a website solely focused on the issue of bringing back the matte screen. See the number of pro-matte comments already at the MacMatte website. Reply
  • drbrady63 - Thursday, June 18, 2009 - link

    I am trying to identify if a new macbook pro 13" would be adequate for editing with Final Cut Express, and for that matter, Final Cut Studio. Unfortunately, it has a 5400rpm hard drive and that is not good for editing. But, I wonder if an optional ssd would be fast enough??

    I would use the 13" for more mobile work and dock it with a larger monitor for more involved editing work.

    Any thoughts on this would be appreciated.

    Dan Brady
    Reply
  • richmoffitt - Sunday, November 09, 2008 - link

    This is an uneducated guess, but I'm pretty sure that Quartz works in ways similar to X11, where changing graphics drivers requires a restart of the window manager.

    You're right though -- this is only a software problem and can hopefully be fixed in the near future (if it's a big enough issue for their user base anyway).
    Reply
  • scipi - Monday, October 27, 2008 - link

    Hope the quality of the components is better than the first gen MacBook Pro's. Mine is on its second H/D, gone through 2 logic boards and now needs a third, this time outside of warranty. Wont be buying another Apple again which is a pity because OSX is great. Reply
  • Zebo - Saturday, October 25, 2008 - link

    Vista is bloated resource hogging junk - You should have tried the OS many of use still use - Windows XP for battery life. I get over 4 hours on my R31 thinkpad with winxp pro.
    Reply
  • Ronbo13 - Saturday, October 25, 2008 - link

    You photos comparing the glossiness and reflection on the screens was not fair, though. Please notice that the laptop on the right (the new MBP) is reflecting a portion of wall that has direct sunlight shining on it, and the laptop on the left is reflecting stuff that's in shadows. So even if the screens were equally reflective, the one on the right would show tons more reflections.

    Come on, people. Normally you guys pay more attention to details. That's just sloppy.
    Reply
  • ioannis - Saturday, October 25, 2008 - link

    nop, you are wrong. Both of them reflect stuff that have direct sunlight. Notice Anand's reflection for instance, or the wall on the left hand side of the old MB and the wall on the right hand side on the new one.

    I'm referring to this:
    http://images.anandtech.com/reviews/mac/MacBookPro...">http://images.anandtech.com/reviews/mac/MacBookPro...
    Reply
  • Enrox - Saturday, October 25, 2008 - link

    Anand, why don't you test Vista installing it on the new MacBook Pro without using Bootcamp, you need to wipe out the drive and create a MBR partition and use Vista x64 SP1 (it supporta EFI), the only thing you need to know is that at startup you have to press the Alt key and manually select the Windows disk in order to boot from it, beside that everything else seems to work just fine with the Vista native installation (tested on a white MacBook Penryn 2.4 GHz 4GB ram).
    It would be very interesting to see if you get the same exact battery life numbers bypassing Bootcamp.
    Reply
  • JonnyDough - Friday, October 24, 2008 - link

    Until Apple stops being so shady, I won't have anything to do with them. Reply
  • aos007 - Friday, October 24, 2008 - link

    I have brand new Vaio Z laptop and I can get 5 hours battery life IF I disable Vista sidebar. It does not matter whether there's no widgets running, it seems to use 10-15% of CPU time no matter what. This translates into a big loss of runtime - I'd get 3.5 hours versus 5. Unfortunately, I like Sidebar as there are some useful widgets, as well as for eye candy so I feel Vista is crippled without it.

    So the question is whether you disabled Sidebar during Vista testing? I am guessing not since it runs by default and if so, that may be part of your answer.
    Reply

Log in

Don't have an account? Sign up now