Apple's 15-inch 2010 MacBook Pro: More Battery Life Tests, High Res Display Evaluatedby Anand Lal Shimpi on April 24, 2010 1:57 AM EST
- Posted in
- MacBook Pro
Snow Leopard Battery Life Issues with Flash: Fixed
Snow Leopard brought forth 64-bit versions of many Apple applications. Finder, TextEdit and Safari (among others) could now run in x86-64 mode. However, last year I found that browsing websites that used Flash with 64-bit Safari dropped battery life by over 40% in Snow Leopard compared to 32-bit Safari in Leopard. To fix the problem you had to force Safari to launch in 32-bit mode.
While Apple never recognized the problem nor discussed why it happened to begin with, it appears that it’s since been addressed in OS X 10.6.3. The results below tell all:
|Snow Leopard Battery Life Improvement|
|Flash Web Browsing Battery Life||OS X 10.6.1||OS X 10.6.3|
|15-inch MacBook Pro (Mid 2009) - Core 2 Duo 2.53GHz||277 minutes||344 minutes|
Flash web browsing battery life improved tremendously from the original release of Snow Leopard. While it's still not as high as in Leopard, it's a definite improvement.
Windows 7 Battery Life
A handful of users asked that I look at battery life under Windows 7. Being your humble servant, I of course obliged. I attempted to recreate my OS X benchmarks under Windows 7 as best as possible, using Chrome and Windows Media Player in place of Safari and iTunes of course. You can’t draw any conclusions about OS X vs. Windows 7 battery life from these numbers however. As I mentioned in the original review, under Windows 7 the MacBook Pro keeps the discrete GPU enabled 100% of the time. There’s no way to shut it off. Battery life will inevitably be lower than OS X.
|Windows 7 Battery Life|
|Light Web Browsing||Flash Web Browsing||XviD Playback|
|15-inch MacBook Pro (Mid 2010) - Core i7 2.66GHz||255 minutes||255 minutes||153 minutes|
It appears that the dGPU being on all the time is the reason we can't get better browser battery life regardless of workload. The XviD playback test also takes a hit compared to OS X thanks to the discrete GPU. In our previous articles we found OS X to be roughly equal to Windows 7 in XviD battery life.
Post Your CommentPlease log in or sign up to comment.
View All Comments
ltcommanderdata - Saturday, April 24, 2010 - linkIt's kind of disappointing that the Core i7 takes a noticeable battery life hit compared to the Core i5 considering they both in the same 35W TDP target. Could you double check the clock speeds of the GPU in the Core i7 model? The low-end Core i5 model you previously showed a GPU-Z shot of was definitely underclocked, but perhaps that's because it's replacing the previous pure 9400M model, whereas higher model MacBook Pros might have higher clock speeds which could help explain the battery life difference.
As well, since you have Windows installed on the MacBook Pros, could you do your round of mid-range laptop gaming comparison benchmarks as was done for the ASUS G73Jh? People don't buy MacBook Pros specifically to game, but people with MacBook Pros no doubt do play games, so I think it'll be useful information. Running it against the Mid-2009 17" MacBook Pro you have would also be good to see the improvement of the GT330M over the 9600M GT.
I'm also curious to see if the limited thermal room of the MacBook Pro limits the effectiveness of Turbo Boost. It might be interesting to chart the CPU frequency of the Core i7 over an hour as it cycles through an intensive benchmark to see if it can hit the top Turbo Boost bin (3 bins for 2 cores) and how long it can sustain that before the heat sinks saturate and the clock drops back down to normal. Of course, without comparison to other laptops with more thermal room, it's hard to tell the "normal" or optimal Turbo Boost behaviour of the Core i7, but it should be something manufacturers should consider to differentiate themselves from others using the same components.
aj28 - Saturday, April 24, 2010 - linkAnand,
Pardon if you've already covered this and I missed it, but have you done any tests comparing the effects on battery life by the use of Flash 10.1 versus Flash 10.0 on any of the Apple/Optimus GPU-switching i5/i7 notebooks? If Flash 10.1 kicks on the dGPU, is there a chance that the i5 might be able to power through it with ease to the point that you could actually get better battery life by not using GPU acceleration?
I know that goes against common wisdom, and I'm sure you would be better off if we were talking about a low-power integrated chip, but it seems to me that the mid-to-high-end dGPU chips are horribly inefficient for this type of thing in cases like these where they can, unlike a desktop, simply be switched off completely in favor of an iGPU which has to be powered all the time anyway, even if in a low-power state.
Thanks for your reply if you get around to it!
flgt - Saturday, April 24, 2010 - linkI think you've already stated why they won't put much effort in OS X. It's a huge population of STUDENTS. And students don't generate revenue like fat corporate and government accounts.
I’m sure the Mathworks don’t want to artificially limit their customer base, but if it is coming down to a decision between OS X or Windows/Linux it’s obvious which one will take priority.
For the vast amount of companies Windows and Linux deliver equivalent or better hardware at much better price points. They also host more of the client applications that drive productivity, which could be argued to be more important than the OS itself.
You could argue it is important to reach students who one day might be paying corporate users, but if a student learns MATLAB on their Mac they can easily be just as productive on a corporate PC. There is no real incentive to make the experience better for a minority of non-paying users.
mbene12 - Sunday, April 25, 2010 - linkI know it is totally off topic, but this is exactly why I left Matlab for python. Their lack of native Mac support. As I transitioned from student to professional I just didnt see a reason to pay $4000 a year for base license and libraries which performed poorly when I could get most of what I needed done in a mix of Python and C for free.
erple2 - Monday, April 26, 2010 - linkI think that thinking is a double edged sword. Students (and more importantly, Universities) are a major piece of the Matlab licensing revenue. I know that the Educational discount is pretty decent, but that's still a significant chunk of change. I'd wager that if you looked at the total number of licenses the Mathworks distribute, Education would be an even player with Corporate and Government levels.
The other argument (that you can get more hardware for the money) has (almost) always been the case with Apple, however. There were a few notable exceptions (Apple had the first Core2 based desktops and laptops - I can't remember if they were on the forefront with the i7 based desktops, though), plus when DDR3 first came out in a SODIMM format, Apple was curiously reasonably priced.
rawd - Saturday, April 24, 2010 - linkHi Anand, Cody Kreiger on Macrumors has already written an app to monitor GPU usage and it resides in the taskbar
bitninja - Saturday, April 24, 2010 - linkThanks for the mention rawd! Anyone interested in seeing the MacRumors forum thread where it's being discussed and fleshed out can visit it below until I get a formal page up on my website:
Thanks, be sure to leave a comment with your thoughts if you try it out!
rcocchiararo - Saturday, April 24, 2010 - linkIs it posible to get a 15 inches macbook pro with a glossy/matte high res display from, lets say.. amazon/bestbuy/newegg/whatever ?
or only directly from apple ?
Computer Scooter Joe - Saturday, April 24, 2010 - linkThanks for the info, just one more question. Do the fans spin up as a result of the dGPU being on? Or are they generally fairly quiet?
I understand that it depends on the application, but im talking about just idiling at desktop or web browsing or working in Word.
maxxl - Saturday, April 24, 2010 - linkI can't post this command here unfortunately, becouse of an error that appears when I try to, so look at this thread for "ioreg -lw0" command:
With the model number you can find manufacturer easily, i.e. here: