Slashdot Mirror
IE8 Beats Other Browsers In Laptop Battery Life
WARM3CH writes "AnandTech tested a laptop with an AMD CPU, a laptop with an Intel CPU, and a netbook to compare battery life while running Internet Explorer 8, Opera 10, Firefox 3.5, Safari 4, and Chrome. They tested on simple web pages and flash-infested ones. IE8 had the best battery life on both laptops (followed by FF + AdBlock), and Safari had the worst battery life. On the netbook, Chrome was slightly ahead of IE8. The report concludes: 'Overall, Internet Explorer and Firefox + AdBlock consistently place near the top, with Chrome following closely behind. Opera 10 Beta 3 didn't do as well as Opera 9.6.4, and in a couple quick tests, it doesn't appear that the final release of Opera 10 changes the situation at all. Opera in general — version 9 or 10 — looks like it doesn't do as well as the other major browsers. Safari is at the back, by a large margin, on all three test notebooks. We suspect that Safari 4 does better under OS X, however, so the poor Windows result probably won't matter to most Safari users.'"
IE8 + adblock would give even better results!
Seriously though, how can you browse the web *without* adblock? I've shoulder surfed people doing it, and I'd rather eat my own hand.
Flash is a pig, no matter what browser you use.
Nothing beats Links or Lynx when it comes it this.
Did they have total control over exactly what ads appeared on the sites they visited (obviously excluding any of the tests running adblockers)? If not, then that introduces a decent variable right there. I give very little credence to tests like this one due to pretty obvious flaws in their methodology.
So IE8 is more battery friendly? Is that before or after having to install a virus scanner to keep an eye on what IE is doing?
Take Nobody's Word For It.
Given how far behind technically IE is otherwise, I think this is called "grasping at straws".
Internet Explorer is 'just' a shell around Explorer - all the components it needs are pretty much there and often locked into memory (which means not swapped out, and disk access is the mindkiller I mean batter killer). I imagine this is sufficient to cover the difference.
Still not giving up my Flashblock+Adblock+Noscript though. Especially on the laptop.
Install privoxy, configure it as proxy, configure it to block ads. Problem solved. And it's not really all that difficult, either. Ans is useful for other things (using TOR, etc.) too.
No, really. It's a meaningless number.
There's no control in this experiment (and, no, I don't mean "control group.") The fact that they were flogging away at public, and probably dynamic (read: inconsistent) websites totally invalidates the entire comparison.
If Anand wanted to take it seriously, they should have eliminated more variables. If they'd set up a dedicated, light-weight web server running in a controlled minimalist environment (bare Slackware+Apache, perhaps?) somewhere on a dedicated LAN, that would have been be a good start. They might even have used a RAM disk to ensure consistent access times to the data being served.
Hell: They should have even measured the battery voltage both before and after the tests, to eliminate (or at least quantify) any incongruity in the charging circuit's behavior. And they should've made sure to rotate their testing, so as to average it out as the battery ages (which it quite measurably will in these relatively-abusive full-charge - full-discharge tests).
But they didn't do these things. And it might seem like I'm splitting hairs here, but the results are close enough that hairs must be split.
Meanwhile, I think battery life while browsing is an interesting and very practical metric which is often overlooked these days. I applaud them for attempting and documenting such a feat, which I'm sure was relatively time-consuming, and I admonish them for doing a piss-poor job of it.
(And, no: I don't care which browser "wins." I have most of the tested browsers installed on my own laptop, and for me, it would be instructive to know which one will conserve battery life best in times when I know I'll be without power for a long period of time.)
Kid-proof tablet..
No, seriously, who cares?
If you get 2 hours of battery time, this gains you about 2 minutes and half.
For 5 hours of battery time you get 6 minutes extra.
If you really want to extend battery time, turning down the screen brightness by a notch will probably have more effect.
I skimmed the article and couldn't find anything on their method. There is a lot of relevent data that i simply couldn't see
1) Which has the lowest wakeups/bat usage once the page is rendered (some of us still read content instead of loading pages all day)?
2) Which has the lowest wakeups/bat usage on an active page, facebook,gmail,etc?
3) Which uses the least CPU/bat to render pages?
4) is there any difference in CPU/bat usage of flash?
Then there is so much to be asked about the method:
a)Was the environment controlled?
b)as i understand it, using a mouse uses quite a bit of battery, was all the navigation done using keyboard shortcuts?
c)was it scripted? (if so where all scripts equal?)
d)Was this done continuously rendering pages?
e)Does it really matter which lasts the longest if it takes longer to render the pages?/How long did each browser spend rendering pages? (e.g if chrome spend 10% less time rendering pages then you could end up with 13 more mins to actually viewing pages)
IranAir Flight 655 never forget!
Except that IE is not the slowest browser of all. In fact IE is probably the fastest browser in completing a page load and this is why MS picked this comparison when they claimed that IE is the fastest browser. As Mark Twain said "Get your facts first, and then you can distort them as much as you please." IE is very slow when running JavaScript but this is something completely different.
... those people aren't.
"Each test was run at least twice." If they were run at least 10 or 20 times you'd be able to estimate from the variance in the scores if the differences were significant.
The netbook had almost identical measures for all except Safari (caveat to significance, as above). Does anyone think it matters that the two laptops were running Vista and IE8, a fairly integrated collection of software, likely installed together, whereas all the others were thrown on top of an operating system that never could get the hang of running much more than itself.
Anyone want to put odds on whether the difference in drive activity in trying to (1) run MS operating system with MS vs. non-MS software and (2) run stuff installed together vs. installed after, would be proportional to the observed differences in battery life?
"I may be synthetic, but I'm not stupid." -- Bishop 341-B
You should use your favourite browser, because you'll be more productive in it.
Teach yourself to read or write 10% faster, and that'll dwarf the savings a different browser provides.
Ok. Let me get this straight. The conclusion is to use IE 8 because it uses the least battery life? Presumably, that implies (loosely) that it has the most effective algorithms for rendering modern pages. AnandTech should really compare apples to apples, and leave the orange out of the picture. What good is a modern browser that saves a bit of battery life, when it doesn't have a working Javascript garbage collector to free up memory on Javascript-heavy sites? I suspect that any user who's IE8 browser session just caused their Windows[File] Explorer to crash due to memory resource starvation might not care about how much battery life their IE8 session just saved them. I could be wrong, of course. "They tested on simple web pages..." --kudos. Because, that's what surfers are most likely to encounter on today's modern world wide web. My impression is that this study is seriously flawed, although I might have missed the point.
So you propose, AC, that adding more random noise to the test would improve the reliability of the results moreso than controlling the test environment?
I'm no statistician, either, but your proposal sounds like it would be more difficult (define "random" in this context, including time of day), more time-consuming, and less accurate.
Kid-proof tablet..