Slashdot Mirror
Browser Power Consumption Compared
theweatherelectric writes "Over on the IE Blog they've posted a power consumption comparison of the five major browsers. They write: 'Power consumption is an important consideration in building a modern browser and one objective of Internet Explorer 9 is to responsibly lead the industry in power requirements. The more efficiently a browser uses power the longer the battery will last in a mobile device, the lower the electricity costs, and the smaller the environment impact. While power might seem like a minor concern, with nearly two billion people now using the Internet the worldwide implications of browser power consumption are significant.'"
At least at this point in my computing; I'm all for low power consumption computers that are small and quiet.
That said, as far as browsers, I run Chrome.
Gone!
It is on msdn.com. Can we consider this a partial and fair article? I'm asking, not accusing.
Efficiency (is greater than) Features.
Since when could slashdot not show a greater than symbol?
Gone!
Since they're not the fastest, they're claiming their the most power-friendly.
"We did it on purpose.. see?"
If you can't compete on innovation, and you can't compete by bullying standards bodies, and you can't compete by leveraging your monopoly, and you can't compete on performance, and you can't compete on security....well, at least you can say you use less power.
And yes, when you work for the same company that wrote the freaking operating system, one would hope that IE would use the least amount of power.
Whatever.
Use the HTML entity > to get >, < for <, and so on. Slashdot accepts most common HTML entities, but alas—not unicode.
How can I believe you when you tell me what I don't want to hear?
When I'm on my netbook, I want a browser that gives me the most battery life possible. Unfortunately my netbook doesn't have meaningful GPU acceleration, so their comparisons don't do much for me. Is IE9's rendering anywhere near as power-saving with software rendering? They also don't account for the battery saved in FF/Chrome by blocking intrusive graphical ads and their related javascript/flash. They also don't test real-world Javascript-heavy web apps like Gmail, or having multiple tabs open/opening at once.
The graphs also blow the differences out of proportion. The Chrome/FF/IE numbers are all within 15% of each-other most of the time, while the graphs make IE9 sometimes appear with a very wide lead of half the power usage.
Either make a > sign with >, or set your Comment Post Mode to "Extrans".
Since when could slashdot not show a greater than symbol?
Um... when did Slashdot support greater-than characters in comments? Try the HTML entity, > (>). You may also be interested in less-than (<) and ampersand (&). Others can be found here.
"The state is that great fiction by which everyone tries to live at the expense of everyone else." - Bastiat
IT'S A TRAP!
Posting on a M$ blog about an M$ product? Don't you believe it.
Wrong. If I write crappy software that hits your laptop's disc every 10ms, is the laptop the inefficient part? The hardware guys need to invent better speculative laptop disc caching technology? Of course software can consume too much power.
OMG.. that right-hand picture.. In FF4, this end of the rack spaces are almost perfectly in line with the scrolling demarcations. Rapidly scrolling up and down makes it look 3D-esque and screws with my mind.. :O
I just hope I wasn't the only one tripped out by the visual effect. :)
"The true measure of a person is how they act when they know they won't get caught." - DSRilk
Well, when you program a browser to do less than everyone else's I would assume it doesn't need as much power.
The real Sig captains the Northwestern. This one captains
Keep it simple, just disable flash in entirety. You will reduce your power consumption as flash is a poorly coded. Run the numbers with and without flash and see what a difference it makes. This is arguably one of the reasons Steve Jobs won't allow flash on the IOS...
While power might seem like a minor concern, with nearly two billion people now using the Internet the worldwide implications of browser power consumption are significant.
I wonder when the day will come when the government starts mandating energy efficiency requirements in software, much the same way they do appliances, cars and other things. I wonder if such rules would apply to open source, or other freely exchanged software.
If I can be modded down for being a troll, can I be modded up for being an orc, or a balrog?
Anandtech:
The amount of power your hardware uses is not constant, it depends on what you run on it. If you run something that's CPU-intensive then your CPU will consume more power than otherwise.
The power rating on your PSU is not constant, it does not suck the rated power at all times and waste the unused portion as heat. The rating is a maximum.
That said, considering all the power wasted by an OS like Windows and whatever other programs you may have resident in RAM for convenience's sake, and plug-ins like Flash that are ubiquitous even on netbooks, measuring the difference in a browser's power consumption is probably laughable.
Software can have a large impact on power consumption because, quite simply, your hardware consumes less power when it does less. So giving your system more work requires more power.
There are many reasons for this, from the architecture of modern CPUs to memory and disk usage. If you use a program that is inefficient and takes more operations (say O(N) vs O(2^N)) then power will be wasted, regardless of what hardware it is executing on.
NEVER to post stories that link to companies' PR releases? I don't care if it's Microsoft, Google or Mozilla, I think it's quite evident that these comparisons are all very biased (and in two days you'll have similar comparisons from the other browser makers with exactly the opposite results).
See subject run.
This is why you need to build a "mobile web browser" as opposed to just a "browser". This is why Myriad, Isis and Polarity browsers were bought out - they provide a specific function for a select group of devices at the right power level with plenty of functionality.
There's simply no way to put IE9 on most devices and expect anything pleasant to happen to your battery.
When the foot seeks the place of the head, the line is crossed. Know your place. Keep your place. Be a shoe.
Comment removed based on user account deletion
http://upload.wikimedia.org/wikipedia/commons/c/c0/Blank.gif
... functionality to follow real soon now
Possting anonymously not to whore karma.
What's the situation on IE integration in Windows these days? Is it still true that IE is really kind of always running when you start up Windows, like it was in Windows 98 or XP? If so... could you say that running just IE is more power efficient than running other browsers along with it?
Pretty good is actually pretty bad.
They completely missed the fact that the only browser in their list which requires Windows is IE9. I'm guessing that ANY of the browsers would beat IE9 hands down if it were running on linux instead of Windows 7.
1. make it compile
2. make it work
3. make it fast
Hitting the disk every 10m incurs a performance penalty.
Not necessarily. If nothing else is using the disk and you spawn a thread to do nothing but sleep 10ms, seek to a semi-random spot on the disk, and write "Hey, hard drive, what's up?" you will have no noticeable performance problems until something else needs the drive.
You could do nonsense math in a loop in a background thread, which in a multi-core system would heat the processor up good and toasty without any real performance hit as long as the other 3 cores are idle.
Neither of those would actually ever happen, but functionally equivalent operations implemented by incompetent boobs could do something similar. To a lesser extent, even a competent programmer, knowing that normally there's a ton of computational power to spare, might not give a dang that his function is sucking up 20% more CPU than it needs to.
I have a laptop that's just fine with 65 watts, and a desktop that barely gets by with 750 watts.
It's not the software that sucks down the power; it's the hard drives and graphics card.
There's no -1 for "I don't get it."
There are certain flash ads on some web pages which make the fan of my laptop turn on. There are also badly written javascripts which do the same - for essenrially doing nothing.
Google is Skynet and I am a Google Borg unit. I have not used IE (Internet Excuses) since FF came out. Now it is Chrome all the way. I wish Google would have accepted my passioned plea to have been one of their testers on the Chrome LT. I am Droid X, My Wife is Droid X, my Son is Droid. Skynet has taken over my entire house and family.
All of these tests appear to have been done with a wired LAN setup, but power consumption matters most when you're mobile, and the power draw from Wi-Fi will far outstrip the draw from any browser in typical usage. Who cares which browser is more power efficient if the technology you need to access the Internet in the first place draws several orders of magnitude more power?
Your graphics card will not be sucking down power unless you run software that actually uses it.
{inquisitive commercial voice}
If we really wanted to optimize power consumption, wouldn't we blindly stick to simple text-over-the-web?
{/inquisitive commercial voice}
{commercial authoritative voice}
NO! We homo sapiens have complex needs and wants that do not simply conform to plain text. Let us consume RICH MEDIA!! {br}
Windows 7 PERFORMS BETTER than any other version of windows, and it takes MORE ENERGY to do so!!! {br}{hr}
MORE is BETTER!!!!
{/commercial authoritative voice}
{cue to an excited actor-CEO leading a chant}
MORE-elopers!!!!! {b}MORE-elopers!!!!!!{/b}{b}{i}MORE-elopers!!!!!!!{/i}{/b}
{/cue to an excited actor-CEO leading a chant}
{cue to an actor-nerd, sitting in the basement, which is really a soundstage, typing on his workstation, which is really just a monitor and keyboard hooked up to an iPad, inexplicably the Mac Apple glow emanates from the side of the monitor}.
I have come up with a power-friendly optimization for my browser. The WORLD is my OYSTER!!!!!!!! {!--Steven, this is an in joke because nerds don't get to see many "oysters," if you know what I mean--}
{/cue to an actor-nerd, sitting in the basement, which is really a soundstage, typing on his workstation, which is really just a monitor and keyboard hooked up to an iPad, inexplicably the Mac Apple glow emanates from the side of the monitor}.
{fade to black}
{!--{/fade to black} wait, this doesn't make any sense. How do we resolve this?--}
Yeah, kinda laughable in the bigger scheme of things, BUT I have to hand it to Microsoft for their part in getting Adobe to fix a problem where Flash prevented Windows from autosleeping. That was huge energy waste when multiplied across probably millions of home PCs that would normally have been asleep for at least 12 hours a day.
What's more, that was part of a wider Microsoft effort that promotes efficiency across their products, including in data centers where a 10-15% efficiency win can really add up. So a begrudging kudos to MS on this -- and I hope FF kicks IE's lilly white hiney in every regard next go-round!
responsibly lead the industry in power requirements.
Why is being energy efficient so frequently expressed in the most ingratiating and sanctimonious terms? I like using less power, too, but I'm not going to pretend for a minute that it makes me a more moral and deserving human being.
I think like most geeks, getting more work done with less energy input is inherently valuable -- at a minimum your batteries last longer. But I can't help but want to waste energy when energy efficiency becomes a question of faith, and I'm pretty sure a lot of other people who would otherwise find great appeal in what essentially amounts to getting more for less are turned off by it as well.
Next go around? With Flashblock and AdBlock it already does in the real world, because it's doing probably 5% of the work on the real web. Same goes for Chrome, the most efficient work is the work I never have to do.
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
time to trade in your HBGary Astroturf license. this one's failed.
They really should try flash heavy sites like YouTube.
I can have my battery life cut in half when using Chrome 10 on YouTube; so much so that I actually have to switch back to Firefox for extended browsing when I'm on the road. It's pretty poor because even if the video has stopped and it becomes an idle page it can still sit at 10-15+% while doing absolutely nothing (so I don't see how they can claim rendering speed is the cause).
What type of memory are they using in a useful working computer system that only draws 0.257 watts?
IE9 currently is confined to Windows Vista and Windows 7, the two most bloated, power-hungry versions of Windows around. Maybe Microsoft should start telling the billions of computer users to ditch Microsoft Windows and move over to a more efficient, less resource-hungry, operating system.
Testing about:blank is one thing, but how about typical usage: At least three sites displaying different sites. As it is, we're testing the effeciency of the rest of the browser framework as much as anything, whereas with several heavy tabs running, you'd be able to test how well it scales up to normal workloads.
Also, it would have been nice to see power consuption graphs for running a Youtube video(flash vs webm vs x264) in the various browsers, using plugins as needed. I mean, a very common web usage scenario is playing one or more youtube videos in a row. Running a JS benchmark? Once or twice a year max for most people.
Also, I see their graphs as showing Firefox winning. IE doesn't even count; It doesn't run cross platorm(Windows, OSX, X-based Linux at least).
What I want to know, however, is how to -increase- consumption for better performance: Sure, on a laptop you want battery life, but on your big honking desktop? I'd take performance over effeciency any day - I've got plenty of spare cycles it can use.
I'd have liked to see pages that specifically use flash heavily (youtube or some kind of flash-based game) compared here too. I've noticed that Flash in Chrome causes my fan to spin up like crazy. Other browsers, not quite so much.
Mozilla Firefox is crashing almost daily for me, between full crashes and times that it leaks enough memory that it hits 1.5GB and stops responding, or burns the whole CPU core and nearly stops responding. Firefox 4.0 is a bit better than FF3.x - at least sometimes when it crashes it's not burning the whole CPU, or not leaking all the memory it can get. There's really no bloody excuse for it - you'd think this was the 20th century or something, and we knew better back then too.
I've moved about half my work over to Google Chrome, though there are things that don't work reliably on Chrome and information that I don't trust Google with. Occasionally it'll crash and turn all the pages into "Oh, Snap!", especially for random news sites like you get by opening up everything in FARK in separate tabs. One other important difference between Chrome and Firefox is that Chrome splits things up into more processes, so damage is a little better contained, but there's no easy way to say "kill ALL the Google Chrome, restart and restore all the pages" the way there is with Firefox.
IE7 has been reliable, and at least it has tabs, but it's pretty lame and doesn't give me all the ad-blocking script-blocking protection that the other browsers do, so I only trust it for a few work-related pages and a few pages that are IE-specific and fail on other browsers.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Where this really matters is for mobile phone and tablet platforms, where people expect days of use without a recharge. Moreover, on mobile devices, the display and radio dominate power consumption especially when using a web browser.
Run, subject, run.
"This post contains words, known to the State of California to cause thought. Wash brain thoroughly after reading."
I'd say this is unbelievable, but it's all too believable -- grasping at anything to show that we are better than them.
Just how is this relevant in the real world? Oh, you measured a difference -- but is this measured difference repeatable over repeated runs on the same system let alone repeated runs on a wide cohort of systems, and is this difference really significant to a user? Let's see -- a ten minute difference over a 220 minute period, that's a whopping four percent! Run these tests a few more times and tell me if that difference is statistically significant (let alone clearly outside the expected margin of error) on the same machine, let alone over different machines.
Oh, and that run time graph is right out of "How to Lie with Statistics" -- cutting off the zero to make the difference in the length of the bars look more significant than it actually is.
I've got a much better one -- this whole "green" thing only appeals to people concerned about energy conservation. There's a bigger audience out there to grab for...
Let's take the current versions of all these browsers, and count all the "1" bits and all the "0" bits in the binaries.
Wouldn't it be great, absolutely stupendous, if IE9 had the smallest difference between the numbers of "1"s and "0"s? Then we could claim, correctly, just as correctly as with this energy consumption thing, that IE9 is the LEAST DISCRIMINATORY browser out there! All those other browsers out there DISCRIMINATE against "0"s (or they discriminate against "1"s) -- just look at the numbers! Do you want to use a browser that is blatantly discriminatory?
So now you could claim that IE9 is not only greener but it's also less discriminatory!!!
Well, unless you count discriminating on the basis of intelligence...
Your system doesn't really use 750 watts. Depending on the configuration it barely uses 80-150 watts on idle and when browsing the Internet and only when you play games or do some heavy stuff like video encoding, it gets close to 300 watts.
People got accustomed to power supplies with big watts number because in the past manufacturers were actually lying about how much their power supplies can "deliver" but nowadays there is very little difference in the price of components for a power supply, between let's say a 400 watts power supply and a 750 watts power supply, so it's not worth it for manufacturers to make "quality" power supplies at low wattage.
If you remove the dedicated video card and use a motherboard with integrated video with not so many overclocking features (like 8+2 vrm chips for the cpu, which cause power loss), change the processor to a dual core close to what regular laptops use and enable the power saving features, you'll get close to an average of less than 100 watts. Won't get close to the power efficiency of a laptop because the power supply is designed to get peak efficiency at about 50% of it's rated power, not at 100 watts.
Very low power usage. I guess some windows updates go through IE, but nothing else.
So you missed all of the Total Power Consumption graphs then - their Y axes ranged from 10 to 15-26 watts.
Firefox 4 is still tops in speed, power consumption, and "mod-ability". I've used chrome, opera, and others in past but since Firefox 4 every single test I've ran compared to other browsers Firefox 4 is #1 in leaps and bounds. Mozilla has really pulled ahead of the competition with this version. Can't even perceive what is coming next in firefox 5
"Wrong. If I write crappy software that hits your laptop's disc every 10ms, is the laptop the inefficient part?"
Yes, because you could have bought one with a SSD.
"National Security is the chief cause of national insecurity." - Celine's First Law
That's just wrong. Flipping bits increases entropy. If your software is instructing the hardware to flip bits, it is instructing the hardware to consume power.
If you don't know where you are going, you will wind up somewhere else.
Can you run IE on an ARM processor yet? I'd like to see IE9 running on Intel Atom (or whatever the lowest power x86 processor supported is), versus Firefox or Chromium running on Linux/ARM. Then we should see some significant differences.
I like the ampersand, because you're showing someone how to make one with HTML entities, you have to spell it, &amp;
Can you be Even More Awesome?!
The faster the browser, the less energy usage... (setting aside the CPU, Memory, GPU usage it used to make it faster), the idea is, the longer your computer is running, the longer you are using your monitor, and harddrive, and memory, and CPU, etc... the more power you use. Simple! Duh!
Of course Galactic is going to work best with IE9! It's designed by the same org!
In time, most HTML5 performance hiccups of most compatible browsers will be fixed, leaving this site pointless.
What was the "News Site"? How do we know it wasn't specifically designed with IE9 in mind, or had help from the someone close to the IE9 project? It didn't say "unaffiliated News Site", or "Independent News Site". It could just as easily be msnbc.com, which is affiliated with Microsoft, and thus will probably work best with IE9.
One more thing: Why were the browsers jumping around to different spots on each summary graph? That was really annoying to have to follow!
Spork.
P.S. Spork.
Fortunately, using your GPU to render webpages means less load on your CPU, which means your CPU can run at a lower power mode. Also, if you render your webpage in half the time, both your GPU and CPU can go to idle faster. It's not so simple. In fact, if you look at this blog post, the fully GPU-accelerated browsers (Fx 4 and IE 9) are actually using significantly less power...
Warning: Following this link may result in your taking actions that benefit the human race. The range of potential negative consequences include a decline in the frequency of Middle Eastern wars due to a lack of interest, a dramatic reduction in the amount of greenhouse gas emissions, and the defunding of America's Republicans.
A shameless plug that's sorta-kinda on topic.
Orwell: "In a Time of Universal Deceit, telling the Truth is a Revolutionary Act"
Anandtech has done this before. They found that Firefox + AdBlock Plus yielded the best browsing battery life because Flash Ads are a huge power drain. It doesn't matter how good IE or Chrome get, I am still going to use Firefox because of NoScript and AdBlock Plus. Yes, I know that Chrome has AdBlock, but it doesn't really block the ads; just hide them, which completely defeats my goal; using less bandwidth. As sad as it is, not everyone can get unlimited internet.
Source:
http://www.anandtech.com/show/2834/5
Did they tested "power consumption" of IE9 vs FF4 with extension that remove all the crap from webpages? I can imagine that after removing all the talking and moving Flash ads, the browser would require less resources.
Ubuntu is an African word meaning 'I can't configure Debian'
which in a multi-core system would heat the processor up good and toasty without any real performance hit as long as the other 3 cores are idle.
One, two, three, multi, five, six, ...
If my browser consumes a kilowatt of power, but renders a page in a millisecond.... then this would use less energy than one using 2 watts, but takes a second to render. The whole test is just wrong!, they should be calculating the energy usage, not the power usage.
Who on EARTH do you & others like you *THINK* they're fooling?
What on Earth are YOU talking about??? What do you think that I am fooling anyone about? I state that a blog about IE will be (by definition) a partial source. How is that trying to fool anyone? How does that make me a fanboy?
And a fanboy of what? Do you think I am for or against Microsoft? Seriously, I can't think how you could think that the line of mine that you quoted could in any way be labelled as fanboyism. Of all the things that I have posted, this must surely be the LEAST controversial statement ever to come from my keyboard!
I guess mobiles were excluded from this study since IE9 doesn't run on them? Because anyone who cares about low-power browsing would have to swap in an ARM chip for Intel. And in practice they would be running MobileSafari on iPad which can browse the Web over Wi-Fi for 10-12 hours on a single charge with a smaller battery than any of the PC's Microsoft tested.
Basically, excluding mobiles is like doing a portability study and only including desktop PC's, not notebooks.
Even if you had to run an Intel chip and cared about low-power, obviously, you would run a Mac, which has way, way better power efficiency than Windows. The Mac manages the CPU and GPU better, uses the GPU exclusively for drawing the interface, and sleeps and wakes and idles much, much better than Windows.
So once again, we have Microsoft acting like the whole world is just Windows, and therefore not telling the whole story.
Browser benchmarking tools are mostly ridiculously biased - or to be very generous, mysteriously seem to perform far, far better on the browser from the same company that created them. And not just the commercial browser vendors, the little guys are guilty too.
So it should come as a surprise to nobody that Microsoft uses their own benchmarks in this test as well as an unnamed news site (can you say CHERRY PICKING?), and in the end IE happens to come out on par with or better than all the competition, in this test performed by Microsoft, using Microsoft testing tools.
"When information is power, privacy is freedom" - Jah-Wren Ryel
But how did you spell what you just wrote? ;-)
&amp;amp;?
Is 1563649 a prime number?
the electricity, not oil, dumb ass.
It would have the benefit of replacing coal and very little impact on anything in the mid east
The Kruger Dunning explains most post on
Good to know that you're willing to rule out electric/battery, hydrogen, and fuel cell-powered cars as well as electric trains, buses, and other forms of mass transportation.
Of course I should have clued in to your superior knowledge and ability to predict the future from the fact that you chose to use two words to inject your personality into your comment instead of one.
Orwell: "In a Time of Universal Deceit, telling the Truth is a Revolutionary Act"
For "Scenario #3: Power Consumption on News Site", try to watch your CPU usage when you go to cnn.com (or amazon.com or ebay.com). In my humble experience, I get very high CPU usage just after I visit the page and during the page rendering process, especially when it's a java-intensive news site. Looking at the Scenario #3 graphs, I expect to see a high transient power (or energy) peak in the beginning which falls gracefully when the page has finished rendering. I don't expect to see a flat line (plus-minus background task consumption noise) the whole time through. This makes me think that they actually plotted only the power consumed after the page has been fully loaded, meaning that all Scenario #3 graphs actually do not include data after the true time origin (0.0 min).
The above flatline illusion is carefully staged by Scenario #2: about:blank. It is a flat line, but that's normal - about:blank rendering uses negligible CPU or GPU. Therefore Scenario #2 is used to prime your perception with the illusion that there are no transients in such graphs. In Scenario #3, the time origins in graphs might well be the true ones.
In conclusion: scenarios #1, #2 and #3 seem to represent steady state consumption rates after the pages have finished rendering. Scenarios #4 and #5 also feature steady state graphs, but these are more meaningful because they're HTML5 animations.
In addition, when browsing news sites, you have to scroll a lot to read a long article. Scrolling seems to consume a lot of CPU if your gfx card is rather old. In addition, scrolling on GPU-accelerated browsers will reduce CPU usage, but increase GPU processing. I'd really like to see some scrolling power requirements, which IMHO represent the realistic use of a browser (rather than that of men staring goats.)
Finally: What's the use of making graphs of 7 minutes of staring at a blank page, or the same news page in scenario 3? Are browsers expected to suddenly change behavior and begin consuming power after say 5 minutes of staring at a blank page?
If they opted to measure energy and not power consumption (like member tonywestonuk insightfully remarked a few comments above), they'd have to produce cumulative energy charts starting at true zero origins and my bet is that the final picture would be much different...
As the Anandtech article points out, despite the hardware only getting half as much battery life under Windows as it would on a Mac OS, they class this performance under Windows as "decent". As their testing shows various laptops from Asus, Toshiba, MSI, Gateway, fare WORSE running Windows than Apple's hardware.
Just say no to w3schools... http://w3fools.com/
I've moved more than half of my standard browser windows over to Chrome (mostly stuff I don't care if Google snoops on, but not reading news online because Chrome still chokes on that.) Gmail is one of the things I run on Chrome.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks