The Great Browser Performance Showdown
One of the most popular posts on this blog is about a comparison between Firefox and Chrome. That was when I switched over to Chrome from FireFox, and have never looked back. This month Opera 10 was finally released, and I have also since installed FireFox 3.5, Safari 4, and IE 8. So, I thought I would do a Browser showdown to see how these latest and greatest browsers perform against each other.
Firefox versus Chrome versus IE versus Safari versus Opera.
And so I present, the Great Browser Performance Showdown:
Setup and Preparation
A little note about the setup I had in place and the preparations I did before testing out the various aspects of these browsers.
- Windows Vista x64, 4 GB RAM, 10K rpm HDD, Core 2 Quad Q9550.
- Fresh Windows install.
- Fresh Chrome, Opera, Safari, IE, and Firefox installs.
- No plugins or addons on any browser.
- Set each browser to load a blank page upon startup.
- Warmed up each browser to visit several sites just to fill up cache and cookies randomly for each browser.
- Updated each browser to the latest.
- Each browser has a Quick Launch icon in my Quick Launch bar.
And for the record, here are the versions that each browser was running.
- Opera 10.0, build 1750
- Firefox 3.5.2
- Internet Explorer 8.0.6001
- Chrome 2.0.172
- Safari 4.0.3
Before I start, I have to say that this comparison is not very scientific, but is 100% unbiased. Also, I am not comparing features that each browser has, but comparing only performance while performing normal tasks. Features are a whole different ballgame.
Here we go.
Startup Time
One of the things that I want my browser to be able to do the most is to start off very quickly – it should launch before I am done moving the mouse after clicking the icon to launch it. So, how do all these browsers perform while starting up.
The method I used to test this simply classifies these browsers into two parts: Blazing Fast and Just Not There. This is more a feeling test – if I feel that the browser is taking a pause before launching, its Just Not There. And the results are:
Blazing Fast
- Chrome
- Internet Explorer
Just Not There
- Safari
- Opera
- Firefox
Even I was surprised by the results. Of all of them, Firefox is the slowest with Opera being the second slowest. I have to of course say, that the difference is really a split second. With Chrome and IE, you can say that they start instantaneously, while Opera and Safari, I felt a momentary lag before they start up, and with Firefox, it was quite noticeable.
Another way to confirm how fast each browser loads is to check for how long does the busy cursor in Windows Vista show? With Chrome and IE, I hardly see the cursor, while with the others I actually see it go round.
Another thing to notice is that the startup time increases over time. In case of Firefox, once you install some plugins, the startup time really takes a plunge (from past experience) and the same goes with IE. Even with Chrome, it won’t be so blazing fast after you have used it a bit.
Conclusions:
- Chrome and IE win.
Memory Usage
The other thing that is important is how much memory does the browser use. If you are anything like me, you will have 30-40 pages open at the same time, and before you know it your browser alone is taking up more than a gigabyte of your RAM.
The other thing to know is how quickly does the browser release memory as pages are closed and tabs are shut down. For this test, I am going to start the browser, and open 6 pages, and note the total memory consumption for all of the browser processes after each page is opened. Then I am going to close them one by one and see how the memory is released. In order to measure how the memory is released, I am going to take the measurement 30 seconds after closing the tab. I am going to use the same 6 pages for each browser.
Here are the results:
In the chart above, the x-axis is the number of pages open. The following observations can be drawn with this data:
- IE and Chrome take the most memory as you open more pages (IE is the highest).
- IE and Chrome release memory much better as the pages are closed.
- Firefox, Opera, and Safari don’t release memory as much as pages are being closed. This is obvious to see when you compare the memory that was being used when the browser was started initially to when the pages were closed to zero (Firefox 20 to 53, Opera 24 to 69, and Safari 22 to 79)
- Opera uses the least amount of memory of all the browsers per page.
So, as you go about your day, opening and closing pages, there is a big chance that you will have a sizable memory consumption when using Opera, Safari, or Firefox. While if you are using IE or Chrome, and you need to free up some memory, you can simply close some pages instead of the entire browser.
Some more observations that I made during this testing:
- Chrome and IE both use multiple processes for loading multiple pages. This is probably what causes the higher memory consumption, and also allows the faster memory release.
- While Chrome uses one process per page, IE doesn’t always spawn a new process per page. Chrome releases memory almost immediately, while IE takes up to a minute to release memory.
Conclusions:
- Chrome and IE win because of how quickly they release memory.
- If you are not a heavy user, then Opera wins.
Browsing Speed
This is the one that most browsers keep talking about. How fast does it actually load the page. And there are plenty of sites out there which have very scientific measures of browser performance. However, in my daily use, I want to see how fast these browsers load the sites that I visit. For this test, I have cleared my caches in all the browsers, and I am going to test them against 5 pages each (the same pages). Here are steps I am going to use to measure the time:
- Clear cache
- Start time on pressing Go
- Stop time when page stops loading
- Repeat 3 times and take average
Here are the results:
In the chart above, the x-axis represents the 5 different websites (LifeHacker, CrunchGear, YouTube, MidEastYouth, and StackOverflow) and the average time for all the sites.
The following are the obvious observations:
- Opera is painfully slow.
- And Firefox is blazingly fast.
- Chrome and IE do pretty well in most situations, with Chrome being faster more often.
- Safari is just about average.
One thing I would like to point out, is this. While all other browsers don’t change anything in the open window when you clear browsing data, Opera clears out the entire environment (closes the open page).
Conclusions:
- Firefox is the winner.
- IE on average is second best, but Chrome is probably a better contender. The average suggests that Chrome does very badly on certain types of sites.
Verdict?
Here’s the summary of my testing:
- Firefox wins – startup time and memory usage not withstanding, it is the one that gives you the fastest browsing.
- Internet Explorer 8 is a very compelling contender.
- Chrome is still holding its own. I am probably going to stick with Chrome for now.
You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.





September 4th, 2009 at 3:36 am
I use Chrome regularly now. Along with Firefox. Chrome should be having some plugins soon. I’d like to stop the ads and the automatic playing of videos.
September 4th, 2009 at 9:46 am
Well yes, the plugins are still what is lacking in Chrome. But I was very pleasantly surprised at how quickly Firefox loads websites.
September 6th, 2009 at 4:31 am
I’ve read just about everything on your site and still don’t have the answer to my original question . . . . which of all of the browsers is the most secure? If someone were crazy enough not to have any anti-virus + software installed on their machine, which browser would be best to use. Yeah, I know, none, but I’d really like to know, if you had absolutely NO protection, which browser would be safest?
September 6th, 2009 at 4:53 am
That’s an interesting question…. and I have to be honest – I don’t have an answer ready…
I use OpenDNS and so I don’t need to depend on a browser all the time for security…
Having said that, I think the fact that Chrome doesn’t allow plugins, might probably make it the most secure – but I don’t know.
September 8th, 2009 at 2:46 pm
To tell you the truth, no browser is really secure. The security has to start with yourself.
Do not donwload unintended stuff. Don’t put your personal information in some untrusted website.
Do your due diligence and you will live virus free.
September 8th, 2009 at 3:57 pm
Well, true, and I agree. But the sad truth is that the majority of people don’t know any better, and that is why Browsers and Operating Systems are better if they can stop those who don’t know any better from hurting themselves.
September 8th, 2009 at 1:09 pm
This is a nice post.
I agree Firefox starts slowly, but it runs faster. I can’t use any other browser than Firefox because it has plugins that take care of my social networking and several other things I can’t do without.
September 8th, 2009 at 1:53 pm
Yup, you sure are right on that count…
September 8th, 2009 at 3:55 pm
I believe for browsing speed it depends on the javascript on the websites you surf.
Opera is painfully slow for loading javascript, but definitely not for normal page rendering such as images loading or even text loading.
September 8th, 2009 at 3:58 pm
Well, yes Javascript is a big factor of course. But almost all modern sites have some javascript…
September 8th, 2009 at 4:32 pm
i think you say chrome is starts faster time beacuse firefox running with few plugin.If you remove the plugin i think firefox will be winner
September 8th, 2009 at 4:48 pm
Zeeshan, as I said in the review. For this testing, all 5 browsers were installed fresh, with all their cache, plugins, history, etc. cleared in the beginning.
So, for Firefox, there were no plugins installed.
September 8th, 2009 at 5:32 pm
IE(even IE8) is a real pain in the ass for the web developer like me. It doesn’t pass Acid 3 test yet (http://acid3.acidtests.org/) while opera and safari scores 100/100, FF 3.5 scores 91/100 and chrome 100/100 (with linktest fail) but IE8 goes horribly wrong for me scoring as low as 14/100 with whole page getting disturbed. I know what you did was the performance test, but with such a poor quality and lack of standard(still) , I don’t give it a damn and will advise all to dump that browser coz it still SUCKS!
September 8th, 2009 at 7:41 pm
Well, of course – from a developer standpoint, the picture is probably very different. This post is purely from an average user standpoint
September 8th, 2009 at 5:38 pm
Good post by the way
September 8th, 2009 at 7:41 pm
Thanks, heh heh
September 8th, 2009 at 9:26 pm
OK, a couple of questions on your review.
(1) Startup Time
I don’t know about you, but I feel startup time would be utterly irrelevant for the vast majority of users. I open my browser when I turn my PC on, I close it when I turn it off. It is by far the least commonly carried out action in anyone’s browsing time, why do you give it such prominence here?
(2) Memory Usage
You say: “how quickly does the browser release memory as pages are closed and tabs are shut down” – I wonder though, what programs are they “releasing memory” to? If you have no “competing” high memory apps running concurrently on your system, an efficient browser will HOLD ON to memory unless it is explicitly requested by another program would it not? Did you run anything concurrently for this test?
(3) Browsing Speed
This one I really don’t get at all. You say here: “For this test, I have cleared my caches in all the browsers”. WHAT? Cache is one of the primary mechanisms for browsers to SPEED UP webpage loading times. Surely clearing that out renders this entire section of the test utterly irrelevant does it not? How many users clear out their cache before each page they visit?
Above, at the beginning of the article in Setup and Preparation you’ve said: “Warmed up each browser to visit several sites just to fill up cache and cookies randomly for each browser” – what was the point of this step if you go on to clear it out again?
September 8th, 2009 at 9:47 pm
Nice questions. And here are the answers, let me start with the last ones first:
3. The warm up was for the first test because as browsers build browsing histories, form filling data, auto complete, quick dial, etc, they slow down in their startup time. So, to ensure that all browsers have relatively the same usage time behind them. This will give it a more fair platform for startup.
And clearing out the cache for the last test was necessary to measure the speed for loading the pages from ground up. A lot of people do visit a lot of new sites every day, so it does matter how fast browsers are at loading new sites.
2. Memory Usage – you have a valid point here, however, I do have some past experience using FF, where I used to have it using a large amount of memory, and if I now have to run something like Visual Studio, it wouldn’t let go of memory even when I used to close almost all tabs. The only option was to close the browser. So how quickly memory is released is important IMO.
You need to remember that not every one is running awesome hardware with lots of RAM and very fast disks (not counting me). For example, this is not a problem on my desktop, but on one of my laptops with only 1GB of RAM, this becomes an issue very quickly.
1. Startup time – again a lot of people do close their browsers often enough – and I am not the only one. Or start new instances. One of the primary reasons why I shifted from FF to Chrome a year ago was because of how quickly it loaded whenever I needed to run it.
In case of FF, I used to have multiple instances of it open all the time because if one tab misbehaved it would crash the entire browser. So, my way of operating it was that I would have my key tasks open in one FF and I would keep opening and closing new instances for other work throughout the day, so the app loading speed mattered quite a bit.
Finally, at the end of the day, I would also like to fall back on my disclaimer – this is not a scientific experiment, and the relevance of results is purely subjective to every individual.
Cheers.
September 8th, 2009 at 10:33 pm
Great response (and thanks for the personal email too).
1) Startup – I can definitely empathise with the issue of Firefox crashing and requiring a restart, however stability apart I don’t think restarting a browser multiple times (or requiring dodgy multiple instance hacks) should be considered “normal”. A stable browser will require none of this, and should ideally be started once, and closed once. If your browser is crashing all the time, that’s a bigger issue than performance, and perhaps you should switch browser on the back of stability not performance.
2) Again, I can empathise with the case of opening up Visual Studio (or whatever high-memory app) and Firefox not freeing it TO the newly opened app. However unless you actually try running apps concurrently DURING TESTING I don’t see how the results can be of relevance. It’s possible that the browser might free up memory to newly opened apps – you don’t know until you try it.
3) People do visit new sites daily, but the vast vast majority of what they visit is not new – people return to read new blogposts on blogs they read regularly, they post in forums they visit regularly, they revisit webmail and online services of which they are a member, etc.
So even if this statistic is somewhat relevant, I don’t think it’s very relevant. Testing how fast cached pages load would I believe be a far MORE relevant test.
September 8th, 2009 at 11:13 pm
Once again, all relevant points, and I agree. This is by no means a very comprehensive test, and I do agree that I probably need to add two more:
1. Test memory release when other apps are contending for them.
2. Loading frequently visited websites.
Will probably do them soon
September 8th, 2009 at 11:25 pm
Great to hear. I look forward to seeing them. I was quite hopeful coming to this blogpost as most such tests seem to be related to artificial and even more irrelevant javascript benchmarks – this seemed at least to be attempting a realistic, real-world comparison.
September 9th, 2009 at 3:27 pm
my vote is for firefox
October 1st, 2009 at 11:53 am
Cool, just added some of your results to WikiVS in the performance section (http://www.wikivs.com/wiki/Chrome_vs_Firefox#Performance).
Keep up the research and may the best browser win!