Browser Wars: Ultimate Browser Benchmark: Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

By | September 7, 2009


With the release of Opera 10 Final, it’s time to find out, who is the king of performance in web browser area.

The following browsers were tested (with no settings changed):
Internet Explorer 8
Firefox 3.5.2
Firefox 3.6 (Alpha)
Chrome 2.0.172.43
Chrome 3.0.195.10 (Beta)
Chrome 4.0.203.2 (Alpha)
Opera 10
Safari 4.0.3

As there is more than 1 benchmark, we will give points after each benchmark for the following results:
3 points for 1st place
2 points for 2nd place
1 point for 3rd place

Let’s begin with classics: SunSpider and Google’s V8 benchmarks*.

Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4
*Internet Explorer 8 crashed.

Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

Peacekeeper benchmark results
Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

CSS rendering benchmark (~2500 positioned (floated) DIVs)
Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

Table rendering benchmark (~21000 td’s)
Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

Celtic Kane JavaScript benchmark
Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

RAM usage after web browser launch
Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

RAM usage with 5 tabs opened
Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

CPU usage during launch
Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

Cold and warm start test
Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

And finally, page load time* (1 point per site)
Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4
*As for Wikipedia, the following page was loaded.

Results

Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

However, if we remove alpha and beta browsers from our benchmark, results are as follow:

Chrome 2 vs. Chrome 3 vs. Chrome 4 vs. Opera 10 vs. Firefox 3.5 vs. Firefox 3.6 vs. Internet Explorer 8 vs. Safari 4

And here you have it. Google’s Chrome web browser is clearly in the lead, followed by Firefox 3.5 and Safari 4. Speaking of Opera, hopefully, by the time Opera 10.10 comes out, things will improve for them.

[digg-reddit-me]


About (Author Profile)


Vygantas is a former web designer whose projects are used by companies such as AMD, NVIDIA and departed Westood Studios. Being passionate about software, Vygantas began his journalism career back in 2007 when he founded FavBrowser.com. Having said that, he is also an adrenaline junkie who enjoys good books, fitness activities and Forex trading.

Comments (64)

Trackback URL | Comments RSS Feed

  1. bogas04 says:

    Really interesting…
    But if you see stability,customisation and frequency of visible bugs then definitely Firefox overpowers every browser

    But then these are features and cant be bench-marked

    Lets wait for FF 3.6 Beta amd 3.5.3 , there must be lot more by FF . Chrome is still there where it was months ago,,,,,,,,,,fastest but featureless and less fruitful , and i dont care about others :P Since they either are slow or dont support sites and other general features (like common hotkeys etc)

  2. whatever these tests say, i find Chrome and Opera serve me best, firefox with essential add ons often hangs on my machine ! i am using Opera 10 now, it’s awesome !

  3. Norbert says:

    I have some doubts about the Firefox benchmarks. Practicaly every user install a bunch of addons without which firefox is a pretty ordinary browser. But the benchmarks test only the addon free firefox right? I find it’s a bit unfair towards browsers such as Opera which are INTERNET SUITES – meaning come packed with features (torrents, IRC, mail client etc.) which neither FF nor chrome do not have.
    I would really look forward to a benchmark that compares a FF and Opera with the first having such addons installed that it has same or similar features as Opera.

  4. Dan says:

    I see a ton of artificial JS tests. (V8, Sunspider are webkit friendly therefore it should not be part of this test, Peacekeeper was not designed for certain browsers though) Since when was a memory usage bad? That is what makes the browser faster. Also O10.1 should have been tested. More CSS is used in webpages than JS, therefore this is not a clear depiction of a browsers speed. How about a real world test?

    • It depends on sites you visit.

      I don’t use any web mail, but as far as I know they are all JS based (gmail, yahoo, etc)

      As of memory, that can be also true.

      • Daniel Hendrycks says:

        Top 4 sites according to Alexa. Stats thanks to http://www.websiteoptimization.com/services/analyze/

        Google-
        (Bytes)
        HTML: 3858
        HTML Images: 11084
        CSS Images: 0
        Total Images: 11084
        Javascript: 0

        Yahoo-Unable to fetch stats

        Facebook- (bytes)
        HTML: 9943
        HTML Images: 589
        CSS Images: 115998
        Total Images: 116587
        Javascript: 11544
        CSS: 17911

        Youtube- (bytes)
        HTML: 17205 3.63
        HTML Images: 33507
        CSS Images: 0
        Total Images: 33507
        Javascript: 33055

        As you can see Javascript is the the most used code type in the major sites. But sites like gmail use alot of js, so a fast js engine would be handy for those sites.

        Bye :)

        • Daniel Hendrycks says:

          Edit (under youtube omit “3.63”)

        • Foo says:

          With Firebug I get:

          google.com
          HTML + Embedded CSS + Embedded JavaScript: 3 160bytes
          JavaScript: 6 316bytes
          Images: 13 386bytes
          Images (Logo): 7 582bytes
          Images (Sprite): 5 804bytes

          yahoo.com
          HTML + Embedded CSS + Embedded JavaScript: 34 664bytes
          Images (26): Sorry, I’m not going to count those. But about 6kB each so 159 744bytes

          facebook.com
          HTML + Embedded JavaScript: 9 952bytes
          Javascript: 11538+54865+20794+3925+486 = 91 608bytes
          CSS: 3967+3995+6298 = 14 260bytes
          Image (Requested with ‘AJAX’ on load): 129bytes
          Images: 67+522+398+2704+1857+8938+42569+129 = 57 184bytes

          youtube.com
          HTML + Embedded JavaScript: 16 029bytes
          CSS: 15 776bytes
          JavaScript: 31 065bytes
          Images: 13185+1793+3687+3681+3600+3687+9716+3610+3693+3622+3568+1831+3595+3663+41 = 62 972bytes

          favbrowser.com
          HTML + Embedded CSS + Embedded JavaScript: 28 314bytes
          CSS: 2850+2571+15009 = 20 430bytes
          JavaScript: 57276+496+3251 = 61 023bytes
          Images: 2550+12112+1044+44928+7101+1225+574+1007+20365+9055+4741+1653+2702 = 109 057bytes
          Flash: 1050bytes

          Small interesting fact, only Google (includes YouTube) seems to send the content-length header for HTML.

          As you may have noticed, most of those sites use embedded JavaScipt and CSS so it’s hard to count their sizes. Also, some CSS is duplicated by first applying a base template and after that making changes yet again. On that you could possibly add data sent and handled by JavaScript in cases such as auto-completion. Additionally some sites might block parts of content depending on the supplied user-agent etc.
          Data size is a crappy way to measure importance either way. For instance, most of the examples on Chrome Experiments use less than 20kB of JavaScript, mostly being around 15kB from the ones I took a look at. Or ‘in other words’ it doesn’t take the same time or power to compute 1+1=? as it takes to compute 5!!=? even though their length is the same.

  5. Nick says:

    Ok. Let’s wait Carakan in Opera, but anyway Vega is more important.

  6. Chas4 says:

    What are the specs of the test computer used?

    • Windows 7
      Core 2 Duo E6750
      4 GM of RAM

      However, it was tested on a virtual PC (fresh installations, etc.)

      • Chas4 says:

        I wonder if anyone has done the test on an older system that most people might have

        256 mb of ram (maybe 512mb)

        A 1 ghz single core cpu

        I don’t think many people have a newer pc most I would guess would stick with the same one for years

  7. John says:

    Safari 4 running on Snow leopard kicks chromes butt =p

  8. Hellspork says:

    This is not a test of anything. Memory use is a subjective and pointless waste of typing.

    http://my.opera.com/hellspork/albums/showpic.dml?album=875612&picture=11984306

    How many other browsers allow you to do something this stupid? Racing through the internet in Windows 98, with only 32MB of RAM? If those benchmarks were right, why does Chrome feel so sluggish and jerky with WinXP and two gigs of RAM?

  9. nobody says:

    I REALLY REALLY wonder in what test Opera would win..

    somebody here mentioned ‘real life test’ – please, post a link if you can :)

    as for the slow CPU’s, I have such a laptop – p4 s468, 2ghz (VERY slow), 512mb and what? chrome is the fastest there to start, to close and to use..

    • johnnysaucepn says:

      Benchmarks are artificial by definition. They test one thing, under one environment. They are by no means any use for producing any kind of conclusions about any browser.

      People go for easy numbers, like memory usage – without considering caching, default settings, etc. This is lazy, and useless for comparing anything. There are methodologies and statistical techniques to follow here.

      You can test how fast javascript interpreter is in a given set of tests, but that is ALL you’re testing. Drawing a conclusion from this is pointless unless running test suites is all anyone is going to do in their browser.

      • Eice says:

        Oh, really? And what’s so artificial about benchmarks? Will browsers suddenly perform any better or worse when it’s not a tester who’s using them?

        You need to come up with a better, more scientific refutation instead of simply saying that tests are invalid. If you do have a legitimate argument, then let’s hear it. Otherwise it’s just painfully obvious what you’re trying to do, Opera shill.

      • V-2 says:

        Well, even the above tests show clear correlation between benchmark results and the probable “user experience”.

        For example, look how long it takes for Internet Explorer to load a MySpace website – which heavily relies on JavaScript, and therefore IE’s slow engine has much of an impact here.
        Over 3 times slower than new Chrome…

        Average users may not care about “artificial” benchmarks, but they DO visit portals such like MySpace or Facebook pretty often.

    • Lettlurt says:

      Benchmarking Browsers with Real Websites: Chrome, Firefox, Opera, Safari, IE
      http://www.codexon.com/posts/a-real-benchmark-real-websites-with-chrome-firefox-opera-safari-ie

      • Foo says:

        More like “benchmarking static HTML with failing scripts”. Mr. ‘Easily Fooled’, you truly are the fool.

        • Thoe says:

          Ah, so if anyone else wins, the test is 100% correct, but if Opera wins, the test is automatically invalid?

          Wow, you ignorant fanboy trolls sure are amazingly hypocritical…

          • Foo says:

            Just pointing out a few of the faults with his “test”. But yea, tests based on something concrete, like execution time of a specific snippet of code, are much more useful than some random ‘I believe this to be faster metric’ like the link you are spamming. Heck, even Peacemaker is more useful.
            He should at least clean up any external data, get a proper host for his files and learn what a website is.

  10. Pseudoscion says:

    I use Opera as my main browser, although am forced to use Firefox with IETab to access certain facilities provided by my company. I try Chrome every so often to see how it is comming along, and cannot use IE8 due to it crashing, and very occasionally I’ll try Safari, although it sucks on XP.

    Theses benchmarks are all very well but as has been said repeatedly, until you use a browser, in fact multiple different browsers, for extended periods of time then you don’t really get a feel for the speed and responsiveness of them.

    My 2 cents is that Chrome is overall the fastest (although there are some sites where it crawls), but Opera still holds it own by being very responsive, and you hardly notice the difference in speed between Chrome, Firefox and Opera.

    I feel it could also depend a lot on the PC being run on, cpu, ram etc.

  11. photosynthesis says:

    Interesting, opera falls far behind in largely all these tests I have seen.
    Looking forward to Opera making their own javascript benchmark :D
    btw, i’m using opera

  12. Wallpaper says:

    Nice article. I myself using opera 10 with firefox 3.5 for addons
    Opera is by far the most ease of use browser

  13. V-2 says:

    Well, the improvements of Firefox 3.6 were supposed to be focused on speed, esp. start-up time. It doesn’t look promising if it’s actually slower than the predecessor (even being an alpha version and all).

    I also second to the observation that add-ons slow Firefox down a great deal.
    On my PC – an old one to be sure, but not quite taken out of the museum of technics (CPU 2.8 GHz / 2 GB RAM / XP SP3), with all add-ons – I think I use about 10, and only 2 or 3 could be called “heavy” – it needs 30 seconds to start up. I mean – come on!! That’s ridiculous.

    That’s why I use SRWare Iron (Chrome derivative – read about it!) as my default browser, and Firefox as the “secondary one” (for “real”, lengthy work).

    • Andrew says:

      Yes, when I read the article I found it strange too that Fx3.6 is worse at startup than 3.5. Maybe it was some buggy nightly build.

  14. Jarel says:

    These tests might work for some, but not all and they surely aren’t definitive. Safari 4 is by far the fastest browser on Snow Leopard (for obvious reasons). Also, it depends on your view of what’s good and bad (like memory usage).

    I think what’s important and definitive here is that Internet Explorer performs like a dog that’s been hit by an 18-wheeler. :-P

  15. Euch says:

    there’s something I don’t get about the sunspider benchmark
    even though it clearly says chrome is the fastest and then goes safari, ff, opera and IE I was a bit skeptic, so I did the following test:
    I’ve installed all 5 browsers on my computer, opened them simultaneously and started the test on all of them together.
    even though the numbers gave away was in that order – the fact was that the first one to finish the test was opera, after it came ff and chrome nearly together, the safari (which was behind IE for most of the test but managed to get past it towards the end) and ie8 finished last.

    that made me believe someone is lying about those tests… have no idea who or why, but I wouldn’t trust these tests – I’de rather compare browsers side by side and see for myself which one I’de rather work with – and I suggest you do the same, regardless on what you think is the best browser

    • Andrew says:

      Running tests in 5 browsers simultaneously? You are insane!

    • Foo says:

      You do realize that your computer works with limited resources, right? The only thing your ‘test’ showed is that your OS put higher priority for the Opera process during that time period, most likely you had that browser focused while waiting.

  16. Ice Ugwa says:

    This stats didn’t say any thing on flock 2.5.2 browser. I use it and i enjoy it alot. I use the social networking features alot. I as well use chrome. Its ultra fast. Recently download opera but can find anything interesting about it. IE is the worst browser i have ever seen. It sucks badly. I recommend flock browser to you all (can be downloaded at http://www.flock.com) its very cute and good

  17. Aerialsky says:

    Where’s Acid3 Browser Test? http://acid3.acidtests.org/

  18. vlad says:

    I appreciate your search of information and sharing your results with others. Year from now browser load can be different but the attitude to get the best from the present moment is rewarding.

  19. Google chrome isn’t doing to badly, I might have to try it out some more instead of firefox.

  20. machan says:

    As of now, there is nobody who can beat Safari 4. As the caption says, the fastest browser.And it does not get slower when time passes by.

    Firefox is crashing constantly this week. Who knows whats the problem. And it crashes completely.

    IE 8 is far better browser than any IEs before. Opera 10 holds the same rank.

  21. Paul says:

    Google chrome is and will be the best browser for me. It’s very fast and now it has extensions too. It’s the best browser.

  22. mbah bejo says:

    nice post, may be chrome is great