fast firefox

Now that Firefox 3 is approaching the home stretch it is important that Mozilla starts to throw in performance improvements to really make the browser purr. Over in our forum xpgeek pointed out that a Profile-Guided Optimization (PGO) build of Firefox 3 had been created which greatly improved the performance of JavaScript in the browser. While PGO itself hasn’t yet landed in the nightly builds there have been some significant improvements to the JavaScript engine.

I’m sure what most of you care the most about are the facts, and so I’ve compiled the results of the SunSpider JavaScript Benchmark test for each of the different browsers. All of the tests below were performed on the same Windows machine, and the Firefox 3 nightly builds definitely came out on top. Here are the results sorted from best to worst (each one is hyperlinked to the full stats):

  1. Firefox 3 Nightly (PGO Optimized): 7263.8ms
  2. Firefox 3 Nightly (02/25/2008 build): 8219.4ms
  3. Opera 9.5.9807 Beta: 10824.0ms
  4. Firefox 3 Beta 3: 16080.6ms
  5. Safari 3.0.4 Beta: 18012.6ms
  6. Firefox 2.0.0.12: 29376.4ms
  7. Internet Explorer 7: 72375.0ms

It’s important to know that every time you run the SunSpider Benchmark it conducts each test five times, and the result is the average of the five tests. So it is a rather thorough test, and definitely shows off the speed improvements that Firefox 3 is going to be bringing to the table.

What does this all mean for you? Depending on what browser you typically use you may not notice a huge speed difference, but the change will be the most noticeable on sites that use JavaScript heavily. With the Web 2.0 era upon us all JavaScript speed enhancements are welcomed with open arms.

Firefox 3 Beta 4 is expected to be released in the next few weeks, and you can expect to see these (and many more) improvements shining through!

There Are 81 Comments

  1. With all the stuff there adding you’d think it would be slower.

  2. Is there any difference between Opera 9.26 and 9.5.x? I knew using the nightlies the the javascript felt a lot more responsive, but boy this test really shows how much it shines above 2.x. It’s results like this that make me want to use FF3.x as my default browser. Once Foxmarks works with 3.x it’s bye bye 2.x. Everything else I care about already works.

  3. Good news all around, think I’ll still wait for the RC1 to come out before trying to dedicate myself to using it 24/7. :)

  4. leland wrote:
    Is there any difference between Opera 9.26 and 9.5.x?

    Yep, Opera 9.5 has a rewritten JavaScript engine, so it should be much faster in the tests. Besides, according to Jeff Atwood, who made some tests in his blog back in November, Opera 9.2x fails some tests, don’t ask why.

  5. FF3…well lets just say I really can’t use any other browser out there. :twisted:

  6. IE7 sure is incredibly slow. I wonder if they will join this speed race with IE8.

  7. Google wrote:
    With all the stuff there adding you’d think it would be slower.

    That’s true, but this is not the overall performance of the browser. This test only analyzes the JavaScript engine, but I would actually say that the overall performance of the browser has significantly improved as well.

    leland wrote:
    Is there any difference between Opera 9.26 and 9.5.x?

    I tested Opera 9.26 and there was an error that caused it not to display the results of each test. That’s the reason that I didn’t include it in the list, but it looked generally the same as 9.5.

    Change wrote:
    IE7 sure is incredibly slow. I wonder if they will join this speed race with IE8.

    I’m hoping that Microsoft will get with the program because competition in the performance arena is the only way that other contenders will push to be better.

  8. Just one issue about the tags: “Freeware”. Well, it is actually Open Source not Freeware.

  9. nix wrote:
    Just one issue about the tags: “Freeware”. Well, it is actually Open Source not Freeware.

    We do realize that, but for the sake of simplicity we consider open source software to also be freeware since it technically is. Freeware encompasses all software that is delivered without charge, and you don’t have to pay for Firefox.

  10. Can some one run it against this:
    [glimr.rubyforge.org]

    WebKit is awesome on this one. I tested this with cairo-openvg+shivavg and some of them give ~200fps where Firefox gives me ~30fps.
    I hope they get cairo rendering right!

  11. I don’t know what was going on with your firefox 2.0.0.12 test…but i did the test with the same version and get significantly faster numbers.
    Total: 13387.4ms
    [webkit.org]

  12. It should be noted that both IE6 and IE7 exhibit pathologically bad performance on one of the SunSpider tests in particular. I forget which one exactly, but on that single test their performance is 10-15X worse than all the other browsers. If that one test is removed, IE still doesn’t come out on top, but it’s at least in the same general range as the others.

  13. I just tested 2.0.0.12 again and got a faster yet result.
    Total: 12394.6ms
    [webkit.org]

  14. Leif wrote:
    I don’t know what was going on with your firefox 2.0.0.12 test…but i did the test with the same version and get significantly faster numbers.
    Total: 13387.4ms
    [webkit.org]

    Huh, I don’t know what to tell you. I did all of the tests on the same machine though so the results should be accurate relative to each other.

  15. You should thank the FreeBSD devs, they made it possible :D

    [ventnorsblog.blogspot.com]

  16. I would like more detail on your results because I get:

    IE 7 38412.0
    Opera 9.5 17497.2 but it appears to get stuck on ‘crypto-aes’ and I have to click the mouse to make it continue.

    In a real world test, Google Maps barely moves under Opera 9.5 and IE has no problems and it is quick.

  17. You are probably running the test on a significantly faster computer. The numbers are not the important part, the numbers will vary between computers, the percentage differences are the important parts.

  18. WinXP/Sp2-C2D@3GHz/2GB RAM

    Safari 3.1 Beta (525.7): 2575.2ms
    Firefox 2.0.0.12: 13087.6ms
    Internet Explorer 7: 20844.4ms

  19. On my 10-month-old MacBook Pro, running the latest Safari (WebKit) nightly, I get 3294.2ms…is my machine *that much faster* than the machine used in the article, or is WebKit just that far ahead of FF?
    [webkit.org]

  20. Surprised wrote:
    I would like more detail on your results because I get:

    I don’t know what other information I can give you. The full results are already linked to for each test.

    Steve wrote:
    The numbers are not the important part, the numbers will vary between computers, the percentage differences are the important parts.

    Exactly.

  21. Webkit nightly is way way way way way way way way way way way way way way way way way way way way way way way way faster than FireFox can ever be. XPCOM hello? Webkit doesn’t use that heavyweight abstraction, it gets speed through shear simplicity. The webkit guys are mostly bored apple developers who don’t want to actually do their work so they spend all day going aspie on the speed tests, the #include src files into singular units to take advantage of optimization they pad out structs to match base2 boundaries. You cannot claim you are fast without comparing yourself to Webkit nightly.

    Sure you might eventually beat them because they are too scared to do anything about their VM other than to tweak it and complain about the KHTML patches.

  22. I thought the guy above was a troll but a friend redid the bench mark with his browsers:

    * Nightly Webkit r30627 – 3205
    * Firefox Nightly3.0beta4pre – 8930

  23. Michael Brian Bentley

    “On my 10-month-old MacBook Pro, running the latest Safari (WebKit) nightly, I get 3294.2ms…is my machine *that much faster* than the machine used in the article, or is WebKit just that far ahead of FF?”

    I think that their benchmark numbers are generated on Windows, and that the Safari port to Windows is slower than the Mac OS X version of Safari.

  24. WinXP/Sp2-C2D@3GHz/2GB RAM

    Safari 3.1 Beta (525.7): 2575.2ms
    Firefox 3 Nightly (02/27/2008 build): 2719.4ms
    Firefox 2.0.0.12: 13087.6ms
    Internet Explorer 7: 20844.4ms

  25. I can verify that my WebKit nightly build (unmodified) is edging out FF3 beta, and that IE7 is really, really slow.

    The test on which IE7 really falls behind is called “string-base64″. IE7 does this test about 40 times slower than my FF beta from two days ago.

    On the upside, the beta is extremely stable, and has been running for about two days straight, heavy browsing, and has about a constant 10 or so tabs open. I haven’t seen any major delays or hangs, only minor pauses here and there.

  26. It’s really silly to fight over which browser is the fastest, when it comes to a difference in milliseconds usually.
    The important fact is that they’re all (Except the IE team, perhaps) working on improving their javascript execution speed, and that’s good news for most of us.

    With that said, I, personally, am waiting anxiously for the performance improvements of Firefox 3.

  27. It would be great if you actually used a benchmark test that:
    1) Was standards compliant.
    2) Actually checked to make sure the results were correct.
    3) Didn’t spew out skewed results because the lack of #1 caused #2 to fail, which causes some browsers to report insanely bad results (IE 7).
    4) Wasn’t based off code written by one or more of the browser developers.

  28. In this case the difference isnt milli8seconds, but for example, FF3 is TEN TIMES as fast as IE 7. Thats hardly a small difference!

    And if you surf a lot, even small improvements add up to a lot of time. Surfing a whole day, IE 7 users would be waiting ten times as long as FF3 users – this may be a lot of work time!

    But then again, I dont consider IE 7 a browser. Its not standards compliant etc.. But the major players are going into a lot of effort it seems.
    Back in FF2 times I just had to use Opera, Firefox experience on Linux was so bad. Now Ive been using Swiftfox 3 betas for a few months already, since its been so good even in beta (Swiftfox is a optimized build for Linuxes). Its stabler, hugely faster (on my AMD x2 pages render in an instant, I can barely bat my eyelashes and its rendered), and its got great features I love – like the new address bar / history thing. Im loving it :) good job guys and gals!

  29. Opera 9.26: [webkit.org]
    Firefox 3.0b3:
    [webkit.org]
    Firefox 2.0.0.12:
    [webkit.org]
    Epiphany 2.20:
    [webkit.org]

  30. :arrow: Including Tamarin (or Tamarin-Tracing) in Mozilla (and Firefox) should speed it up even more. Tamarin is the VM for ActionScript (Adobe’s implementation of JavaScript 2 aka ECMAScript Edition 4) that is used in Flash Player. The integration project of Tamarin into Mozilla is called ScreamingMonkey (while SpiderMonkey is what Mozilla currently uses).

    [mozilla.org]
    [wiki.mozilla.org]

  31. LM wrote:
    WinXP/Sp2-C2D@3GHz/2GB RAM

    Safari 3.1 Beta (525.7): 2575.2ms
    Firefox 3 Nightly (02/27/2008 build): 2719.4ms
    Firefox 2.0.0.12: 13087.6ms
    Internet Explorer 7: 20844.4ms

    Thanks for performing the test with the Safari nightly versus Firefox 3 on the same machine. While Safari might still be faster the difference is pretty small, and from what I’ve heard from Mozilla they haven’t finished optimizing the JavaScript. So by the time that Firefox 3 is released we might just find out that it is faster.

  32. [webkit.org]

    Opera 9.5 beta (build 3807)

  33. so? i hate javescript! i always uninstall it and never visit sites that use it…

  34. fd, then Lynx is your browser! You can do away with the mouse and fancy high resolution GUI too in favour of a lightning fast console. :-P

  35. Impressive improvements on the latest build of Firefox 3.
    Hope Opera guys put its hands on the JS engine to improve its performance.

  36. I do not care too much about this actually. At least the pages I browse do not need JS too much and definitely as much as that benchmark is going with its testing. Rendering speed is much more important.

    And what really pisses me off about FF under Linux is that the multithreading is still sucking that much. It still occurs rather frequently that some action in one window locks up all other windows for some time or forever. That a plugin problem in one window can lock up and crash the whole browser with all windows is also a nuisance.

  37. fd wrote:
    so? i hate javescript! i always uninstall it and never visit sites that use it…

    Well, you visited our site and we use it. If you came from Slashdot then they used it. It would be pretty hard to specifically not visit sites that don’t use JavaScript.

  38. Taking a closer look at the benchmarks being run, these numbers don’t seem to have much impact on typical Web 2.0 websites, because SunSpider does not run any DOM tests (anyone knows why?).

    So I wouldn’t expect too much of a performance gain on a typical AJAX heavy website.

  39. Ok speed is great. But these alternate browsers will NEVER gain more then a few % marketshare (nd from home users only) if they do not add:
    - Central installation via Active Directory
    - Central patching/reporting via WSUS
    - Central control via Group Policies

    It’s the companies where the big marketshare is. Internet Explorer 6 and higer support these critical centralised and intergrated administration features. It’s a Windows world, deal with it. No admin can roll-out Firefox or Opera without these features..

  40. Descargar en ESPAÑOL para Windows

    [ftp.mozilla.org]

    Download

    Other Languages: [ftp.mozilla.org]

  41. Download Firefox

    But will FF3 use less RAM? I have only 64mb of Ram on Windows 98, Firefox 2 just gets too sluggish and takes too long to start up and uses up too much ram for me so I have to use Opera 9.25, which works just fine on 64mb ram on W98.

  42. I’m afraid Firefox 3 won’t run on Windows 98.. :roll:

  43. I think it’s time for you upgrade young chap.

  44. Nice work guys,
    One thing i have noticed though is how javascript performance degrades very much when using linux instead of windows. Is this a known behaviour? OS bug?
    Especially redraw operations like Drag and drop manipulation of the dom, cause the brwoser to crawl under linux

  45. Nemlah wrote:
    Nice work guys,
    One thing i have noticed though is how javascript performance degrades very much when using linux instead of windows. Is this a known behaviour? OS bug?
    Especially redraw operations like Drag and drop manipulation of the dom, cause the brwoser to crawl under linux

    I’m not sure whether that is a bug or not. I’m sure they still have some fine tuning to do so things will probably get better by the final release.

  46. For all the wonder seekers out there I have used firefox for quite sometime and I am not seeing the improvements that beta 3 is offering in fact I am starting to notice that perhaps IE7 is faster, much faster than firefox. Too much is being added/deleted and people are starting to get too secure with running firefox as their main browser. Fact, firefox is slower upon startup, loads slower, and on and on. I am not a proponent of IE but it is just as it is, firefox is outsmarting itself and the people using it. The only time and that is a maybe, the firefox outduals IE is when using stumbleupon but I cannot even be sure about that right now. Bob

  47. bob, your first name wouldn’t happen to be MS, would it?

  48. Yes, SunSpider isn’t a complete benchmark of browser rendering. However, it is a pretty good indicator, since a lot of sites use javascript, and browsers themselves implement it.

    Therefore, a difference of 100-300ms in Sunspider isn’t enough of an argument to claim that browser A is faster than browser B.
    Nevertheless, SunSpider is a pretty good indicator overall as to whats quick and what isn’t.

    To those witty people that claim that webkit is “way way way way… faster than firefox could ever be” – get your facts right. The latest firefox 3 nightlys are easily on par with webkit on windows.

    Plus, it makes no sense to compare firefox 3 with webkit. The former is a nearly finished product, while the latter is just a showpiece engine, which is by no means ready to go public as a product.

  49. 5899.6ms with Latest Minefield 3.0b4pre build

    [webkit.org]

  50. Jeefy, result is pointless. Different computers will give different times, on different OSs. Unless you do other tests to show relatively what’s quicker, there’s no point in posting a random result. I could get a faster time by using a fast comp, and similarly a slow time using a slow comp.

  51. Ton wrote:
    I’m afraid Firefox 3 won’t run on Windows 98..

    It was expected. It’s not Firefox requirement, it is requirement of engine. In any case, Windows 9x is dead and impossible to use in Internet without powerfull firewall.

  52. Oops, I messed up quoting…

    fd wrote:
    so? i hate javescript! i always uninstall it and never visit sites that use it…

    JP wrote:
    I do not care too much about this actually. At least the pages I browse do not need JS too much and definitely as much as that benchmark is going with its testing. Rendering speed is much more important.

    You guys should be aware that JavaScript is used in more places than you seem to think. Some basic browser functions are coded in JavaScript. Not to mention extensions, which heavily rely on it. So, even if you don’t visit JS-intensive sites, or even disable JS altogether, you may still have a nice perf gain.

  53. ============================================
    RESULTS (means and 95% confidence intervals)
    ——————————————–
    Total: 2477.0ms +/- 0.4%
    ——————————————–

    This is on the latest WebKit build for OS X on an 8-core Xeon Mac Pro (2.8 GHz)

  54. Calgary Guru wrote:
    ============================================
    RESULTS (means and 95% confidence intervals)
    ——————————————–
    Total: 2477.0ms +/- 0.4%
    ——————————————–

    This is on the latest WebKit build for OS X on an 8-core Xeon Mac Pro (2.8 GHz)

    I wasn’t exactly using an 8-core machine on my computer. ;) Just a slightly older Pentium D Dual Core, and I was using the Windows version of Safari which could be a little different.

  55. cool, Ill try this one now. I got some few spare minutes now

  56. Thanks.. Absoultly Firefox :)

  57. An interesting alternative study of Firefox 3 performance on linux : [mininglabs.com]

  58. MMMM i have some scripts in javascript and firefox run more slow than IE, want a test?, simply download this script [dynamicdrive.com] but instead of count 20 characters limit put 1024 or 2048, and you will see what browser run more slow. Firefox is very good and i use it, but is not the pearl of the world

  59. the mozilla 3 is more faster than IE. IE is too slow and can slow down your computer too.

  60. Firefox 3 very fast, fantastic. Thanks

  61. thanks for post / forever firefox 3

  62. With that said, I, personally, am waiting anxiously for the performance improvements of Firefox 3.

  63. cant find where u can download it.how much different is it really.is it worth downloading.also i heard google chrome is really good

  64. Version 3 is junk! I removed it because a lot of my add-ons are not supported with this new version. The browser is slow, the shortcut feature is stupid. Just junk.

  65. Firefox fantastic. thank you.

  66. Not everyone can afford to upgrade from win98.
    Lack of work in a sick economy an all. :(
    And Linux is not always a good alternative.
    I’m hesitant to dual boot, but we’ll see.
    Still, it would be nice to be able to run
    FF3 on win9x. :?

    Besides, how can you truly declare it runs faster,
    until you run it on one of those old win98 machines.
    That’s the real test. :o

  67. 542 ms.

    Safari 4.0 on a Mac Pro:

  68. Nice stuff!!!

Leave Your Comment


Message is the only required field.
Emails are not published.