Firefox 3.0 Beta 4 - Benchmarked

Firefox 3.0 Beta 4 - Benchmarked

Summary: And it's out! Mozilla release Beta 4 of Firefox 3.0. And it is good!In this post I take a look at some of the changes in this beta and benchmark it against other popular browsers.

TOPICS: Browser

And it's out!  Mozilla release Beta 4 of Firefox 3.0.

Firefox logoThis latest beta refresh includes includes over 900 additional improvements related to the previous Beta 3 release. At this stage most of the improvements concentrate on areas such as security, plugging memory leaks, improving performance and tidying up the UI.

Be sure to check out the Firefox 3.0 Beta 4 gallery!

Firefox 3.0 Beta 4 - Benchmarked

Firefox 3.0 Beta 4 - Benchmarked

Firefox 3.0 Beta 4 - Benchmarked

Firefox 3.0 Beta 4 - Benchmarked

Firefox 3.0 Beta 4 - Benchmarked

Firefox 3.0 Beta 4 - Benchmarked

Firefox 3.0 Beta 4 - Benchmarked

Firefox 3.0 Beta 4 Benchmarked -->

Note:  All testing carried out on Windows Vista Ultimate 32-bit.

SunSpider JavaScript Benchmark

The results of the SunSpider JavaScript benchmark shows that Firefox 3.0 Beta 4 is the fastest browser on the block. 

Firefox 3.0 Beta 4 - Benchmarked

Note:  Neither version of Opera fully completed the MD5 part of the test, and these results have been omitted.

Here are the raw results:


I also ran all the browser through the ACID 3 standards compliance test.  In this test Firefox 3.0 beta 4 comes first and Opera 9.50 beta 1 second.

Firefox 3.0 Beta 4 - Benchmarked

Here's the ACID 3 output for Opera 9.50 Beta 1:

Firefox 3.0 Beta 4 - Benchmarked

Here's the ACID 3 output for Firefox 3.0 Beta 3: 

Firefox 3.0 Beta 4 - Benchmarked

And here's the winner - Firefox 3.0 Beta 4:

Firefox 3.0 Beta 4 - Benchmarked

Internet Explorer 7 fails misterably:

Firefox 3.0 Beta 4 - Benchmarked

Here's what the properly rendered page should look like:

Firefox 3.0 Beta 4 - Benchmarked

Other thoughts

  • Firefox 3.0 beta 4 is a very responsive browser.
  • Memory usage seems to be unchanged from Beta 3.  No signs of spiraling memory consumption.
  • Start up and shut down is nice and fast.

Here's the official download page containing important information about this release.  Release notes here.


<< Home >>

Topic: Browser

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • IE8b1 Acid 3 Score

    Hm, I can only manage to squeeze 17 out of IE8b1. Did you use a different test site? Maybe its the cross-domain thing showing up ...

    Quite odd. Its true on both the 32 and 64-bit varieties.
    • No ...

      ... I just fouled up the sort in Excel ... fixed now!
      Adrian Kingsley-Hughes
      • What about webkit 3.1?

        The latest webkit builds are kicking serious ass in the acid 3 test (90/100). Not sure about javascript performance. The changes will be released as Safari 3.1 as far as I'm aware. Might be worth throwing into the mix, as you have other betas in there also...what do you think?
        • Except

          This is about firefox, and you're forgetting where no one cares about safari. Stay on topic.
          • Except ?

            "Stay on topic."

            It was on topic silly the purpose is to have competitive performance tests against other browsers, beta or not and Safari Beta does belong on the list!
  • ACID3 Results

    You've got your screenshots and results flipped around a bit - the 60 with the 'FAIL' in the left corner is Opera 9.5 beta 1, the next 60 is Firefox beta 3 and the 67 is beta 4.
    • You must be seeing an old version ...

      ... are you still seeing it? That should have only been live for a couple of minutes.
      Adrian Kingsley-Hughes
  • Good test

    Good test, but it would be nice if you did some testing on a more commonplace computer- something like an ordinary XP computer with 1 or 2 Gb of ram. This would be more meaningful to most of your readers- Vista is used only by a small minority of computer users at this time. I question the value of the test to most computer operators when the testing is conducted on Vista, and probably with a VERY high powered machine comparatively speaking. Would such a test have similar results on a more ordinary machine? Not to pick nits, I am simply wondering what applicability your testing would have to most users.
    • Well ...

      ... it would make no difference to the ACID 3 test, and since the same was used for all tests (1GB or RAM) the SunSpider tests should indicate which browser is fastest no matter what system was used.
      Adrian Kingsley-Hughes
      • Thanks for feedback

        To be clear, you are running Vita Ultimate on 1 gb of RAM? Doesn't that make for a very slow machine? It was not clear to me if results would be hardware dependent. And you think Vista vs XP would make no difference? Just curious, given the internal operations of how those OS's work is kept concealed in many ways. Not dissing your test.
        • Not really

          If you take the time to configure vista to the featureset recommended for your hardware configuration, it runs substantially better than xp.

          However, on a webbrowser test, you wouldn't see any significant difference unless the software were specifically written to run slower on one specific operating system. One thing I would like to see is them compare x64 and x32 versions.
        • Just!

          Test it yourself, Download, Install and try it out. I have it on Vista Premium (2 Gig Ram), Vista Basic (1 gig Ram) and XP Pro (4 gig ram). Firefox is fast no matter what system. Explorer was still trying to open up but yet Firefox was ready and browsing in the blink of an eye!
  • Your results are already outdated.

    In my article, my results show Safari Webkit scoring much higher in Acid3. It should place Safari at #1.
    • Actually it's a tie!

      Last wednesday both Opera and Safari Webkit achieved Acid3 100/ even your tests are now this current rate of development, in six months this whole discussion topic will also be outdated...
  • RE: Firefox 3.0 Beta 4 - Benchmarked

    I suspect that this is a comparison of shipping and beta browsers and not random nightly builds and that's why you don't see opera weekly builds or webkit nighlies in the chart. When Opera or Apple ship their next betas, I'd expect to see updated results.

    - A
    • Exactly ...

      ... otherwise I'd be here for days!
      Adrian Kingsley-Hughes
  • RE: Firefox 3.0 Beta 4 - Benchmarked

    Below is the results from my Firefox on the Sunspider test for WinXP Pro 32bit. On a system with the following: AMD Athlon 64 X2 5200+, 4GB memory.

    RESULTS (means and 95% confidence intervals)
    Total: 19420.2ms +/- 0.6%

    3d: 2415.6ms +/- 1.0%
    cube: 600.2ms +/- 2.9%
    morph: 1424.8ms +/- 1.2%
    raytrace: 390.6ms +/- 0.2%

    access: 1634.6ms +/- 3.7%
    binary-trees: 175.4ms +/- 29.6%
    fannkuch: 540.8ms +/- 2.0%
    nbody: 587.4ms +/- 2.9%
    nsieve: 331.0ms +/- 2.5%

    bitops: 4760.0ms +/- 0.5%
    3bit-bits-in-byte: 356.6ms +/- 2.5%
    bits-in-byte: 387.6ms +/- 2.3%
    bitwise-and: 3565.8ms +/- 1.1%
    nsieve-bits: 450.0ms +/- 3.6%

    controlflow: 125.0ms +/- 0.0%
    recursive: 125.0ms +/- 0.0%

    crypto: 715.8ms +/- 2.3%
    aes: 259.4ms +/- 4.1%
    md5: 190.8ms +/- 4.5%
    sha1: 265.6ms +/- 0.3%

    date: 5081.4ms +/- 0.3%
    format-tofte: 1153.2ms +/- 0.7%
    format-xparb: 3928.2ms +/- 0.3%

    math: 1312.4ms +/- 1.8%
    cordic: 575.0ms +/- 1.5%
    partial-sums: 465.6ms +/- 1.9%
    spectral-norm: 271.8ms +/- 3.8%

    regexp: 884.2ms +/- 4.6%
    dna: 884.2ms +/- 4.6%

    string: 2491.2ms +/- 0.9%
    base64: 537.4ms +/- 2.0%
    fasta: 469.0ms +/- 0.0%
    tagcloud: 394.0ms +/- 4.2%
    unpack-code: 703.0ms +/- 2.0%
    validate-input: 387.8ms +/- 2.3%
  • Great Benchmarks What about RoboForm

    I think the benchmarks are great, but when will I be able to use RoboForm on it. I get tired of go back and forth between 2 and 3, just because I need to use Roboform. What have you heard?
    • Siber Systems will probably make a compatible version when

      Siber Systems will probably make a compatible version when it comes out of beta.

      it took forever before they made a version to work with 2
      SO.CAL Guy
      • roboform for Fx 3 b4.

        currently supports b3-b4. They do a silly version check inside their extension. When I was a reviewer, I'd mention that they don't need to do an independent check, the functionality is already built into Fx every time I'd approve the extension.

        Ah well, least they're keeping somewhat up with the beta builds, only a couple days behind now.