DISQUS

satine.org: satine.org » SquirrelFish Extreme: Fastest JavaScript Engine Yet

  • Damian Cugley · 1 year ago
    Your charts show Safari Windows at 7.9 to and Safari on OS X at 13.7 -- are these on the same hardware? On the face of it it looks like there is something surprisingly slow about the Windows version...
  • oliver · 1 year ago
    Who cares for JS speed as long as the browser can be crashed by Flash content.
  • ethana2 · 1 year ago
    Yay progress!
  • Charles Ying · 1 year ago
    Damian: Oops, no the Windows test system is on different (slower) hardware. I suppose I should try it on the same hardware, but I don't have Bootcamp on this iMac. I'll try a Parallels version soon and post the results here.
  • coolfactor · 1 year ago
    @oliver

    It sounds like you have a chip on your shoulder. There's plenty that can contribute to an unstable browser. If you can point me to a flash file that will crash a given browser *every* time for everyone, then your point will stand. If it only affects your system, then you have somewhere to start to solve the problem.

    Way to go WebKit Team! This is fantastic. 280slides.com runs amazingly well in the latest WebKit nightly.
  • Andy · 1 year ago
    Maybe Oliver's point was that Adobe should pull their fingers out of their buts and do some optimisation of Flash on Mac OS X. It that wasn't Oliver's point, then it can be mine instead.
  • Jon · 1 year ago
    Why don't you include IE8 in these benchmarks? As a web developer I am loving the focus on javascript performance for the next generation of browsers, but IE is always a [limiting] factor in terms of knowing how far I can push the browser.
  • Enola · 1 year ago
    Jon:

    IE8 is ~ 10 times slower than the others.
    So it doesn't really qualify for this benchmark.. It would only make the bar graph unreadable.
  • Another Jon · 1 year ago
    I think Oliver was talking about Chrome's independent process model that lets Chrome survive Flash crashes, as well as any others for plugins. Which would be nice to see in Safari.
  • Vincent · 1 year ago
    It would indeed be nice to see a slower (e.g. IE) JS engine compared to it, even if its bar is so small it's unreadable. It'd be a great demonstration of the amount of progress made in a small amount of time.
  • buddyglass · 1 year ago
    I'm not sure what sort of system you used, but when I tested the three browsers on Windows XP running on a 2.4ghz P4, I found that Chrome totally smoked FF and Safari 4.0 (not using Squirrelfish Extreme) on Dromaeo.

    FF was approximately twice as slow as Chrome, and Safari 4.0 was approximately 1.33x as slow as FF.

    Chrome also performed ridiculously better on Google's own V8 tuning benchmark:

    http://code.google.com/apis/v8/run.html
  • Charles Ying · 1 year ago
    buddyglass: Safari 4.0 (I assume you're using 4.0 Developer Preview) is not the latest version of Safari + WebKit. The 3 browsers I'm comparing are using the latest nightly builds taken from last night, 2008-09-18. I'd be interested to see your comparison using the latest builds on your system.
  • buddyglass · 1 year ago
    True, the Safari 4.0 beta didn't have the latest version of Squirrelfish. But I was using a nightly build of FF3.1 as of approximately 3 weeks ago, and testing vs. the initial release of Chrome. You have FF handily beating Chrome on Dromaeo, which was the reverse of what I saw. I'll download nightlies of FF and Webkit, and test vs. the latest Chrome, on SunSpider, Dromeaeo and Google's V8 tester.
  • buddyglass · 1 year ago
    So I just finished testing, and my results match up more with yours than with what I remember from the last round of testing I did. Now that I think of it, though, I may have done my earlier tests on an iMac (Core 2 Duo) running XP inside a Parallels VM. Maybe Chrome takes better advantage of the two cores on that system? Or maybe I'm just on crack. Here are my results, with links:

    Webkit: r36682 (9/20) + Safari 3.1.2 (525.21)
    Minefield: 3.1b1pre (9/20), JIT Enabled
    Chrome: 0.2.149.30

    SunSpider:
    Webkit: 2814.0ms (http://tinyurl.com/4fwloh)
    Minefield: 2767.4ms (http://tinyurl.com/4z8gv4)
    Chrome: 3316.8ms (http://tinyurl.com/3osox2)

    Dromaeo:
    Webkit: 11346.40ms (http://dromaeo.com/?id=43093)
    Minefield: 13321.00ms (http://dromaeo.com/?id=43094)
    Chrome: 16296.00ms (http://dromaeo.com/?id=43089)

    V8 (higher is better):
    Webkit: approx. 480 (varies)
    Minefield: (fails)
    Chrome: approx. 880 (varies)
  • buddyglass · 1 year ago
    Charles, I posted my numbers last night, but Wordpress seems to have eaten the posts. Either that, or they were too long. Let me know if I should try to post them again.
  • buddyglass · 1 year ago
    Oh. Duh. They're right there. I look like a moron now. :)
  • movie fan · 1 year ago
    it's funny, the more i use Chrome (for windows), the more unstable it seems to get... crashes a lot more, can't handle sites with flash, hangs every time i close a tab... all that to say, i'm switching back to Firefox
  • DaVince · 1 year ago
    "Speed is not everything and we are not suckers for browsing speed. We want functionality!"

    In the case of JS, speed is functionality. You can make so much better and more complicated web apps if the underlying engine - JS, can handle so much more than before.
  • pingooo · 1 year ago
    It would be nice to see the scores of each individual tests since different real world web applications use javascript differently.
  • gigi · 1 year ago
    And why they don't test it vs Internet Explorer? I think IE it's the fastest.
  • Tony Arnold · 1 year ago
    @gigi: I believe one of the links in the comments above mentions that IE is about 8 times slower for JS than the slowest implementation of the other major browsers. IE is (and probably always will be) a lumbering monster that seems incapable of keeping up with it's competition.
  • e-okul · 12 months ago
    Hi,

    Charles Ying compiles a few more comparisons
  • Jay Godse · 10 months ago
    This is interesting because it shows how fast Javascript can go. All we need now is a good web framework similar to Rails or Django, but based on JS. Then server-side JS can take off.

    One thing these tests don't take into account is when you have multiple tabs open in FF or IE, memory does leak a lot, and one hung script can bring down all tabs. Just like Windows 3.1 or Windows 95 from 15 years ago. Because of this, I still prefer Google Chrome because of its one-process-per-tab model. However, many browsers will do the same within the next 18 months, and then JS speed will become the next performance factor.
  • Exteme Mobile Phone Spy · 10 months ago
    Firefox is approximately twice as slow as Chrome, and Safari 4.0 was approximately 1.33x as slow as FF.