Forgot your password?
typodupeerror
Programming Software

A Real World HTML 5 Benchmark 163

Posted by timothy
from the global-variables dept.
KidCompy writes "The newest browsers boast huge performance improvements, but how much do you trust benchmarks trotted out to prove those claims? Do they reflect the real uses to which developers will put HTML 5 and JavaScript? We've extracted several benchmarks from our existing programs to measure actual versus theoretical performance."
This discussion has been archived. No new comments can be posted.

A Real World HTML 5 Benchmark

Comments Filter:
  • How in the world does a site get Slashdotted as soon as its article as posted? I mean, there weren't even any comments yet when I clicked it!

    • Your connection must suck, it loaded in under 5 seconds for me. Well... I know that most Slashdotter's are under 25 and pretty impatient when it comes to load times.
      • Re: (Score:2, Funny)

        by santax (1541065)
        Are you suggestion that most slashdotters are female?
      • Your connection must suck, it loaded in under 5 seconds for me. Well... I know that most Slashdotter's are under 25 and pretty impatient when it comes to load times.

        Worked fine for me too, loaded in seconds.

        • LMAO at "under 25." I'm under 25 and don't have a single tech friend that could keep up around here. No way. Best guess is 30's and up.
  • I don't have much faith in this benchmark, or the company/their programmers, for that matter.

    My browser gets identified as:

    Browser Family: safari Browser Version: 534.6

    Oddly in contrast, the "About Chromium" has a somewhat different version and "Browser Family". (A later build, not sure which at this point.)

    Interestingly, my browser didn't perform all that well on any of the tests.

    • We're just reporting what JQuery reports. We'll get a better browser identification library and get more accurate values.

    • by Carewolf (581105) on Saturday December 25, 2010 @06:26AM (#34664846) Homepage

      Chrome is reporting both as safari and as chrome, it even includes a safari version number, as a minimum safari-version it is compatible with. When Chrome is spoofing to anyone not specifically testing for Chrome, it is hard to blame anyone misdetecting them.

      • IE spoofs as Firefox (Score:2, Interesting)

        by tepples (727027)

        When Chrome is spoofing to anyone not specifically testing for Chrome, it is hard to blame anyone misdetecting them.

        That's nothing. Both Safari and IE report as Firefox. The "Mozilla/" in the user agent string represents Netscape Navigator, and the last versions of Netscape (8 and 9) were customized versions of Firefox. Moreover, WebKit browsers such as Safari and Chrome spoof as Konqueror and specifically recent versions of Firefox ("KHTML, like Gecko")

        • by Carewolf (581105)

          It all depends on where you place the name, but in essence you are right. It is a complete mess :D

  • The Most Useless Benchmark(tm)

  • Browser: Firefox 3.6.13
    OS: Windows Vista Home Basic
    CPU: Intel T1600 @ 1.66GHz
    RAM: 2GB of RAM

    Benchmark #1: 328 iterations
    Benchmark #2: 10 iterations
    Benchmark #3: 3005 iterations

    FWIW

    • Browser: Firefox 3.6.13
      OS: Windows Vista Home Basic
      CPU: Intel T1600 @ 1.66GHz
      RAM: 2GB of RAM

      Benchmark #1: 328 iterations
      Benchmark #2: 10 iterations
      Benchmark #3: 3005 iterations

      FWIW

      Sorry for replying to myself, just noticed 5132/50000 rwb points for the above benchmark. The overall score is shown at the top of the page.

      They should show it on the bottom, since that's when most people will look for it.

  • Chrome > Opera > FireFox
  • 1 - 684
    2 - 55
    3 - 8499

    12508 / 50000

    • That says blazin' fast to me, you got a third of the ideal # of iterations on bumperbots. Was that on Safari?

        • Browser Family: safari Browser Version: 533.19
        • iMac 2.66 GHz i5 8GB
        • 11240/50000 rwb points
        • 622
        • 44
        • 7610

        For what it's worth, there was a lot of stuff open (mail, photoshop, skype, etc.)

        • The code only maxes out one core at the moment, so your experience is in-line with what I see as well (I run while compiling and running the dev server with the box gettin' all swappy, and it runs okay even then).

  • by Haedrian (1676506) on Saturday December 25, 2010 @04:58AM (#34664672)

    Strangely enough I don't think bots which smack into each other and have collision sensors are very much real world. I don't plan on using my browser to animate bots colliding into each other in the forseeable future...

    • by bloodhawk (813939)
      I don't think any of the tests were representative of real world HTML performance let alone the sprite collision detection, the article and the site are just a bad joke.
    • I don't plan on using my browser to animate bots colliding into each other in the forseeable future

      Video game developers do. If HTML5 proponents want it to replace Flash, it needs to be able to do so for FarmVille, Tetris, and all the other popular browser games.

    • Not to mention, it kind of looked like they were running some sort of BASIC on top of JavaScript...

  • How much of this is dependent on the pc?
    • by rhade (709207)
      and how do we know this is being done the most efficient way? The code needs to be peer reviewed
      • Most efficient? No. First of all, we want to accentuate the positive with HTML5, and that means leveraging Web Workers and freeing up the DOM thread to do DOM-like things. That is tricky though, lotsa delicate messaging required between modules, and we only halfway refactored the code for that before the Christmas release deadline hit. So, I anticipate all Web Worker-capable browsers will double in Tasty script interpreter performance once we get a chance to implement that. But, as for now, we're singl

  • Latest Chrome on a Dell Inspiron 1501 laptop, AMD Turion 1.6 GHz 1 GB RAM, ATI Radeon xPress 1150 using UMA
    #1 - 503
    #2 - 37
    #3 - 6670

    Your browser's total score is 9446 out of a possible 50000

    IE8 same machine:
    #1 - 94
    #2 - 1
    #3 - 465

    Oddly, I cannot seem to copy and paste from IE.

    A second run on IE8:
    100/1/1215... it seems like minimizing the browser increases performance.
    Let's try minimized on Chrome:
    541/44/6701 - slight improvement. - Your browser's total score is 9884 out of a possible 50000

    Let's try Chrome in a

    • by ThreeGigs (239452)

      Egads, forgot to mention Windows XP. Seems Chrome beta (9.0.597.19) is faster on XP, even on a 4 year old laptop if my results on Chrome are directly comparable to others.

    • YES! Minimizing IE6-8 *DOES* improve performance, and I found that interesting as well.

      My take is that IE is "close" to the kernel in a lot of ways, and so you get them turning off their blitter or whatever (even though the scene is still rendered in the offscreen buffer). IE really is a mix of very fast and very slow parts (mostly the JavaScript engine is the "slow".) You know, while I was developing that code, I found so many interesting ways to hack the code to squeeze out marginally better performan

      • by gbjbaanb (229885)

        I'm praying IE9 will rock.

        why? there's no reason to stick with a browser anymore, they're completely interchangeable, so if IE9 doesn't rock (or to put it another way, as IE6->8 don't rock) then get yourself Firefox, or Chrome or Safari or Opera. Really easy, and you'll get used to the interface in no time at all - in fact, you might like some of the fancy bits in some of the other browsers and think "why the heck did I ever use IE?"

        Results for FF 4.0 beta 8: 469/30/4835 (7837 total). I'd have expected m

        • completely agree, just want naive IE users to get a decent experience whenever they upgrade to the next version of windows. (windows 8 will be one of those rare win upgrades I will recommend to f&f when it comes out)

            I am displaying the standard "use at your own risk" message box when users launch the shell from the main site on IE and we detect there is no canvas tag support.

        • by smash (1351)

          If you're in the corporate world there are plenty of reasons to stick with IE. Its already there, it can be secured adequately (security zones, filtering firewall at the edge), group policy and it actually works with DHCP proxy autodetect, which mozilla and chrome do not. Which is a big issue if you have mandatory proxy usage inside your network and roaming users who want their browser to work without it when outside.

          The UI issues are a non-issue, its the under the hood stuff that keeps IE on corporate

      • by Z34107 (925136)

        I had high hopes for IE9 as well, but they didn't pan out. On my system (2.93GHz Core2 Duo, NVIDIA GeForce 8800 GTX) IE9 got 4846 points versus Chrome 8's 9911.

  • Chrome 8
    Kubuntu 10.10
    Phenom II X3 720
    4GB RAM (in 32-bit)
    Radeon 4700 (with fglrx driver)

    11141 points.

  • My Results (Score:4, Interesting)

    by bgarcia (33222) on Saturday December 25, 2010 @06:34AM (#34664858) Homepage Journal
    My results (running on quad-core Windows Vista 32-bit):

    Chrome 8.0.552.224: 8641
    Firefox 3.6.13: 5082
    Internet Explorer 8.0.6001.18999: 2145

    • My android phone (HTC desire/2.2) scores 2427. If this benchmark is to be believed, my phone is better than your quadcore machine for browsing JS heavy sites if you're using IE8.
      • by bgarcia (33222)
        Ha! I didn't even think about running this on my phone.

        I just ran this on my brand-spankin-new Nexus S running Gingerbread & got a 2482.

        Wow, that's just a really pathetic showing by IE.

    • by root_42 (103434)

      AMD Athlon(tm) X2 Dual Core Processor BE-2350, 2.6.35-22-generic Ubuntu x86_64 GNU/Linux

      Chrome 8.0.552.224: 8008
      Firefox 3.6.13: 4395
      Konqueror: Did not pass... Hans during the first benchmark

    • by Bengie (1121981)

      i7 920(2.66) stock on Win7x64

      Chrome 9.0.597.19: 14,993
      IE 8.0.7600.16385: 2438

    • > My results (running on quad-core Windows Vista 32-bit):
      Quad-core running 32-bits?
      And Vista?

      I guess Santa doesn't read ./

    • by yoshman (1416413)
      Built in browser, HTC Desire, 1GHz ARM Cortex A8, Android 2.2: 2404 (137/1/1690)
      Chrome 8 (64-bit), Dell E4300, 2.4 GHz Core2Duo, Kubuntu 10.10: 11306 (652/61/7057)
      Rekonq (64-bit), Dell E4300, 2.4 GHz Core2Duo, Kubuntu 10.10: 10006 (576/27/6634)
      Chrome 8 (32-bit), MBP 2010, 2.4 GHz Core i5, OS X 10.6: 11475 (630/56/7686)
  • by ponos (122721) on Saturday December 25, 2010 @06:50AM (#34664896)

    Well, seeing a Mandelbrot algorithm running on an interpreted language on top of an interpreted language and
    struggling on my super powerful quad core makes me suffer. I had coded the Mandelbrot fractal in assembly
    and it ran faster on a 80386...

    Now get out of my lawn...

    • by Seth Kriticos (1227934) on Saturday December 25, 2010 @08:50AM (#34665074)

      Don't you worry, one of these days, someone will write a full 386 emulator in JavaScript bringing the full Web 2.0 fidelity to your lawn.

      And if you have an especially powerful rig, you'll even be able to use the 'turbo' button.

      • by Anonymous Coward

        I think someone is working on this in minecraft

    • by gbjbaanb (229885)

      yeah, mind the grass.

      I was surprised at the language used: the basic code at the side showing the code it was running, complete with gosubs, I did think "WTF", but then I read the rest of the site - particularly the bit "ClubCompy is an innovative new service for kids of all ages to learn about computer programming?" and it all became clear, and took me right back to the old days when I was learning programming using code just like that.

      Ah, happy days. I'm old enough now not to be surprised that things come

    • by devent (1627873)
      True. But had your Mandelbrot 16 million colors, HD resolution of 1920x1680 and could millions of clients access it with a generic software (the browser), which runs on Linux, Mac, Windows both in 32bit and 64bit and run it for them self?
  • 600/47/7291 One core maxed out during the tests. The other 7 sitting idle. So if the tests were multi-threaded the score would have been 86808 out of a possible 50000. :)
    • by Anonymous Coward

      Safari is just fast. Even on my iMac core2 duo 2.0Ghz (2007) i got 9340. However, Firefox 3.0.19 on MidnightBSD only scored 4029 on an AMD phenom 9600. On the same system, Chromium scores 10617.

  • 89/1/599 , iOS 4.2.1
    • On a HTC Desire/Android 2.2 froyo, it scores 2427 (144/1/1627).

      Browser is detected as Safari (as expected, since the default is webkit based).
  • by jovius (974690) on Saturday December 25, 2010 @09:00AM (#34665096)
    MacBook Pro mid 2010 i5 2.4GHz, latest public browser versions

    Firefox: 5055 / window minimized: 4930
    Safari: 10628 / 11210
    Opera: 9121 / 9487
    Chrome: 10903 / 11035


    On virtualized Windows XP home SP3 (Parallels desktop 6):

    Firefox: 5878 / 6749
    Opera: 9170 / 9734
    ie 8: 1463 / 1440
    Chrome: 10920 / 11392


    Another reference point, virtualized Ubuntu 10.10

    Firefox: 5165 / 6040
    Chrome: 10769 / 11064
    Opera: 8942 / 9500


    Chrome was identified as safari 534.10 on all OS's. The results seem to fluctuate a bit from run to run, from 10 to 1500 points (i did some of the tests two - three times). It seems I get different results each time the test is run.
  • Obligatory result (Score:3, Interesting)

    by rshane (27512) on Saturday December 25, 2010 @09:33AM (#34665144) Homepage

    Core i5-650
    8gb RAM
    Windows 7

    Firefox 4.0b8 - 8246/50000
    Chrome 8.0.552.224 - 12611/50000

  • CR-48, Chrome OS 0.9.128.12: 4416
  • AMD Phenom II X2 555 BE 3.8ghz
    4gb RAM
    Windows 7 64bit

    Chrome 8.0.552.224 - 17041/50000
  • Not sure how real any of those are. Bumper cars? A really slow paint function?

    On my anemic (1.6 GHz Atom) system, Chrome gets 3986 while Opera gets 4250 (sorry, no Firefox installed).

  • by Anonymous Coward

    Emulating BASIC programs from 1985 with GOTOs and line numbers in Javascript is what we do on the web all day.

    Hey, did you know that ClubCompy is an innovative new service for kids of all ages to learn about computer programming?

    kids of all ages = 40-something "kids" who are nostalgic for their first home computer.

  • What if I don't need either HTML5 or Javascript? Simple old HTML worked *fine* for a decade to do what it was designed for -- display information (and allow simple forms entry, e.g. electronic transactions). It did *NOT* require HTML5 nor Javascript. Those are applications for people who want to use *my* computer resources for *their* purposes [3].

    It is reasonable to point out that Google has scanned 7 million books, PubMed/Medline has 21+ million records, and Wikipedia has 3+ million articles in English

    • sounds like you don't need an HTML5/JS benchmark all that much, then.
    • by Xtifr (1323)

      What if you don't need a web browser at all? Decrypting 1940s-era "secure" communications doesn't require all these newfangled applications. If the machines built at Bletchly Park [wikipedia.org] were good enough for Allied Intelligence in WWII, they ought to be good enough for you. Or are you trying to say that you're better than those heroes that helped save the world?

      For that matter (and I actually think this may be a better analogy), why do we need interactive terminals at all? Batch processing with punch cards wor

      • by bradbury (33372)

        The point primarily was -- the distribution of simple information, which was what IMO the web was intended to do, does *not* require HTML5, nor Javascript. Indeed Google's Gmail points out that one can produce very functional apps without resorting to Javascript. If even some small amount of effort were put into maps.google.com or mapquest I'm sure those would work just as well without Javascript too.

        Yet I have *yet* to see a benchmark which measures the simple functions which are those needed for 95+% of

        • by weicco (645927)

          I've been saying the same thing for almost 10 years now. Every time I've been blamed as a "troll". I'd like to have web which just displays information without hundreds of lines of CSS masturbation. For any more complex things I prefer rich apps which uses Web Services (or whatever) when communicating with server(s).

          But if five billion other internet users wants to have HTML5, CSS, JavaScript, Flash, etc. etc. who am I argue with them.

        • by DerPflanz (525793)

          The point primarily was -- the distribution of simple information, which was what IMO the web was intended to do, does *not* require HTML5, nor Javascript.

          But, guess what, the web evolved. It isn't used any more for what it was primarily designed for. It evolved into a platform, where we (except you apparently) do shopping, communication, games, etc. The beauty of the whole thing, is the web is both things now. HTML5 lends itself perfectly for the semantic non-javascript web, with the new tags. It lends itself better for forms with the new input-types (why did we have to wait so long for that?). And it also lends itself for being a complete programming platfo

  • I am running Windows 7 64-bit and have IE9 Platform Preview 7.
    Note: This code is 2 releases after the Beta

    On the second test it crashes.
  • '10 MBP 13" 2.4Ghz w/SSD maindisk:

    Firefox 4.0b8 = 6623
    320 / 33 / 4266

    Chrome 8.0.552.231 = 10018
    562 / 48 / 6630

    Safari 5.0.3 (6533.19.4) = 10210
    550 / 39 / 7135

Real programs don't eat cache.

Working...