Web Browser Benchmarks

[Alex Libman]

In light of all the other geeky debates we are having on this forum, this is a specific thread to post Web browser benchmark results for your computer.  These days browser performance should be the #1 most important component of any geek-fights on which system is superior.  Results from multiple operating systems on same hardware would be most interesting, though the new browser performance wars are interesting on a single OS as well.

I started with the Futuremark Peacekeeper benchmark, and I'm going to post my own results in a bit.  I wouldn't recommend this benchmark to others, though, because it takes about ~5 minutes per run, and it has a totally retarded bug where it tries but fails to track your performance across sessions, and sometimes, totally at random, it gives you an error message instead of results after wasting your time.  (ex: "Could not find result for key [blah]. If you trying to benchmark another browser please make sure you have copied the url correctly and try again.")  I'll leave it to someone else to recommend which benchmarks are best.


A few general notes on browser benchmarks:

• Close all non-essential apps on your system except for the browser being tested, with one maximized tab / window running the test.  Don't do anything while the test is running, and even a screensaver coming on could mess up the results.  Move the mouse pointer away from the animation canvas.  Some browser add-ons may theoretically interfere with the results, but I can't think of any that would have a significant impact.

• Make sure all browsers have roughly the same page zoom level during the benchmark.  This is particularly important for IE9, which defaults to a higher zoom level on larger screens, which is actually a pretty good feature, but having to scale to more pixels means an unfair benchmarking disadvantage.  Press Ctrl together with the Plus or the Minus key (which on some keyboards would also require holding Shift) to zoom in or out.

• Don't make any significant system changes in between tests.  If testing on multiple operating systems, please post the details, especially which video driver you are using.

• If you want to get really accurate, run the test 2-3 times on each browser and pick the best time, as the higher times indicate that something random happening on your system interfered with the test.

[blackie]

Seems like you should post a general system benchmark score for CPU/Memory before you move on to other things.

[Alex Libman]

Not really, because this is all about racing browsers against each-other and also comparing operating system performance (if the same browser version is benchmarked on both).  It's not about how well your system performs relative to somebody else, but which OS / browser is able to make the most of it.

My predictions:  Firefox 4 isn't fastest for anybody.  Whether IE9, Chrome, or Opera come out on top will depend on your hardware, with the newest high-GPU computers favoring IE9.  And same browser versions on Windows / MacOS X will beat Linux every time. 

[Alex Libman]

My own benchmark results are delayed a bit, because my main goal is to benchmark the browsers on same (low-end / legacy) hardware with different operating systems, and I didn't get around to some of those OS'es just yet.  Plus I decided to use this UNIX desktop hopping session to give "my long twilight struggle" against Nvidia driver crashes one final try... 

I would very much like to see multi-OS benchmark results from other people as well.  Windows has always been far ahead of Loonix in this regard (ex), but the benchmarks I've seen are now outdated, plus it can always be claimed that the Loonix system wasn't tweaked just right.  So I'm wondering if any Loonix apologists on this forum can show Loonix best Windows...  at anything... 




In the meantime, here are some recent Windows 7 results from CNET:



IE9 seems to be a winner when it comes to games, and Chrome when it comes to raw JS performance.  The tests won by Firefox are less significant, but it's clearly a fast well-rounded browser.  As is Opera, which is absent from their test (Safari is also missing, but I wouldn't call it "well-rounded").

[Turd Ferguson]

WTF????


You dont include the awesome, efficient, streamlined, ad free AOL browser??? This benchmark test is BULLSHIT!!!!

[Alex Libman]

Yes, I am conspiring against AOL!  I was a major AOL hater since 8th grade, because that's what all the idiots were using, while I used a dial-up MajorBBS which provided a UNIX shell (it was FreeBSD actually) and SLIP/PPP.  Won't even e-mail a hot girl with an AOL address! 

Seriously, modern AOL software is just Internet Explorer with a different interface thrown on top.  IE, Firefox, Chrome, and Opera represent unique HTML5-compatible implementations.  Safari shares the same base as Chrome, but it has a 5-10% market share (among a particularly coveted demographic of rich yuppies who like to overpay for shiny things) and it is sufficiently different from Chrome (i.e. a different performance-competitive JS engine) to warrant inclusion.

[Cognitive Dissident]

Safari shares the same base as Chrome, [bigoted unnecessary bullshit omitted] and it is sufficiently different from Chrome (i.e. a different performance-competitive JS engine) to warrant inclusion.

It also predates Chrome, so it could hardly be a variant of Chrome.