Slashdot Mirror

TFA uses http://jsbenchmark.celtickane.com/Run.aspx which is a joke.
A useful benchmark is Futuremark's Peacekeeper, really, since it tests a wide variety of common tasks. On my machine, Chrome's the fastest at raw JS, but (by far) slowest at rendering...besides Firefox, which is actually slowest at -every- benchmark -every- time (by a typical margin of 5-10x or more; 4 RC is even slower than everything else on its own benchmarks like Kraken).
Even Opera (with no hardware acceleration at all) beats Chrome at complex graphics and rendering on canvas. Chrome is also the only accelerated browser to get incorrect rendering/redraw on many of the various Canvas acceleration tests/demos.
IE9 is the fastest at rendering complex stuff, while still keeping up with the pack on regular JS, which I dare say is a useful area to be #1 in.
If the browser compiles all of your JS really fast, but then takes a lot of extra time to actually display it, you're still bottlenecking as if you had an incredibly slow JS engine, just at a different part in the average case.

If you do HPC in your web browser via JS, Chrome is definitely the way to go, though.

Ah, interesting. The source has actually changed significantly from the last time I looked at this. Or the article author is on crack about what benchmark it is.

I assume they ran http://jsbenchmark.celtickane.com/Run.aspx which actually looks like a legitimate benchmark. Celtic Kane used to push a dom/css/etc benchmark which basically did things like "measure speed of CSS animation" by setting style.x and style.y and the like.

Running a Javascript speed test I found at: http://celtickane.com/webdesign/jsspeed.php, I get the following results:

Firefox 3.0
===========

Try/Catch with errors 19
Layer movement 94
Random number engine 21
Math engine 64
DOM speed 387
Array functions 6
String functions 9
Ajax declaration 33

Overall: 633

Minefield (Firefox 3.1)
=======================

Try/Catch with errors 15
Layer movement 57
Random number engine 16
Math engine 19
DOM speed 181
Array functions 7
String functions 7
Ajax declaration 31

Overall: 333

Which, by my simplistic reckonings, is about a 2 fold increase from FF3.0 to FF3.1

Impressive to say the least :-)

James

Cause we all like benchmarks.... Here is a breakdown of some browsers' times: (all were freshly installed and used http://celtickane.com/webdesign/jsspeed.php for tests) minefield 3.1b2: ~193ms chrome 0.2.149.30: ~234ms opera 9.61: ~203ms internet explorer 7: ~2328ms safari 3.1.2: ~203ms These were all done on the same computer.. this is why we have competition kids..

Grammar checking, resizable text boxes, automatic language translation, fast and efficient javascript engine, just for starters. [...] The references I made to OS X were mostly with regard to problems with failing to properly code for that platform and take advantage of the ways it is superior to Windows.

Opera doesn't have any built-in spelling or grammar checking, so that hardly counts as a "feature copied from other browsers" (unless you mean they copied the absence of a built-in checker?). Under Windows / Linux, Opera uses GNU Aspell and, under OS X, it will use the operating system's checker. I guess that contradicts your theory that "Opera is not coded to take advantage of the features offered by OS X".

Resizable text boxes are not defined by the current HTML / CSS standards. If and when resizability becomes a valid property of text boxes, recognised by the W3C, I'm sure Opera will implement it. Allowing the user to resize text boxes on pages that weren't coded to deal with that can seriously break their layout. If the web site creator wants the boxes to be resizable, right now, he or she can implement it with JavaScript (and this works fine in Opera). Opera also does not support some proprietary Mozilla CSS extensions, for example, but will support the "official" (W3C) equivalents.

I'm not sure what you mean by "automatic language translation". If you're talking about Safari's Language Translator widget, that's a 3rd party plug-in and, in any case, Opera can do the same thing; you just need to select some text, right-click on it and select the language (no need to install 3rd party widgets or cut & paste the text). Opera has done this since before Safari even existed, so again I don't see how it can be considered a feature copied from another browser.

Finally, I also don't see how "fast and efficient javascript engine" counts as a "feature copied from other browsers"; when the current Opera JS engine was released, it was 4 times faster than pretty much any other browser of the time (still is the fastest in most interactive operations, although string handling is a bit slow - possibly due to the extra privacy / security features in Opera). Personally I've never felt that Opera was "slow" at handling JS (can't say the same about Firefox 2. Firefox 3 is much better, but still not as responsive as Opera in several sites I visit).

Gecko based browsers still run it twice as fast as Opera.

Latest "non-partisan" benchmark I could find online at an English-language site (September 2008):

http://celtickane.com/webdesign/jsspeedarchive.php

Opera 9.5.2 runtime = 420
Firefox 3.0.1 runtime = 538

Firefox (Gecko) is 30% slower (or was, one month ago - that may have changed, but most people don't update their browser every day, so my original point stands).

Sunspider is a "core" JavaScript benchmark that doesn't test any web page manipulation or interaction. That's fine if you're planning to code JavaScript applications, but won't tell you much about how well a browser handles JavaScript used in an interactive web page (which is what really matters to 99% of end users).

BTW, if you look at the results of Google's V8 benchmark, Google Chrome is more than twice as fast as any other browser. And I bet Microsoft could come up with a benchmark where MSIE would be #1. That doesn't mean it'll translate into the real world.

Granted this is anecdotal, but on my home PC I've been using Google Chrome exclusively for a couple of days now and I really like it. It doesn't take several seconds to boot like Firefox has for a while now.

Performancewise, I compared my current Firefox window with ten tabs open (Slashdot, Kotaku, Penny Arcade, Google, Yahoo, CNN, Craigslist, Pokemon.com, Newegg, The Tech Report as test tabs) to Google Chrome with the same ten tabs. Firefox came out to 123MB, Chrome was 179 BUT Firefox was chugging as the 10 pages were loading (this was on an old single-core Athlon 64 with 512MB ram), while there was no slowdown whatsoever on Chrome on the same machine. Sure, don't use it if you only have 128MB of RAM, but at that point I don't think you should be using Firefox or Windows XP anyway.

The Javascript performance in both are pretty good, but Chrome's is slightly better. On Celtic Kane's Javascript speed test, Firefox 3 usually gets around 380ms and Chrome gets around 80-100 less.

http://celtickane.com/webdesign/jsspeed.php

Google Chrome lacks some of the polish and features that Firefox does (no built-in RSS reader is noticeable), but I really like the responsiveness and speed. For now I'll stick with Chrome and see what Firefox 3.1 and the new Chrome has to offer.

Wow. Your comment is so insightful.

Do you have any idea how PGO works? Because you seem to be under the interesting impression that it is an optimization for the SunSpider benchmark. It is not. Also, the performance increase of the pre-b4 builds is mainly due to engine optimizations rather than PGO, even though it definitely contributes

And if you STILL think that PGO and the latest buids are rigged to score highly on SunSpider, why don't you try out some other javascript tests, such as http://celtickane.com/webdesign/jsspeed.php , or for that matter any other javascript test. You will find that the results are consistent, and that the pre-b4 nightly beats the latest opera build.

So please, before making comments based on your lack of knowledge, do download the tests and run them yourself.

There's no need to misguide other people just because you don't like Firefox, or whatever other reason you may have had for writing a biased unresearched comment.

Looking at javascript speed tests, it seems it is going to be slower. The test reported here show that javascript in Firefox 3 is 20% slower compared to Firefox 2, and 4.4 times when compared to Opera. http://celtickane.com/projects/jsspeed2007.php
I don't know where Mozilla is spending all its money.

Quick test on a 2GHz Pentium Mobile:

Firefox 2.0.0.6: 1250ms
Konqueror 3.5.6: 750ms
Opera 9.23: 360ms

Looks like the relationship between firefox and opera is about the same as in the measurements given on the benchmark page.

without units. 281ms per what? Apparently a bunch of tests listed on http://celtickane.com/projects/jsspeed.php

Now my question is, how significant is ~500 ms for these tests? All I care about is how long it takes to load a typical webpage I surf, and for me, Firefox seems almost instantaneous for most pages. "Smacks silly" my be an overstatement.

Firefox is and always has been faster (uses less CPU) and more efficient (uses less memory) compared to IE and even compared to Opera.

Maybe more so than IE but not Opera. Search "Firefox Opera speed test" on Google. A good selection of links, especially this one, showing Opera's speed. I also challenge the memory usage - someone up the page mentioned having Firefox on all day and using 103 meg. I have 15 tabs open right now, it's been open all day and my memory usage has barely exceeded 50 meg. No memory leaks, nothing.