The Mozilla/WebKit Arms Race
March 20, 2008 – 7:27 pmBiologists talk about a phenomenon called the evolutionary arms race. Cheetahs, for example, only survived if they were fast enough to catch the slowest gazelles. Gazelles, on the other hand, only lived to produce offspring if they could outrun the fastest predators. These are powerful evolutionary forces, and as a result both species have adapted to become almost comically fast.
Something similar is shaping up in the browser space with WebKit’s vertiginous rise. Since WebKit swept onto the scene as the engine powering Safari, AIR and the iPhone browser, the heat has been on Mozilla, and they’ve responded with major improvements to their rendering engine’s memory consumption and performance. I was speaking to some developers at a major Ajax shop yesterday, and they were blown away by the performance boost they’ve experienced since upgrading to Firefox 3.
Now the WebKit camp has struck back with the release of Safari 3.1:
Safari loads web pages 1.9 times faster than IE 7 and 1.7 times faster than Firefox 2. Safari also runs JavaScript up to six times faster than other browsers, and is the first browser to support the latest innovative web standards needed to deliver the next generation of highly interactive Web 2.0 experiences.
Apple pointedly fails to provide figures for performance relative to the latest Firefox 3 beta, but according to the same Ajax shop, WebKit is back in the lead by about 10-15%. It’s only a matter of time until Mozilla responds in kind; innovations like Tamarin Tracing (just-in-time compilation using the virtual machine donated to the Mozilla Foundation by Adobe) will eventually take their JavaScript performance to the next level.
And what about IE? My Ajax contacts tell me it isn’t even on the map. They even claim that, for their app at least, the new IE8 beta is even slower than IE7. With two outstanding, standards compliant, open source browser implementations vying for supremacy, Microsoft had better get its act in gear or risk going the way of the dodo.
9 Responses to “The Mozilla/WebKit Arms Race”
You can try Safari 3.1 vs. a Firefox beta or nightly yourself, on WebKit’s own SunSpider test collection at http://webkit.org/perf/sunspider-0.9/sunspider.html to see how the JS performance compares. I think you’ll find the results interesting.
Apple’s numbers are on a beta version of Safari (?) and rely on iBench 5.0 (in case you have IIS and the Helix Streaming Server around to test with), so you’ll have a hard time reproducing them yourself.
By shaver on Mar 20, 2008
It’s perhaps most instructive to compare Firefox nightlies to WebKit nightlies. I believe the lead has switched quite a few times in the past few days.
By Maciej Stachowiak on Mar 20, 2008
Indeed, it’s been good times.
By shaver on Mar 20, 2008
Well there you go, a proper arms race indeed! Someone should set up a website that pulls down the respective nightlies and charts the progress on a day-to-day basis. Hmmm…
By Matt on Mar 20, 2008
The best thing IMHO is that speed is another compelling argument for “normal” users to switch. Firefox has gained a lot of market share among normal users because it was safer than IE, now it also has the speed advantage
By Vincent on Mar 20, 2008
Most exciting to me is that Safari supports referencing SVG images from CSS!
The Firefox bug has 80 votes:
https://bugzilla.mozilla.org/show_bug.cgi?id=231179
By Chris Hubick on Mar 20, 2008
Standards support for stuff like SVG and CSS is certainly benefitting from this arms race as well, but I agree with Vincent that performance is more of a banner issue that is likely to mean something to “normal” users.
By Matt on Mar 20, 2008
(I thought this was in my previous comment, but I guess not.)
Comparing nightlies is fun and easy, but people shouldn’t expect Mozilla to be taking major optimizations right now. We’re in stabilization mode for FF3, and we aren’t going to be jamming things into a release that would risk site breakage or stability — other groups, obviously, can choose other paths, depending on how they prioritize benchmark performance vs. other things. There are lots of excellent optimizations to come in Mozilla’s JS (some with patches waiting for after FF3 ships), and we haven’t even started dipping into the Tamarin-Tracing bag of tricks, or doing any interesting compile-time bytecode optimizations.
But if you’re going to compare nightlies, you might find it especially interesting to compare builds on mobile devices, like the n810…. I’m afraid someone forgot to tell those machines that Mozilla’s supposed to be the slow one.
By Mike Shaver on Mar 22, 2008
Just watching the Acid 3 progress and speed of the Webkit nightlies is impressive! Webkit is sitting in the mid-90s, in terms of compliance, and moving swiftly. How long has it taken Mozilla just to get Acid 2 compliance? I do not hold out hope that Acid 3 compliance will be present for 6 months or more.
I think it is fair for Apple to compare only against released browsers. Note that they did not include IE8 either. When Mozilla and MS actually release their browsers, currently in development, then that is the sign they consider them complete. Until then we would just hear arguments that they are not code complete, debugged, or optimized. Released products are the best mark for measure.
By Colin Scroggins on Mar 23, 2008