Newer Chrome, Firefox show speed improvements

The SunSpider test shows Tracemonkey-enabled Firefox leading the newest Chrome build in JavaScript performance--for now at least. Test results reported in seconds, so smaller is better.

(Credit: CNET News)

Correction 12:00 p.m. PDT: This report has been updated to reflect Firefox performance with the TraceMonkey JavaScript engine enabled, in which case Firefox is fastest at the SunSpider test.

With new beta versions out for Firefox and Google Chrome, I thought I'd see how things have changed when it comes to testing the speed of JavaScript, the programming language that powers many cutting-edge Web applications such as Gmail and Google Docs. The answer: both browsers made big strides, but Firefox still beats Chrome on one widely-used performance test.

When Chrome was released, I ran Google's JavaScript speed test on Firefox 3.0.1, the initial Chrome beta, Internet Explorer 7 and 8 beta 2, and Safari 3.1.2. Chrome led the speed test with an overall score of 1,851 and Firefox in second place at 205. A bigger score is better on this test.

Running the same test on the latest developer version of Chrome, 0.3.154.3, boosted the browser's score to 2,265--a 22 percent increase. And Firefox jumped 15 percent to 235. Firefox 3.1 beta 1. However, that test measures Firefox without its new TraceMonkey JavaScript engine enabled; a bug in TraceMonkey trips up the test by invoking a print dialog box. (There aren't any new versions of Safari or IE to test, though Safari likely will see a boost from its earlier score of 170 from the SquirrelFish Extreme JavaScript engine.)

In September, Firefox backer Mozilla countered Google's benchmark suite, spotlighting Firefox's superior results using the SunSpider speed test. Here, TraceMonkey works, and Firefox maintains its lead over Chrome and the others.

The newest Chrome beta is 22 percent faster on Google's own JavaScript benchmark. Firefox's performance score increased 15 percent--but that doesn't factor in TraceMonkey, because a bug breaks this test.

(Credit: CNET News)

I couldn't run SunSpider when Chrome was released because the site was out of commission that day, but it's up and running again now, so here's the latest results for the four browsers--and bear in mind here that a smaller score is best for SunSpider: TraceMonkey-enabled Firefox led with a score of 2,257; Chrome was second at 2,904; Firefox 3.1 beta 1 with no TraceMonkey next with 4,233; Safari 3.1.2 followed at 6,351; and IE 8 beta brought up the rear with 9,025.

Mozilla's Mike Shaver said the bug that impairs the Google JavaScript test--one reason TraceMonkey isn't enabled by default--should be fixed "soon." For those who want to try the their own tests, Tech-Recipes has useful instructions on how to enable TraceMonkey.

Why you should care
Why does all this matter? A few reasons.

First of all, JavaScript is widely used in innumerable ordinary Web sites, and Internet companies have found that even small improvements in Web page responsiveness increases user interaction with their sites. A snappier response is better for everyone.

Second, for the more avant garde, JavaScript powers many sophisticated Web sites and Web-based applications, endowing them with features such as drag-and-drop, pop-up dialog boxes. Faster JavaScript means companies such as Zoho, Google, and Yahoo can build more features into the Web applications and that users will find those applications easier to use. And these more interfaces are spreading to mainstream sites, too.

Last, on the programming front, JavaScript is vying with other technologies for building rich Internet applications. Microsoft steers developers to Silverlight, Adobe Systems continues to improve its Flash and Flex technology, HTML itself is getting more powerful. And of course there's the larger competition between Web-based applications and those that run directly on the PC, such as Microsoft Office.

A final note: The same benchmark caveats apply this time around, too. Your mileage may vary--my tests were on a dual-core Lenovo T61 with Windows XP. There are other performance attributes that affect Web browsing besides JavaScript performance. And even in the narrower realm of JavaScript, benchmarks like these don't necessarily represent the workloads that will have you pining for a faster machine.