Opera 10 final is out, Chrome just turned one year old, and Firefox 3.5 has settled into stability. It's time once again to break out the timers and speed test today's web browsers.
Like all our previous speed tests, this one is unscientific, but thorough. We install the most current versions of each browser being tested—in this case, Opera 10, Chrome's development channel 4.0 version, and the final Firefox 3.5 with security fixes—in a system with a 2.0 GHz Intel Centrino Duo processor and 2GB of RAM, running Windows XP.
We use Rob Keir's tiny timer app to time each browser in the common ways that leave users waiting: a "cold start" right after boot, a "warm start" after already having run once, and waiting on eight tabs to load up. We run each test three times and average the results, tossing out any obvious irregularities.
We also run each browser through Mozilla's Dromaeo JavaScript testing suite, which itself is an aggregator of Apple's SunSpider suite, Google's V8 tests, and a few unique items. These tests run each browser's JavaScript interpreter and engine through a series of situational exercises. Mozilla's Firefox Director Mike Beltzner told us that JavaScript test results can be compared to horsepower, and can be tweaked by engine developers—still, a holistic solution isn't in our hands at the moment. Anyone feel like coding up a fake Gmail that captures test results?
Finally, we take a Windows Task Manager measurement of how much memory is being used at startup and after those eight tabs are loaded. The eight tabs are the same as in the last set of tests—basically, each browser's home page, and then the Google home page, Lifehacker, Gizmodo, and YouTube thrown in for good browser-dragging measure.
On with the results!
Test 1: Boot-up and page loading—Winner: Google Chrome!
Normally Chrome rules the roost on start-up times, both warm and cold. (Incidentally, the development channel version I loaded on XP turned my mouse into an hourglass for a few seconds on every new load, both cold and warm. That could indicate some issue with the system, but I was running Chrome on a plain vanilla, just-installed copy of XP Service Pack 3.) Chrome 4 is, however, a lot faster on my Windows 7 RC system, but that's not what this test was about. So, here's the results for the three latest browsers:
For comparison's sake, here's a combined ranking of all the browsers from our last test and this round. As shown here, Chrome's stable version is the overall champion:
When the browser's open, though, Chrome's 4.0 version handily beats even its stable predecessor, so perhaps that's a willing trade-off. You'll only load your browser a few times each day, but tab loading is something of a by-the-minute affair for many of us. Here's the eight-page loading for this round:
And the aggregate results from the last two rounds:
Test 2: JavaScript—Winner: Google Chrome!
Not much new to report here, except that Chrome 4.0 shows gains in JavaScript speed over its stable release. Opera 10 and Firefox 3.5 perform the same as their last beta releases—reassuring for our testing purposes, at least. Here's the three current contenders:
And the full leader board:
Test 3: Memory Use—Winner: Firefox!
Google Chrome's 4.0 version is definitely grabbing more system resources and revving its engine at startup to deliver performance. Between Chrome 2.0 and 4.0, there's a notable increase in memory use in multi-tab situations, and a smaller gain at startup. Firefox continues to be impressively efficient on first load. Here's how the newest trio fared:
And the combined results of all the modern Windows browsers:
Firefox and Opera show themselves to have fairly consistent and usable betas, while Chrome reveals that its bleeding-edge development trunk really is different from what the safe-as-cotton users are getting. Anything catch your eye in this multitude of minutiae? Have your own suggestions on what you'd like to see tested for the next round? Tell your referees in the comments.
