As I reported earlier, I installed Mozilla Firefox 4 from the PPA. On the side, I have been playing with the development versions of Google Chrome for some time. I use the dev versions because it isn’t my primary browser and its fun to toy with the bleeding edge. The latest dev for Chrome is version 12.0.712.0 as I write this. BTW, Firefox 4 had over 5K downloads a minute and totaled over 15.85M in just 48 hours. Congratulations, Mozilla!
After playing with Firefox 4 for a few days, I decided to take a shot at Micro$oft’s speed and HTML5 compatiblity tests. They intended those to highlight Internet Explorer 9, but I thought that it would be fun to see how Firefox 4 compared with Chrome 12 dev. I actually intended to run only the FishIE Tank speed test, which is very cool, but later decided to expand the comparison.
Before testing, it’s important to ensure that both browsers are properly set to use hardware acceleration. This helpful page tells you how. Firefox 4 worked fine without the tweak, which didn’t make any difference. Chrome, however, needed its experimental 2D hardware acceleration feature manually activated. The Compositing setting no longer exists in version 12.
Lastly, the underlying hardware isn’t all that important since the comparison was run within minutes of each other on the same box. But for completeness, I’m running 64-bit Ubuntu 10.10 Maverick on a homebuilt powered by an Intel Q9650 Core 2 Quad overclocked to 3.3 GHz with 8 GB of DDR2 SDRAM. My video card is an EVGA GeForce 8800 GTS w/640MB of DDR3 SDRAM. I ran each browser separately, with only Shutter and Krusader running in the background (besides my standard configuration). I used the 64-bit versions of both browsers, along with 64-bit Flash support.
On to the results…
The IE Fish Tank loaded fine in both Firefox 4 and Chrome, but it behaved a bit differently in each. The fish swam smoothly and consistently in FF4, and FF4 maxed out the frame rate at 60 fps with 20 and 50 fish. At 100 fish, FF4 bounced between 50 – 60 fps. Here’s 20:
Chrome 12 dev, OTOH, never actually stabilized at any setting. With 20 fish, it would “hover” between the mid-40s and mid-50s, then pause and drop to around 20. Here’s about the top end:
Here it is after a particularly bad pause:
That was the worst case I saw. Normally it would drop to around 20 fps. It seemed like odd behavior. Chrome really wasn’t much worse with 50 fish, but really choked above that.
M$ had a cool graphics accelerator test that shot bubbles out of a champaign bottle. Although intended to celebrate IE9, it proved an appropriate accolade for the new Firefox 4. Firefox 4 pegged at 60 fps, which bested Chrome 12’s 29 fps:
Next I moved on to an HTML5 demo, which was interesting but not particularly exciting. The Power Meter supposedly gages how easy a browser is on your battery. Firefox 4 did quite well:
Chrome supposedly hammers a battery if you believe M$:
Both Firefox 4 and Chrome 12 did fine in displaying all the other HTML5 demos which required manual interaction. I didn’t encounter any issues in either browser.
Lastly, I tried three more of the Speed Demos besides the FishIE Tank. Again, I surmise that these were clearly compiled with the idea of showcasing IE9’s strengths while glossing over its weaknesses. That said, these tests were less consistent between Firefox 4 and Chrome.
One for which I almost did not take screenshots was the Flying Images. The screenshots were difficult because the images’ speed and position followed the mouse pointer, sometimes obscuring the score. The pictures don’t do justice to the speed at which the images spun. With 144 images spinning at max speed, Firefox 4 held a steady 60 fps, but dropped to bounce between about 55-60 fps with 196 images, spending most of the time at 60 fps:
Chrome 12, OTOH, hung down around 46 fps with 144 images and 32 fps with 196:
If this were a “bash Chrome” post, I’d stop there. However, Chrome 12 did much better at two other speed tests. Firefox 4 didn’t do quite so well on the Preschool test that recited and displayed the alphabet. In fact, Firefox 4 didn’t produce any sound at all for that test. It finished in 11.679 seconds:
Chrome 12 played the sound of children reciting the alphabet and finished in 7.175 seconds:
That said, that Chrome 12 tab locked up after the test, with some video corruption spilling beyond the browser itself. It presented a dialog asking to close that tab, but Chrome closed the tab itself before I could get Shutter up for a screenshot. The Chrome browser continued to function fine after closing the offending tab, so the sandbox worked. That was the only issue I saw with the Chrome 12 dev, but I don’t read much into it because of its raw developmental status.
Last was the coolest test besides the FishIE Tank. Galactic simulates flying around the sun with the inner planets. Firefox 4 did particularly poorly with this one at 191 kps, producing a jerky display and not playing the background music. I did try it in safe mode with all the extensions disabled, but it didn’t make any difference:
Chrome 12 did a little worse with 155.51 kps without its 2D acceleration enabled, but it did play the music:
With 2D acceleration enabled, Chrome 12 kicked some serious butt at over 545 kps:
Chrome 12 had no trouble playing the music in Galactic.
That concluded my goofing around for today. It looks like 64-bit Firefox 4 more than holds its own against the newest 64-bit Chrome 12 dev version. Both, however, still have work to do. Competition will keep both on their toes, and I look forward to more good stuff from both Mozilla and Google in the browser wars.
As for IE9, who cares? It only runs on Micro$oft’s newest expensive operating system. Did I mention that Firefox 4 had over 15.85 million downloads in its first 48 hours? And that Linux is free, as in free beer?