Friday, September 17, 2010

Chrome 6 beta versus IE 9 beta - Showdown!

So, I downloaded the new Internet Explorer 9 beta, which is only available for Windows Vista and 7 (sorry XP) and had mixed feelings. It looked good. It promised much. But I still found the weird webpage scrambling problems, extremely slow performance and was disappointed. Maybe this was the reason it did not get such a wide acclaim online. But anyway, that didn't stop me from taking IE 9 beta and Chrome 6 beta into the ring.

Browser versions and specs:

Here's the version numbers for either browsers. Click the images for better views:

This is Chrome 6.0.472.59 beta...

...and here's "Windows" Internet Explorer 9.0.7930.16406.

And this test was carried out on a PC running Windows Vista Home Premium 32-bit, with the following hardware specifications:

CPU: AMD Turion 64x2 TL-58 1.9 GHz dual-core.
RAM: 2 GB.
GPU: nVidia GeForce 6150 Go.
Hard Drive: Hitachi 150 GB.

So the performances of these browsers can vary on your machine, at times.

Test summary:

So, to test these browsers, I had an unmodified IE9 and Chrome 6 with no add-ons that affect the page rendering. I ran the SunSpider JavaScript benchmark, the CSS tester from Nontroppo (run locally), the Acid3 test and some page load tests that I'll explain later.

The results were graphed and they are discussed later on.

Test Results:

IE9's Acid3 performance
I didn't expect this. Windows IE9 beta failed Acid3. More than that, Chrome passed it. Now, if you're wondering: What is this acid thingy by the way? Here's your answer:
Acid3 is a test page from the Web Standards Project that checks how well a web browser follows certain selected elements from web standards, especially relating to the Document Object Model (DOM) and JavaScript.
When successful, the Acid3 test displays a gradually increasing percentage counter with colored rectangles in the background. The percentage displayed is based on the number of subtests passed. It does not represent an actual percentage of conformance as the test does not keep track of how many of the subtests were actually started (100 is assumed). In addition to these the browser also has torender the page exactly as the reference page is rendered in the same browser. Like the text of the Acid2 test, the text of the Acid3 reference rendering is not a bitmap, in order to allow for certain differences in font rendering.
Chrome's Acid3 performance

The above text is from Wikipedia. Precisely, Acid3 is a "universally accepted" benchmark for a browser following major web standards. A good browser must pass the test. But IE9 did not. I really expected this new, revamped web browser to pass it. Anyway, it reached 95/100, and that's good still.

Google Chrome, as I told earlier, passed it. Everything was perfect.

The SunSpider results! This is my favorite, everytime. This is also a widely accepted benchmarking service, for JavaScript rendering. The browser is tested on various kinds of JavaScript elements, and the results are displayed in the time taken to do it.

Easily put, Chrome 6 was about 1.49x faster than its competitor. This graph shows the results for the two browsers. The smaller the bars, the better performance.
Chrome won, but by a fairly small difference. If IE8 was in the competition, it would be very, very tall. You can see the results of a relatively old IE8 for SunSpider in this comparison that I had done earlier. So, this is pretty good.

The CSS benchmark at Nontroppo was done with the page saved on the hard drive. Both browsers were tested five times. The mean of the five data was taken and compared. See the graph below (the smaller bars, the better):

Again, Chrome turned out to be the winner, though IE was pretty close, compared to its earlier versions' performances.

Finally, I tested the page load performances for the browsers. This can vary by large margins for different people, as I tested this one with only two websites. I used this stopwatch to measure the time it took for the two browsers to load the American Yahoo! homepage and the English Wikipedia Main Page. This was measured with an empty cache for both the browsers. After that, I refreshed the pages to see the cache performances for the two browsers. I got some very astonishing results. But first, here's the graph:

What I called page load (warm) is the refresh time, with the cache
See the trends? Look again. Probably you're looking at the graph without any interest and you want me to explain. See, I read your mind.

Fine. IE9 was slower than Chrome, as expected, but again, much much better than its previous versions. Here's the astonishing part. The cache performance of this new browser was way better than Chrome's. IE and Chrome loaded Y! in 16.56 and 15.26 seconds respectively, but IE refreshed the page in 5.2 seconds flat, while Chrome took 6.22 seconds. Wikipedia's a light website. Still, Chrome did the loading in 5.24 seconds, while IE took 19.37 seconds to render the same thing. Although IE took so long, the page refreshed faster in this browser. It did it in 1.59 seconds, flat while Chrome took another 3.87 seconds to do that.

Also, Chrome loaded the page steadily, displaying element by element and IE loaded much of a page at once and displayed it all together. For some people, like me, this appears to be slow, as I have the habit of using Chrome. But for some other people, this may appear faster, as the page displays quickly. So, using the browser depends on your preference.


Well, Chrome was surely the winner. Anyway, I had expected IE9 to be better, and it didn't live up to it. Still, the app is a newly released public beta, as it is still in the works. It will get better as it goes to stable. I hope that happens. 

Chrome will always remain one of the best browsers. No matter what. Oh yes, and it will remain on the list of free programs a geek should have.

Want to see more fun stuff? See this YouTube easter egg, or if you are more of a constructive person, check out my complete guide to editing and writing for Wikipedia. And there's always the archive to check out for more fun stuff.