18 February 2010

Unofficial browser speed tests

And by “unofficial” I mean tests made on my home computer. Now that launched the first beta version of 10.50, I was curious to see if the advertised speed gains were real. My first impressions after installing the new beta were that it was indeed faster, but nowhere near .

For my tests I used two online benchmarking tools, SunSpider and V8, which were also featured in the official Opera tests. I tested the current browsers installed on my computer (well, except Internet Explorer): Chrome 4.0, 3.6, Opera 10.50 and an older, portable version of Safari 4.0. I opened each of them separately for the purpose of the test and closed the program after getting the results. There were no other major software running in the background, except for the antivirus.

The results were a little surprising, while confirming my initial impressions of Opera’s speed: in SunSpider, Opera came in last, with a time to complete the test about twice higher than Safari. Firefox was second and Chrome first, more than 3 times faster than Opera. I ran the test twice for each browser to convince myself it wasn’t a one-time glitch, but I’ve got very similar results.Browsers speed test SunSpider results

It was also interesting to watch the memory consumption of browsers in Task manager during the test: Chrome and Firefox were extremely steady, completing the test with about the same amount of occupied RAM as they started; Opera’s memory use on the other hand has increased during the test to more than twice the original value. To be fair, it also released a large portion of it after completing the test. Since it’s only a beta version, there is still time to work on these memory issues.

The contrast between Chrome and the other browsers is even more evident in the V8 results. Opera and Firefox are here neck-in-neck, but in last place, with a score about 8 times lower than Chrome. In this test, contrary to SunSpider, larger numbers represent better performance. The results in the following graph are averaged from three runs of the test for each browser.Browsers speed test V8 results

Since V8 is built to test Chrome, the large lead here was to be expected. However, Opera’s test results and other independent tests ranked it very close to Chrome in both of them, so how can this discrepancy be explained?

Different hardware will naturally give different results, but I always thought the hierarchy will be similar, that the fastest browsers will be fastest on a large range of system configurations. Of course, my system is pretty old, with a low amount of RAM, and I think this could explain the very different results I got for Opera. The high performance probably requires a large amount of RAM, so when it’s not available, the browser simply slows down, like a runner gasping for air. Chrome on the other hand proves it can run just as fast on older hardware, so it could be the ideal browser for netbooks and tablets.

Post a Comment