After reading Seth Weintraub’s post on how the upcoming versions of Safari are blisteringly fast I decided to download the latest nightly (WebKit r30123) and check it out for myself.
To check the different browser versions I used the SunSpider JavaScript Benchmark.
The results were pretty amazing – on Vista the performance times came in at:
Internet Explorer 7 -Â Â 66,870.6ms
Firefox 2.0.0.12 -Â Â Â Â Â Â 34,121.0ms
Firefox 3.0b2 – Â Â Â Â Â 29,293.6ms
Safari 3.04 -Â Â Â Â Â Â Â Â Â Â Â Â Â Â 21,930.4ms
WebKit r30123 -Â Â Â Â Â Â Â Â 9,094.2ms
While on OS X:
Flock 1.08 -Â Â Â Â Â Â Â Â Â Â Â Â Â Â 30,476.8ms
Firefox 3.0b2 – Â Â Â Â Â 10,863.4ms
Safari 3.04 -Â Â Â Â Â Â Â Â Â Â Â Â Â 13,534.0ms
WebKit r30123 -Â Â Â Â Â Â Â 5,720.0ms
That’s pretty spectacular performance – and seeing as I use Safari quite a bit on my iPod Touch, I may just have to switch default browsers for a while to see how I get on with Safari Webkit!
UPDATE – post updated with results for Firefox 3.0b2 on Vista
I’d be interested to see how Firefox 3 compares to the new Safari on Windows. It’s good to see Mozilla and Apple are working on increasing the performance of their browsers rather than bloating them with unnecessary new features.
Good point Joe, I will download Firefox 3.0b2 to the Vista box and update this post later today.
Thanks for the suggestion.
I’d actually prefer to see the speed comparison on pages without javascript (meaning the test you use can’t be used), and without content loading from remote sites. Are the tests just showing that one javascript engine is worse than another or that the page renderer is faster.
A lot of the delays (on blogs in particular but elsewhere too) I find are due to various javascript doo-hickies and even moreso from loading external content. Disable / block those features and a lot of sluggish blogs suddenly speed up 🙂 YMMV of course 🙂
A nice benchmark, Tom. I had a play with the Webkit nightly build and was impressed with rendering times no end.
@Hostyle, would Slickspeed be a more acceptable test?
What hardware were you using?
@mj Good question. The OS X tests were done on a MacBook Pro with 2.16GHz Intel core duo chip, 2gb ram running 10.5.1
The Vista tests were done on a Sony Vaio SZ3 Intel Core 2 CPU T7400 @ 2.16GHz 2.17GHz and 2Gb ram running Vista Ultimate.
Okay…so both devices were Core 2 Duo recent machines running at the same GHz and with the same amount of RAM.
Have you tried the Opera 9.5 beta.
they use a new Java script engine.
I’m not sure what it is but i know its a great improvement on
previous versions. which was always much faster than firefox and
IE.
Maybe you should give that a run also and see how it pans out.
Slickspeed … still javascript based though so much of a muchness i would say. What I mentioned is rather hard to automate – more of a “I’d like to see” than “I expect you to do” thing.
I think the point still stands (after reading the second test): javascript speed != page render speed. Render speed is much more important than javascript speed, as long as the javascript on a page is running after load. So while an interesting set of results it wouldn’t make me change browser 🙂