Next-gen browsers: How fast is fast?

A little over a week ago I took a look at Firefox 3.0 Beta 4 and I benchmarked this latest release against the current browser lineup.

Yesterday Apple released Safari 3.1 so the browser landscape has now changed. It’s time to see how Apple’s latest browser fits into the big picture.
Apple makes some bold claims relating to Safari. Have a read of this:
The fastest web browser on any platform, Safari loads pages up to 1.9 times faster than Internet Explorer 7 and up to 1.7 times faster than Firefox 2.

And it executes JavaScript up to 6 times faster than Internet Explorer 7 and up to 4 times faster than Firefox 2. What does all that mean for you? Less time loading pages and more time enjoying them.

To put these results into context you have to also read the small print:
Performance measured in seconds. Testing conducted by Apple in March 2008 on a 2.4GHz Intel Core 2 Duo-based iMac system running Windows XP Professional SP2, configured with 1GB of RAM and an ATI Radeon HD 2600 with 256MB of VRAM. HTML and JavaScript benchmarks based on VeriTest’s iBench Version 5.0 using default settings. Testing conducted with a beta version of Safari; all other browsers were shipping versions. Performance will vary based on system configuration, network connection, and other factors. [emphasis added]

Now if I’m going to pull Apple up over anything it’s over the use of iBench. iBench 5.0 is a benchmarking fossil. It was released November 2003 and it’s safe to say that a lot has changed since then, especially as popular websites are now loaded with JavaScripts and make heavy use of AJAX. Benchmarking a modern browser (and OS) with an old tool seems pointless to me and the results should be taken with a liberal pinch of salt. But it does highlight a problem that when it comes to browser benchmarking tools, there isn’t much to choose from. iBench might be old (and obsolete) but there’s not much else to take its place. We either have to rely on benchmarking against live sites and put up with all the variables introduced by server load, lag times and local traffic bottlenecks, or alternatively rely on synthetic benchmarks which have little in common with the real world.

There are dozens of tests out there that look at page loading times and page rendering speeds, but I don’t feel happy using any of them, and as a result I’ll leave out any discussion of page loading and rendering.

I’m going to confine testing to two tests:
* SunSpider JavaScript Benchmark
* ACID 3

Testing will be carried out on the same hardware and software platform as previous tests were carried out on.


Technorati Tags: ,,

Related Posts by Categories

Widget by Hoctro

Enter your email address:

Delivered by FeedBurner


Source Code