Living the startup life in Atlanta, GA.

There’s a lot of factors that you have to take into account when you pick a javascript framework. Picking the right tool is a balance of developer familiarity, the task at hand, and performance.

Measuring a framework’s performance is tricky. The reality is that no benchmark will account for what your app will do in real life. Making the task harder is that each framework has different features, making a straight head-to-head comparison impossible. However, that doesn’t mean we don’t need to know the relative speed of different frameworks.

I found myself in that position recently. I attempted to gauge the relative speed of Mootools, JQuery, Prototype, and Appcelerator. The rational behind picking these four frameworks is that I understood them enough to create a simple test.

the benchmark

The bechmark that I wrote is a rails app and available on GitHub. The README describes what it does:

This pretty simple test attempts to measure the time it takes to load the framework and do some basic tasks.
The first is to load the page with the W3C css
selector page
pre-loaded into a hidden div. It then dynamically loads in the Georgia 2008 Presidential election
results, again in a hidden div.
Next it loads in the
Georgia 2008 Senate results and
attaches a
click event to all the red elements. Finally,
it loads in the Georgia 2008 election
summary.
All files are served from the local server to avoid mistimes due to network latency.

Because of the way Appcelerator works, instead of using a selector to attach an event, I’ve created
a different version of the “dynamic” html file with “on” expressions set for the red elements. I
feel this test is more fair, since there’s no reason to select through the dom when Appcelerator
processes each element by default.

This is a rails app, so you should just be able to download it, then run script/server.
Navigate to localhost:3000 and you’ll be good to go.

Feel free to fork and improve as you see fit.

results

I ran the benchmarks on my Windows XP vm, with 384MB of RAM running within VMWare Fusion on a 2.2GHz Macbook Pro. These are the times. Appcelerator is included twice. The first is an internal patched version. The second is the latest 2.2.1 release. Columns in Red indicate that an error occurred while running the test.

conclusions

Safari is incredibly fast. If all your users used Safari, the framework you use doesn’t really matter. I’m pretty surprised with how fast Prototype was in IE. It was the only one that was quicker in IE than in FF. Mootools and JQuery had similar performance, right up until it came time to load a very large file in IE. No framework was the clear winner on speed.

Hopefully this benchmark provides a little guidance on which framework to pick. If all you’re doing is catering to Safari users, feel free to go with something like Appcelerator. If you need to target IE users or the general web population, you can’t go wrong with Prototype, Mootools or JQuery, unless you’re loading in very large files, at which point Mootools is out. Otherwise, pick for which one you like better.