Browser performance, Chrome compared.
September 4th, 2008 by 
Afraithe There is a lot of buzz going on now about the new Google Chrome browser, and a lot of different testing of its performance against other browsers.
We take performance very seriously, thousands of ppl use TinyMCE every day, it is important that it works smooth on the browsers, not just loading performance, but also the performance of all the stuff that’s going on under the hood when you work with the editor.
So, the tests we have seen online so far have been very wide spread, showing insane performance improvements, but what are they testing? Is there any real-world implications of these performance tests?
TinyMCE uses DOM operations a lot, we have our own test bed, and these are the tests we are running.
- Serializer - This tests the cleanup functionality, a lot of DOM operations.
- Style parser/serializer - Tests a lot of string and array handling.
- URI - Tests a lot of regexp performance, converting URLs back and forth.
And these are the results.
The performance is measured in milliseconds, lower is of course better.
Summary
We could not install MSIE 8 beta on the same machine as MSIE 7, so no tests for that right now. The testing machine is a Windows Vista 64-bit, 4GB ram, 2.66gz Dual Core. Tests where performed with nothing other than the testing browser running.
Safari comes out on top of performance, with Firefox close behind, however we did not see any improvements when enabling the new JIT compiler in Firefox 3.1 Nightly build.
Results vary a bit up and down, but you can perform these tests yourself at this location.
http://tinymce.moxiecode.com/js/tinymce/performance.html
These tests try a lot of different things at the same time, using the TinyMCE API, the performance improvements that the browser developers have been bragging about does not seem to be there, at least not in the areas interesting for most developers (DOM operations).
Posted in Development, Software
September 5th, 2008 at 09:27
[...] Nachlesen gibt es hier den englischen [...]
September 5th, 2008 at 18:27
[...] Meldungen nehmen kein ende, nun hat sich die Firma hinter dem TinyMCE dran gemacht und einen Vergleich der Browser gestartet. Habe ich eben auf der Infogurke gefunden. Hier gab es auch einen feinen Link zu einem Video der [...]
September 7th, 2008 at 22:16
You should send this data over to the crew at Google Chrome. As TinyMCE is the most important javascript wysiwyg editor around, they need to do something about this!
September 8th, 2008 at 10:13
They probably already know, the “headline” news about the browser being “super fast” is just a show they are putting on to get more press covering the news about this new browser.
And it probably is faster than other browsers, in the most optimized areas, just not in the real-world examples, DOM operations etc.
September 11th, 2008 at 09:46
I am sad, cause it means its not so super fast
September 17th, 2008 at 04:27
This article is the most extreme example of retardation I have ever seen.
Let me ask you a question - if you take an stock html rendering engine (Webkit) and mate it with a superfast javascript engine, how fast will DOM operations on the resulting beast be?
Answer: pretty much the same as the stock renderer mated with a standard javscript engine.
DOM operations and DOM speed come from the renderer - it doesnt matter if you use fast javascript, slow javscript, c or anything; the speed of DOM will pretty much be the same.
September 17th, 2008 at 22:35
What version of Opera was tested?
September 18th, 2008 at 08:44
[...] Les détails sont disponibles sur leur blog. [...]
September 18th, 2008 at 09:50
@damien: The first test (serializer) is the only one using DOM operations. The other to do simple array manipulation.
Also since both Safari and Chrome is using the same engine (WebKit) shouldn’t the performance be the same or better in Chrome it clearly isn’t.
We are not testing the V8 engine alone there are already a lot of tests doing that in this article we are testing the overall speed of Chrome and the result is that it’s slower than Safari in a real world application.
September 18th, 2008 at 09:52
@asdf: 9.60 don’t know the build number.