ongwt.com - Google Chrome and GWT : micro benchmark

Wednesday 03 September 2008

Google Chrome and GWT : micro benchmark

Par dgirard le Wednesday 03 September 2008 12:22 - Browser

I have done a micro benchmark to compare Chrome/Firefox 3.0/Firefox TraceMonkey (3.1) and IE.

Algorithm	IE6	FF3	FF3.1	Chrome	Comments
Computation for(i=0 to 10millions) result+=i;	2800ms	945ms	40ms	200-2000ms	Strange : with Chrome results are not stable, each time the bench is restarted (not reloaded) the result is different
Building a table : for(i=0 to 1000) addALineOnTopOfATable();	1900ms	885ms	480ms	160ms	Chrome is faster
Moving an image : for(i=0 to 10000) moveTheImage();	1450ms	5308ms	912ms	450ms	Chrome is faster

Results are in milliseconds.

It was run on this machine (Windows XP SP2) :

1.Thursday 04 September 2008 11:35 par Tanguy Bayard: Another benchmark can be found here : <a href="http://news.cnet.com/8301-1001_3-10030888-92.html" title="http://news.cnet.com/8301-1001_3-10030888-92.html">http://news.cnet.com/8301-1001_3-10...</a> It uses Google provided benchmarks, and the results are slightly different.
2.Thursday 04 September 2008 03:10 par Zaccret: Should take a look at RAM usage, as one tab stands for one independent process. On my machine, two tabs in Chrome use the same amount of memory as 5 tabs in Firefox 3.
3.Thursday 04 September 2008 03:25 par Zaccret: Should take a look at RAM usage, as one tab stands for one independent process. On my machine, two tabs in Chrome use the same amount of memory as 5 tabs in Firefox 3.
4.Thursday 04 September 2008 08:56 par Plazma: Chrome is fast. The ram usage really shouldn't be a concern as you can always buy more as its extremely cheap. Chrome focuses on what matters. I still primarily use FF but thats only because I'm waiting for customization and add ons.
5.Saturday 06 September 2008 06:44 par keitai1948: Maybe loop unrolling can explain the fluctuation? Maybe the effect of loop unrolling cannot be sensed in the two last tests?