Concerning Flex you are true i made a mistake. I run the test again and the results are almost the same. Even if we can discut about how flex is build don’t you think that only the final result matter. The load is immediate.
For Ext, thank you for the precision. I made the same test using the loadData method. By the way the results are very close to the first one.

With FF the loading times for 10,50,100,500 and 1000 rows are:

0,034
0,071
0,180
0,718
1,858

With IE :

0,029
0,144
0,248
1,025
2,271

That doesn’t surprise me because this test try to evaluate the first load end renderer of the grid. So even if i use a method that change the data you have to build all DOM elements of your grid.

No i don’t made the test with Rialto-GWT. The purpose was not to test Rialto here. I just add the results with rialto js to be honnest. We plan a new release of rialto GWT sooner. At this moment i will test it…..

1. You are only timing building the data array for Flex, not grid rendering. Since Flex is built in Flash and rendering is asynchronous, I would imagine it will be very difficult to get an accurate result even if the call « grid.dataProvider = gamesData; » was included within the timed block.

2. You are completely reconfiguring the Ext JS grid instead of just loading new data. Data in Ext JS is completely separate from the visual component and the loading of new data is done through the Store API (similar to the Flex DataProvider). If all you want to do is test the load/rendering time of new/additional data, I would recommend instead of reconfiguring and recreating the grid from scratch every time, you use the loadData() method of the store. For example:

var d =new Ext.data.Store({data:oData,reader: myReader});
grid.reconfigure(d,grid.getColumnModel());

grid.getStore().loadData(oData);

http://extjs.com/forum/showthread.php?t=34539