This test is completely unfair.
1. Loading the same page 5 times and cut off fastest and slowest.
This is totally BS!!!!!!!!!!!!!!!!
if I create one stupid browser that render google.com in 1000s. Then when reload, it does nothing but repaint the same page / or show the same image on the screen.
Then, with your test, I get the fastest browser in the world, right. But the fact is I got the slowest (dump) browser.
Pls. use something we ACTUALLY do in real life. Open gmail / hotmail for example.
Run sunspider or use some benchmarking web site ( there are a lot out there).