Originally Posted by MrAlex
Now that I'm running benchmarks today, the scores seem to be even. I'm wondering why when I first tested them I got such different scores in the benchmarks.
I haven't tried them, but since you've tried Dromaeo I wanted to offer my feedback on both and I've ran Google v8 since it's a known test.
Honestly I see too many variations with browsers these days and the scores people post on the Internet are just out of my reach 100% of the time, unless they have a slower computer (I still have a Core 2 Duo T9300 @ 2.5Ghz and 4GB of RAM with Windows 7 Enterprise SP1 x64, so I'm on a pretty fast computer for 90% of the tasks).
I have noticed something though and it's a bit interesting: in Dromaeo I, by habit, minimized the browser, as to not bother me. The result: CPU usage went down from ~50% to ~35%. I had to redo the test, just to not have variations.
Either way, if you look at the Dromaeo results you will see that for each test they provide a margin of error. If you look at my tests, the results for tcmalloc version have a higher degree of error (+/- 25% in some cases), while for the installer version the degree of error is smaller -> more consistent results.
In the end, what really matters is the rendering speed for me. The times it takes to press the button -> page finished loading. All browsers are very fast and the differences are minimal these days. Even IE9 is plenty fast nowadays for what most people need. It's weird that Firefox 12 is not really faster than Firefox 10 even 9, while Chrome improved a little. It's time for a redesigned browser and x64 support from Mozilla, as to improve from an already optimized base.