Hi there, thanks for your email:
Obviously in our test we compared with Chrome 4, and not v5. We have been comparing with version 5 in our testing, but the difference was minimal (although Chrome 5 is slightly faster in some areas). I've seen sunspider on several computers, and it's almost always been faster with Opera 10.50, although the build that we were testing is not the exact build that was actually released.
Just now on my laptop here (not the test machine we used at Opera for this purpose) I did a quick test. I have the following specs:
Intel Core 2 Extreme X9100 @ 3.06Ghz, 4GB RAM, Windows 7 64 Bit
I used chrome 5.0.322.2 (the latest as of today) and the final release of Opera 10.50 beta that is now public.
The results I got were:
Sunspider Error Range Extremes
Chrome 5 dev 462.8 474.8
Opera 10.50 491.0 480.0
(The error range was +/- 2.6% on chrome, +/- 2.3% on Opera) Taking these account in a manner that would be most beneficial to Opera gives results that are almost equal... very very close. Granted, they could also work the other way around, but either way, the results were much closer than what you have. I can't say why there is such a big difference in your test, but I'm left wondering if there might be an architectural difference on AMD processors that put us at a disadvantage, although I don't personally have the knowledge to confirm this. I'll ask around.
It does appear that Chrome may have made some improvements that bump their performance a bit in SunSpider, so we will keep our eye on it as well.
I have made a graph of my results. You might choose to ignore the second set of bars, but I just did that to show that the difference is pretty negligible when factoring in the error range - granted Chrome still wins in my quick benchmark, just not by much at all.
I hope this is informative,
J.D. Lien