No, he took a screenshot. Hes using stock speed on video card and CPU. Like I said EVERY single person with a GTX480 or Radeon 5970 had extremely poor scores when compared to those with Radeon 5870s.
There's a bit of uncertainty about whether or not multiple GPUs are getting optimum performance with the benchmark. People who have actually checked their core usage on their Crossfire/SLI setups are saying that they're seeing activity on both cores that is consistent with the benchmark using both (though perhaps not as well as one might hope). That helps to explain why 5970s aren't posting such great numbers...it's basically a crossfired 5870 pair on one card.
As for the GTX470/480s, there's one key thing to remember and that's that ATI has had a lot longer to tweak and refine their drivers for various different circumstances and applications. Fermis have the hardware to post better numbers than a 5870 but they need the drivers to be tweaked and refined to consistently deliver the performance the card is capable of.
So far, we're seeing that for the low res scores, CPU is more important than GPU. For the high res scores, it's somewhat the other way around but CPU is still factoring heavily into the mix.
There are other things to consider when evaluating performance, especially under such a demanding application. Bus speeds and RAM speeds are important when you're moving around vast amounts of data and performing lots of intensive calculations on the fly. You could have a great processor and a solid GPU but if you stuck them all together in a budget motherboard, your numbers aren't going to be the same as someone who went with quality from the ground up.
Edit: This might be a long shot, but I'd like to see the temperatures on the 480s. It almost makes me wonder if they're not throttling back due to overheating.
Edited, Jun 18th 2010 8:47am by Aurelius