[citation][nom]refillable[/nom]What's Up with reviews on other websites?In Techpowerup It trade blows with a 7950, not a 7870?In Hardwarecanucks It also trade blows with 7950, not 7870?And many more...[/citation]
What's up is differences in settings.
As mentioned on the last page, I tested Nvidias settigs and came up with better results for the 660 Ti. it can be summed up in two words: Tessellation, and 8x MSAA
First off, I don't use maximum tessellation unless it provides a visual benefit, and doesn't cause too much of a frame rate hit. In the case of the 660 ti and the games we tested, I found it fails on both counts. Nvidia and other sites crank up tessellation, and I believe this gives the GeForce a sizable advantage in a number of tests.
Also, I use 8x MSAA if the hardware can handle it. in the case of the 660 Ti, it can in many cases, so we enabled it. Our results suggest that this gives it a bit of a disadvantage compared to the radeons.
To be clear, I wasn't looking to cripple the 660 Ti. In fact, I chose the game settings based on what the 660 Ti can handle. I don't have much use for benchmarks that are so low they're unplayable (meaning the detail is cranked too high to be usable) or frame rates too high (meaning the hardware can handle much higher settings so we're not squeezing the most out of it). I try to find settings that give at least 30 FPS minimum and at least 40 FPS average during game play. Settings that are meaningful in a real-world scenario for folks who want the best looking visuals their graphics card can handle.
If you look at the benchmarks, you'll see that the 660 Ti handled every game, at every resolution, with smooth minimum and average frame rates unless (a) the game was so easy to render that even the highest settings provided ridiculously high frame rates or (b) the game was so tough to render at a certain resolution or settings that it dropped just below playable.
In any case, it comes down to this: I don't think the other sites have done anything wrong, their results are just as valid as our own. But it just so happens that the tessellation and 8x MSAA settings have a dramatic effect on 660 Ti performance relative to the Radeons.
It's all different sides of the same coin.
After working with Nvidia and trying out their settings there's only one major discrepancy between results that we can't figure out, and that's Skyrim. We get the same 660 Ti frame rate they did with their save game, but our 7870 result is much higher than theirs. We've replicated this on two of our test machines, with the 7870 confirmed as running at reference clocks, and we haven't found the cause of that discrepancy yet. But everything else seems to sync up.