Power Control settings sat at +20% at all times.
Yes, it would seem that there is only a 1-2% increase with the added shaders at same clock speeds, which one can barely notice. As a matter of fact I was under the impression that unlocking actually gave me less performance than my overclocked 6950 until I took the time to put it through some tests.
These are the results I got:
A) stands for 6950 @ 880/1325
B) stands for 6950 @ 6970 stock
Dirt 2 Benchmark
A) Min Fps: 44.5
Max FPS: 49.5
Score: 4560
B) Min Fps: 43.9
Max FPS: 50.9
Score: 4766
4.5% increase
3DMark 11
A) P4562
B) 4620
1.2%
FFXIV Benchmark
A) 3398
B) 3474
2.2%
Heaven 2.1 cold run
A) 1039
FPS: 41.2
Min FPS 17.8
Max FPS 80.1
B) 1064
FPS 42.3
Min FPS 18.1
Max FPS 82.4
2.6%
Crysis Benchmark
A) Min FPS 26.34
Max FPS 40.50
Average FPS 33
B) Min FPS 25.69
Max FPS 41.80
Average FPS 36.5
10.6% increase
These were all conducted at 1080p with the exception of 3DMark 11.
Now I could point out to you that Haven is tesselation limited, both cards have two tesselation engines, and so theres very little real change there.
All the other results you posted except one are BENCHMARKS, they aren't designed to be absolute graphical performance benchmarks, but system benchmarks. Essentially speaking in the one game you ran, you got over a 10% increase, though that will be a best case scenario most likely.
Dirt 2 is the next closest to a gpu benchmark, that showed 5%, FF as I understand it, is quite cpu limited but I'm not sure on that one, 3dmark sucks balls and has a margin of error run to run of 1-2%, meaning you could be showing best run vs worst run and seeing results much closer than normal.
Haven as said is tesselation limited but still shows an ok increase, that will be mostly clockspeed as opposed to shaders.
As with anything, some games are bandwidth limited, some will be shader limited and some clock speed limited, some a perfect balance of all 3 and scale great when all things are increased.... Crysis.
My main gripe was you somehow using 1fps and 1% interchangably, unless you see a 1fps increase from a base 100fps, you aren't talking about 1%, AT ALL.
A 580gtx is only 10% faster than a 6970 in several things, if thats only 2fps in some cases does that not really count, sure it does.
Faster is faster, slower is slower, 2-10% performance increase, with the two most real world examples being between 5-10%, for FREE.
and i take it you did the Smartdoctor install, enable overclocking range enhancement, uninstall?
I had to re-download ATIWinFlash, then set the .exe to Windows Vista and Admin credentials (for all users). This allowed the app to launch and I could save the default BIOS.
