I was looking through the reviews,to see how much an FX8350 system would consume over a Core i5 or Core i7 system with a gaming load,not a stress test,rendering or video encoding test.
I only found two reviews:
http://www.legitreviews.com/article/2055/13/
http://hexus.net/tech/reviews/cpu/46985-amd-fx-8350-piledriver-cpu/?page=6
The difference seems around 40W at the wall looking at the figures. The first review uses an HD7970 and BF3 and the second a GTX680 and Batman:Arkham City.
First
Legit Reviews in the games they used (Which are all mainstream games) the FX-8350 was on a parr with the i7 3770K.
In fact the only thing that made the 3770K really standout was SupperPi and HiperPI, which it would anyway given that AMD do not support the x87 instruction sets those apps use, at all, the results are purely emulated.
Hexus is more interesting, again the games they used don't show the Intel particularly much better but a little worse that Legit reviews on the same games, But one game is particularly interesting.
Dirt Showdown, which i have, and have benched. according to Hexus my 1090T is 5% slower than the 2500K.
I benched
101 FPS (Avg) on that with my 1090T @ 4.1Ghz and 7870 as you see it clocked in my signature.
LtMatt benched 20% higher than me on a 7970 running at the same clock and on a 2500K running at 4.7Ghz.
LtMatt did that bench on 12.9 Drivers which are 10% slower than the 12.8 drivers i used.
That gives him 30% over me on 2500K @ 4.7Ghz and a 7970 @ 1200 / 1700, vs my x6 1090T @ 4.1Ghz and 7870 @ 1200 / 1500. The performance between his and my CPU is identical in that game.
I call Hexus result there horse manure, whether intentional or not, its utterly wrong and strongly suspect the FX-8350 result they put up is just as wrong.