How good the cards seem to be is much dependent on who does the reviews.... I found THG's review somewhat ...er shall we say "curious".
According to the settings we chose on a per-game basis, picked to maximize visual quality at playable frame rates, the GeForce GTX 660 Ti is close to, if not slightly slower than a Radeon HD 7870. However, after comparing lab results, the outcome of our testing appears tied to the way we picked settings for each game, likely taxing the 660 Ti's memory bandwidth more than less-demanding options would.
Perhaps our settings favor the Radeon cards. Perhaps theirs favor the GeForce-based boards. And maybe the most real-world outcome lies somewhere in between. But, I believe the truth is that GeForce GTX 660 Ti performs within 5%, plus or minus, of the Radeon HD 7870
I don't think I have ever seen a review couched in quite such a dramatic way.
OTOH, Techpowerup comes up with a completely different scenario. In their test suite (Alan Wake, Batman Arkham City, BF3, BattleForge, Civilization 5, Crysis. Crysis 2, Dragon Age II, Hard Reset, Metro 2033, Stalker CoP, Starcraft 2, Shogun 2, Skyrim, WoW, Diablo II, Max Payne 3, Sniper Elite V2), we see the following:
The 7870 finishes with a total of 1236.6 fps and delivers that at a cost of $0.23 per frame.
The 7950 finishes with a total of 1348.2 fps and delivers that at a cost of $0.24 per frame.
The 7970 finishes with a total of 1532.8 fps and delivers that at a cost of $0.30 per frame.
The 660Ti finishes with a total of 1387.4 fps and delivers that at a cost of $0.22 per frame.
The 660 Ti finishes about 3 % ahead of the 7950 but does it for $30 (10%) cheaper