I've always maintained that a lot of the hyped-up technologies, like multi-core processors and high-speed DDR2 RAM, are basically overblown in terms of their effects on gaming performance. The most important thing by far is the graphics card. Forget synthetic benchmarks – what matters is real-world performance. So, I did a little test today.
I went to Guru3D.com, who did a review of the 8800GTS and 8800GTX. Their test system was an Intel E6800, 680i motherboard, Corsair Dominator DDR2 RAM (CAS 4 @ 1142mhz), and the cards in question. Using the in-game test for FEAR (all max settings, 4xAA, 16xAF), in 1280x960 they scored an average of 112 fps for the stock 8800GTX and 82 fps for the stock 8800GTS.
I ran the same test... my system is an AMD 4000+ @ 2.94ghz (single core), OCZ platinum DDR 500, A8N32-SLI deluxe, and an 8800GTS clocked at 660/2000. Using the exact same settings (all max in-game settings, 4xAA, 16xAF), I scored an average frame rate of 101 with the texture filtering set to "high quality" in Forceware. Using the default "quality" setting, which is probably what Guru3D used, I scored an average of 102 fps.
Ironically I think a dual-core or a higher-speed CPU would have helped me score even higher, only because my max frame rate was capped around 255 fps. An E6800 would raise the max frame rate somewhat, which would raise the average slightly (it wouldn't affect the minimum frame rate, which is GPU limited). However at 100+ fps we're well past what any monitor can physically display or what the human eye can perceive anyway, so it's kind of pointless.
What About Higher-Resolution Gaming?
It took some tracking down, but I found a second FEAR test, this time at IT-review.net, done with 8xAA. For some reason 8xAA tests in FEAR are hard to find. Increasing AA has the same effect on performance as increasing the resolution, so I thought it'd be a good comparison.
Their test system was 680i, Intel E6700, OCZ 1000mhz DDR2 RAM and an 8800GTX. At 1280x960, all max, 8xAA and 16xAF, they averaged 81 frames per second. To my surprise, my overclocked 8800GTS actually beat them, scoring an average of 88 frames per second at the same setting.
Both of these systems are cutting edge, and mine is considered yesterday's news. In any benchmark those systems would blow mine away. Yet with my overclocked 8800GTS, I scored 20fps higher than than Guru3D's stock 8800GTS score, and 7fps higher than IT-review's 8800GTX score! All those expensive components – 680i board, high-speed low-latency RAM, high-speed dual-core processor – didn't have much of an effect on their performance. Granted this is just one game, but when you consider the cost difference between my system and theirs, it gives you an idea about where your priorities should be when you build a gaming rig.