Here's an article talking about memory speeds for DDR2: http://www.tomshardware.com/reviews/tight-timings-high-.... There's a similar one for DDR3, but I can't seem to find it right now. Basically, the conclusions were that the DDR3-2000 kits had too slow timings and didn't help except in very specific situations like WinRAR. In games, the differences were 1-2%, and it actually depended on the game as to whether faster timing/1333 or 2000 kits resulted in better frame rates.
This is definitely one of those areas where if you have to ask, you don't need it (even if you're a hardcore gamer). The differences in performance are small, and unless you have very specific needs (I'm a photographer dealing with 200 photoshop files a day and I need my RAM to do x), you're not going to be able to even tell the difference. If it were me, I would go with the faster timings on a 1600 kit.
9 CAS @ 2000mhz = 4.5 microseconds actual latency
8 CAS @ 2000 = 4.0
7 CAS @ 1600 = 4.3
6 CAS @ 1600 = 3.7
So an 8 at 2000 is actually slightly better than 7 at 1600, as far as timings. But at the end of the day, no, none of this will make much of a measureable difference, much less a noticeable one. Changes in CPU speeds will drown out any of these differences in memory, so just pick a rough "best" of the ones that are actually available and don't sweat it.