Each combination of timings vrs clockspeed is going to affect bandwidth/performance differently.
It depends on how tight you can get your timings, or how high you can set your clockspeeds and how loose you need to set your timings to get there. Some setups will allow for tighter timings than others, so that would provide better performance, while others will allow for higher clock speeds and thus better performance.
The only way to know for sure is to try both settings and do the benchmarking on your own system.
I'm just starting to play with this kind of benchmarking on an Intel platform, but my own experience with S939 A64s was that there was little difference between max clockspeed w/ loose timings, or max "tightness" with lower clockspeed.
There is a balance to be found.... obviously running your mem just slightly faster than what allows you to run tight will not be as efficient as running the slightly slower clockspeed w/ tight timings.
Short answer is that there is no real way to tell which works better for your setup other than doing the benching yourself.
__________________ EVGA X58 3X SLI / i7 980X / 3X 4G Mushkin Blackline Frostbite / XFX Pro 1000W / EVGA GTX 680 SC+ 2GB / Corsair F120 / WD 6401AALS / TT Lvl 10 GT Swiftech MCP655 WC Pump / EK XT 240 Rad / 2X Scythe Ultra Kaze / EK Supreme HF / Primochill Myriad Dual Bay Res / Dell UltraSharp U2412M Sabertooth 990FX / PhII X4 965 / 2X4G Corsair XMS3 / EVGA GTX580 + EK Block / Corsair 120G Force GT Corsair TX750 / XSPC EX240 / Swiftech MCP655-B / XSPC Dual Bay Res / ? CPU block / Bitfenix Ghost / Samsung 931BF |