There are consistent and reliable trade-offs between FSB, latency settings and voltage.
If you choose a 4:5 divider instead of 1:1, you would have to loosen the timings IF - I say IF -- you have tweaked them already to the tightest stable settings. I doubt that you have done that, or changed them from their SPD stock values. You might also be able to drop the memory voltage a notch @ 4:5. The results would be a wash with the 1:1 settings and tightest latencies.
I can "measure" these differences in synthetic benchmarks -- the only "objective" measure of comparison which nevertheless may not reflect real-world performance. But I can gauge real-world performance subjectively, and compare maximum achievable game scores at different settings. For that, I have to trust the Everest memory benchies as a reliable relative indicator of "improvement."
If you don't understand my jargon in the first paragraph, I mean, by "trade-offs" -- that an increase in FSB will require looser timings (latencies) at various thresholds as you go up the FSB ladder. Tighter latencies may require higher voltage, even as you also drop the FSB a tad. Conversely, if you lower the FSB, you can tighten the timings. But at any particular FSB setting, tighter timings require higher RAM voltage, and lower RAM voltage requires looser timings.
