Why would you want to waste time on voltages? I'd much rather play with straps, multipliers, dividers, timings, etc.
Brute force cooling and endless fiddling with voltages were the main reasons for me to dump C2Q's. Current X58 system doesn't look much better from what Aigo says, it seems. And if you think about why, it's kinda pathetic.
You see, all these voltage fiddlings in i7 system involve QPI, and #1 reason why Intel developed QPI is da license. Even giving Intel the benefit of doubt and assuming performance was Intel's utmost concern, that would be for multi-socket servers, not single-socket desktops. I have no doubt that Nehalem desktop would perfom just as well using HyperTransport which is royalty-free industry standard. And Intel seems to agree! It will reportedly do away with QPI on upcoming LGA1156. (Though we'll have to wait and see)
If you remember Socket 939 platform, you surely remember how much HyperTransport affected overclocking or memory tweaking - i.e. near zero. On the other hand, on LGA1366 QPI (and related voltages) is tightly interwoven with CPU and memory, and is what dictates EVERYTHING when it comes to overclocking. Yet QPI itself has nothing to do with the performance of a system.. It's just an interconnect between CPU and the north bridge (which is a PCIe controller for the most part) Thus its main purpose of existence is to prevent anyone from making platforms or devices for Nehalem without Intel's permission. And You will not likely see any reviewer questioning why QPI is so important on X58 yet doesn't even exist on P55. But be that as it may.
I have absolutely no interest in wasting my time and energy on tweaking voltages, which contributes nothing to performance gain of itself. I'd leave that to engineers and take for myself multipliers, dividers, and timings instead at any time.