During out testing of the GeIL ONE memory modules, we had to push the input voltage all the way to 3.0 volts - something that can't be good for their longevity

Based on our experience with all kinds of memory modules we can state with some authority that tightened timings are nearly always a better choice than higher clock rates. This show is being replayed in the DDR2 arena at this very moment, in fact: more and more vendors are pushing clock rates as high as they can make them go, and count it only a small thing that input voltages must rise apace. That explains the burgeoning support for DDR2 1066, which means true clocking at 533 MHz to the memory socket, in high-end motherboards nowadays.

But it really doesn't deliver much more performance in comparison to using DDR2 667 down at CL3-2-2-8. DDR2 memory modules have to be really high quality to sustain this kind of use, especially at extreme clock rates. Default timings for all DIMMs that guarantee DDR2 1066 support limp along at timings of CL5-5-5-15, or only slightly faster at best.

Vendors who build fast modules select their memory chips from big manufacturers such as Hynix, Infineon, Micron, or Samsung on the basis of associated clock rate and timing data. That overclocking up to DDR2-1066 or DDR600 requires definite and measurable increases in input voltage doesn't appear to enter into this equation. Above all, we are compelled to observe critically that measurable input voltage increases lead to premature memory chip failures.

But it's nearly impossible to predict when such failures will occur. From one day to the next, a defective memory chip can render an entire memory module unusable. Those who find value in long system lifetimes should therefore avoid radical overclocking. On the other hand, minor input voltage boosts seldom cause problems.

From this perspective, it's entirely debatable whether or not driving these GeIL DIMMs at such high clock rates is a good idea, especially if you want them to enjoy a long trouble-free lifetime.