there is more to it

i asked Kris to comment on this over at OCX but he didn't yet

i don't actually think advocating to leave loadline calibration off is a better idea than having it on because you are effectively going to be giving the CPU more volts overall without the loadline calibration. i gave some extreme examples over there for LN2 benchmarks and the need to give between 1.9 and 2v to CPUs to get max clocks...........you think about it now............gigabyte have very large droop on X38 and in fact it's bigger than it was with their P35 boards which means that you will have to increase your vcore in bios by as much as 0.075v to compensate for the droop in windows. Now think about idle and load cycles with and without loadline calibration and different vcores set and used and the fact vcore fluctuates enormously without controlling droop

once you have a good thought i believe you will arrive at the same conclusion i did and that is that controlling droop on boards that exhibit a fairly large one is more beneficial and will probably save your CPU from dying more than doing without it

how was the data measured and there are no timeframes to spikes so are they just rough graphs, did they do the testing themselves or is some data from other sources........there are a few other things but IMO the aricle didn't make me change my mind about controlling droop