Originally Posted by conrad.maranan
I've noticed a a few people having to drop their CPU multiplier in order to run Prime95 stable. There are others, including myself, that can pass Memtest forever at the multiplier we want, but fail to boot into Windows or maintain some sort of stability during Prime95 using a high multiplier.
I've noticed the above only on Winchester CPUs. I'm starting to fall under the impression that most Winchesters have weak on-die memory controllers to begin with. If there are others that run an FX or 130nm chip with great results, please share your data with us in order that we may draw some guesses/generalizations with the Winnies (as if we don't do enough speculating already).