The mystery of the relationship between Nehalem CPU and DDR3 voltages continues.
We keep hearing that a Bloomfield will become a Doomedfield CPU if you pump up the DDR3 by more than a bit.
This article will get you up to speed on events, but we’d like to make two observations.
First, given that these chips are supposed to show up a little more than a month from now, there’s been few examples of people getting engineering samples and testing them out.
Second, maybe this is just me not looking hard enough, but for the relatively few examples of reported Nehalem overclocking we do have, they seem to all be increase-the-multiplier overclocks, not FSB overclocks. We know that the multiplier in Nehalem is going to be a bit flexible, allowing for an extra multiplier or two. However, this report indicates Nehalems apparently have up to at least a 31X multiplier defined, which hardly seems necessary given the expected speeds of the chip.
There’s also a cryptic statement from the article linked above that “knowledgeable overclockers” will be pleased by whatever it is Intel did.
Add all this together, and it seems like Nehalem overclocking could well be multiplier rather than FSB-based. Do I have much confidence in that prediction? No, but keep in the idea in your head when you see other advance testing in the weeks ahead.