Originally Posted by tehmaggot
I've been fairly concerned with this as well. I find it hard to believe these smaller chips are that much more resilient against voltage. Consider that Intel's 90nm chips were considered safe up to about 1.65v, their 65nm chips up to about 1.5v to 1.55v and their 45nm chips from 1.36v to 1.4v.
I can't help but believe these new 32nm chips will be even more delicate, and yet these review sites are pumping 1.5v+ through them and posting high clock speeds.
This is not a coincidence. If you want to understand more you can read about it here http://www.overclockingwiki.org/inde...iquid_nitrogen
I can say with confidence that the safe threshold for these chips is likely to be below 1.4V, possibly even < 1.35V based on Intel's specs. Electron migration becomes a big problem as the manufacturing process shrinks which also directly implies lower tolerance to voltage.
In one sentence, do not expect to be able to run these chips for 24/7 pumping 1.5V through them (IMHO even 1.4V is ambitious for 24/7). This is likely a publicity stunt to differentiate the SB chips from LGA1156/LGA1366 in order to boost sales. It's actually laughable how many of these early reviews do not compare chips clock-for-clock yet they claim performance boost. Of course you get more performance with higher clock on the same basic architecture.
you can actually refer to i7 970 and i7 980x overclocking to see that it supports what I am saying. They are both manufactured using the 32nm process and are both (while very good clockers) less voltage tolerant for 24/7. The biggest difference is and will be power consumption, clock for clock. 32nm chips are just more efficient overall.
32nm i7-900 series datasheet seen here
lists MAX Vcc at 1.4V. Expect i7-2x00 to be the same.Edited by dejanh - 1/5/11 at 12:09am