reply to Krisnatharok
Re: Nvidia's FU to overclockers I can see where they are coming from. As CMOS scales down the chips are going to be getting more and more sensitive to running beyond spec voltages. As someone familiar with the technology at a low level the voltages that I hear about people applying make me cringe. I have looked at reliability data for CMOS smaller than 65nm and it doesn't take much to reduce the lifetime from years to weeks.
Still this is a bit heavy-handed. A more balanced approach might be to attach a disclaimer that over-volting voids the warranty. That has practical issues though like requiring some sort of detection mechanism, and will still probably get bad press from people who think they should be allowed to overvolt and still have a warranty (which, as someone who designs integrated circuits, is ridiculous). In the short term I can see why they went with this, the detection for those who overvolt is that they will have to physically modify the card. Perhaps there is a way that such modification could be easy built in to the design, but irreversible, on enthusiast cards?
That MSI story is interesting. With the manufacturers in a pinch to differentiate themselves with 'stock' overclocks, I can understand the motivation to push it. One could say that MSI is the one the hook for the warranty coverage and bad press, but I don't think nvidia can ignore that a bunch of cards dying either in warranty or out of warranty is not good for their reputation either.
Most people are not going to seek out let alone understand the root cause of a product failure. I think back to the failure of so many GeForce 8400M/8600M mobile chips (they were basically all ticking time bombs). While nVidia was squarely to blame there, laptop manufacturers shouldered alot of the blame in the minds of consumers. I couldn't help but laugh when someone talked about how HP/Asus/Dell/Apple was junk because it died right right out of warranty, and how they were buying Apple/HP/Asus/Dell instead this time even though they all had the same problems with those chips. (I call this the 'failure roulette' where people switch companies even though the chance of failure is roughly the same). On the same note people with dead MSI overvolted cards could swear off nVidia and switch to AMD.
said by pnjunction: You even see it with the Intel die shrink from 32nm to 22nm--Ivy Bridge CPUs don't OC as well as Sandy Bridge.
I can see where they are coming from. As CMOS scales down the chips are going to be getting more and more sensitive to running beyond spec voltages. As someone familiar with the technology at a low level the voltages that I hear about people applying make me cringe. I have looked at reliability data for CMOS smaller than 65nm and it doesn't take much to reduce the lifetime from years to weeks.
If we lose this freedom of ours, history will record with the greatest astonishment, those who had the most to lose, did the least to prevent its happening.