Voltages and lifespan

[superstition]

I've read so many times about chip lifespan being reduced by excessive voltage, either in conjunction with excessive heat or without.

However, aside from just staying within the manufacturer's suggested range, is there any systematic data that can help people know what level of risk they're taking? Vague comments about lifespan aren't very helpful. There have been systematic tests of a lot of things, but I've yet to see someone try to do this with voltage. I realize that every chip is different, which is why such tests would have to use more than one of a CPU and more than one motherboard (given the differences in vdroop, etc).

 

[Idontcare]

The manufacturer does. Qualification of a new technology node requires thorough characterization of acceptable voltage operation and temperature in order for devices to meet their spec lifetime.

It's not a binary thing, you don't magically cross a threshold whereupon your CPU dies or lives. All voltages will kill your integrated circuits (be it CPU, memory, GPU, etc) eventually.

It is merely a question of time...how long at your operating voltage and temperature is your product expected to live?

The temp and voltage specs are engineered to ensure the CPU has a minimum 10 year lifespan.

This is the physics of the situation. It's not really up for debate whether voltage or temperature kills your CPU any more than it is up for debate whether gravity makes things fall to the ground.

 

[myocardia]

Idontcare is completely right. If you want an answer in more technical terms, here's what our local guru said on this subject, in another thread asking about electromigration, which is what you're technically asking about. It's the second post in the thread, located here: http://forums.anandtech.com/me...ey=y&keyword1=CTho9305