PDA

View Full Version : Is heat the only danger when overclocking?


Mad_Capr
06-29-01, 02:45 PM
I mean, what factors can actually physically damage your components when overclocking? Is it just the heat or can too much voltage damage a cpu even if it is running cool?

How high can you go with the voltage setting on a 1.333 and still be safe? The default voltage is 1.75 and with that I am running around 45 degrees under load. If I can keep the temp below 50 at max load, how high could the voltage go theoretically?

Bender
06-29-01, 03:40 PM
Heat is really the killer for processors. If you can keep the chip very cool you can use some very high voltages. Using a Vapochill as an example; if you had as much voltage as you wanted the processor would only go so high even if heat wasn't an issue. The core of a processor can only use so much voltage even if the core is way below freezing. If you had a cooling system that cooled the cpu to absolute zero the cpu wold still hit a brick wall and not clock any faster after a certain point even if more voltage were used. If you are water cooling you could try up to 2.11 or 2.25 to really stretch it. If your temps were 50 or below I wouldn't worry about a thing. Even if your temps are great using higher voltage physically wears out the chip faster. 2.11 is about all durons and T-birds like anyway so if you have the cooling system to handle 80-100 watts of heat give it a try.

Mad_Capr
06-29-01, 06:50 PM
Would71 (Jun 29, 2001 03:51 p.m.):
Heat is the immediate problem, and electromigration (sp.?) I believe is the long-term problem, sped up by heat. (don't expect a better answer than this until after I've finished my coffee) :D

so what causes (or accelerates) electromigration?

hooziewhatsit
07-01-01, 03:26 AM
If I remember correctly from the tutorial I watched at the company I last worked for, electromigration is when the electrons, which are racing around at crazy speeds, crash into the "walls" that contain them, and from there create little holes which can get bigger and spread to other wires, which is a very bad thing.

I'm not positive about this, but it makes sense at least....

Pitspawn
07-01-01, 03:49 AM
Electromigration is the process where computer components slowly fail over time due to build up of static problems.

Electromigration will ALWAYS occur to your computer components. It is merely a matter of how long it takes to happen. Here is a list of what causes electromigration...

1...Heat. As a general rule of thumb, in theory for ever 10ºC increase in temperature, the component will 'die' twice as quickly.
2...Voltage. Voltage DOES speed up electromigration, but nowhere nearly as quickly as temperature does.
3...Static. If you are charged and you touch an delecate component like a cpu, it will almost instantly half or completly kill the components life expectation.

4...Temperature fluctuations. Although it does not affect electromigration, it does still increase the rate that a component is likely to fail. It does this because if the metal in the components increase and decrease in temperature to quickly microfractures are formed. This can especially be a problem with sub-zero systems. This is why it is said that you should never run cpu-cooling programs with a sub-zero system after stressing your computer. The cpu may drop as fast as 2-3ºC when this happens. Not good. It is also the one main reason that computers can only take one or two boots using LN2 cooling.