Does heat mean speed?

[Scudtech]

Everyone seems to agree that running your CPU cooler is better. But, according to my dad (who apparently knows something about computers since he build one from scratch with a Z80 ), transistors actually perform better when they?re hot, so by decreasing the heat of the CPU, performance actually decreases. Is this thought still valid, or have chips changed enough of the years so that it?s a moot (or inaccurate) point?

 

[Grunt03]

I have mine cooled down to 20 degrees, I can see the preformance
increase.

I would not want to be the one who will do it and be wrong.......

 

[rimshaker]

Absolutely not. Transistor (MOS) performance increases with decreasing temperature. It's an inverse relationship. The colder the temperature, the higher the electron and hole mobilities. Take one of the fundamental device physics expressions incorporating temperature, the Einstein relation: (D/u)=(kT/q) , where u is mobility and T is temperature in kelvins. The mobility is then u=(qD)/(kT). As you see, there's an inverse relationship between mobility (speed) and temperature.

 

[Evadman]

Great RimShaker, his head just exploded

Heat is bad for transisters. Cooler = better = faster. Cooler it is, the more voltage can be bused to pump those holes and electons. Your father probably just made a boo-boo. Happens to the best of us.

Lots of times to me tho

 

[Heisenberg]

All modern semi-conductor electronics can be destroyed by too much heat. One problem is that the resistance of the metals used to lay the interconnects in the CPU increase with higher temperature. It becomes a cycle - resistance causes heat, heat causes increased resistance, etc. Today's faster processors also means the "transistors" inside the CPU are switching extremely fast - faster switching means more power used, which means more heat produced. Bottom line is, adequate cooling is absolutely essential for more speed.

 

[rimshaker]

<< Great RimShaker, his head just exploded >>



Hehehh, they don't call this the Highly technical forum for nuthin...

 

[Scudtech]

Wow... thanks for the great info. I'll have to keep that formula stuff around for physics class next year . I guess my dad didn't quite understand the obsession of getting CPU temperatures down, so I'll make sure to run this by him and see what he says.

And, yes, my head is in quite good condition at this point... It might implode when there is a sure sign of stupidity... but that's another story...

 

[jhu]

aren't semiconductors better electrical conductors at higher temps? or do the impurities change that?

 

[Aquinas]

from what I know cooler = better, unless my EE prof was wrong...but then again he smoked pot b4 class...

 

[rimshaker]

Yes doping plays a significant role in electron/hole mobility. In terms of temperature, mobility is affected by scattering. There are 2 types of scattering that affect speed: impurity scattering and lattice (phonon) scattering. Impurity scattering is due to just that, doping of the semiconductor. In lattice scattering, an electron is scattered within the material due to vibration of the lattice (which increases with temp).

A complete graph of mobility vs. temperature would look like a carrot top ^ ..... With impurity scattering dominating at lower temps, mobility increases with increased temperature. At a certain point, lattice scattering dominates at much higher temps. Then it's the opposite, mobility decreses with increasing temperature. The Einstein relation I mentioned previously was assumed valid when lattice scattering is dominant... . the usual area of operation.