Is tec worth it?

[lel4866]

Looking at the overclocking databases at vr-zone at overclockers.com, I've noticed very few tec setups.

Clearly the consistently best results are from the Prometeia (which, btw, calls into doubt the need to cool other chips on the board in any extreme way).

Now, it would seem that a good non-chilled water/tec system could come close to prommie temps, but I don't see actual results in terms of max overclocks.

Adding a tec and power supply to my existing water cooling project would be about $200, which would bring the whole project to about $400, which is less than the prommie II.

Any comments on why people don't seem to get much better overclocks on tecs that with just water. 5% just ain't worth it to me.

Larry Lewis

[bigben2k]

If overclocking is your objective, you're already passed the point of cost benefit

I wouldn't put any numbers on it, but there is a benefit to TECs, if you can put aside all the work involved to set it up, the cost of the parts, and the running cost (your electricity bill).

BTW, it's also possible to combine a TEC with a phase change unit.

[sevisehda]

There seems to be plenty of super-cooled comps(CPU below 0C). Promethia, TEC, ect... I've only done straight H20 cooling, so far. Is there a point to bringing the temps so low?

Why cool a CPU to -50C? Wouldn't it run just as fast and just as stable at 0C or 20C? Does the super cooling reduce the resistance and allow it to throttle higher?

If you can't tell I'm interested in getting a good length explanation or link to such an explanation, thanks.

[lel4866]

Some possible answers as to why chips work better at lower temperatures:

1. Gate propagation delays oftem go down with temperature.
2. Noise often goes down with temperature.
3. Trace resistance decreases, and therefore noise margin increases with lower temperature (go low enough and its resistance is 0! Everything stops working as we know it).

Larry Lewis