[Tenacious Dyl]

Alright... I would have to say I know a lot about computers in the way of how software works, and how the general build of a computer is, and usually I find myself comfortable w/ my knowledge... however something has been bothering me for a long time.

It is the issue of cooling a processor / ram / harddrive to produce faster results. I understand that this is true, and that cooling a processor especially can improve performance.

My question is what other factors are causing this? Electricity has less resistance when going through warm/hot wires than cold. So why... why would *increasing the resistance* against the electronics actually improve performance?

Yes this is more of a science question... but I am wondering. People have commented about the g5 having such crazy cooling power..... how does this work?

Thank you.
Dylan I.

[djohnson]

I see what you are saying, see if this helps.

When you ru a modern processor like the G5, you are producing an over abundance of heat. The cooling system does not reduce the heat output to room temperature, instead it lowers it from the 200+ degrees it would run without cooling down to a more managable level. You do not want to cool the chip to much though because at a certain point, well below 0 deg C, the chip will not work as you pointed out. About room temperature would be ideal I guess. Hope this helps!

[Catfish_Man]

Originally posted by Tenacious Dyl:
Alright... I would have to say I know a lot about computers in the way of how software works, and how the general build of a computer is, and usually I find myself comfortable w/ my knowledge... however something has been bothering me for a long time.

It is the issue of cooling a processor / ram / harddrive to produce faster results. I understand that this is true, and that cooling a processor especially can improve performance.

My question is what other factors are causing this? Electricity has less resistance when going through warm/hot wires than cold. So why... why would *increasing the resistance* against the electronics actually improve performance?

Yes this is more of a science question... but I am wondering. People have commented about the g5 having such crazy cooling power..... how does this work?

Thank you.
Dylan I.

As far as I know, cooling a processor does not actually make it go faster. What it does do is allow you to set it at a higher speed without it melting or otherwise messing up. This is why people who overclock their comps often have crazy cooling setups. Also I suppose cooling it would contract it slightly which might do something... can't see it doing much though.

[Spliffdaddy]

The enemy of electronics is heat. The best possible operating temperature would be just above absolute zero.

Heat is a byproduct of inefficiency. Friction of electrons creates the heat in a CPU. Ideally, a CPU should only use enough electricity to switch transistors on and off - creating no heat in the process. That isn't possible with today's technology.

Without going well past my field of knowledge by explaining electromigration and other boring stuff - let's just accept the notion that too much heat will cause (temporary) short circuits within the CPU core which lead to all sorts of strange behaviour up to and including an outright stall.

CPU manufacturers expect that their processors will be used in a "room temperature" environment - and they include specific (& required for warranty) recommendations for cooling the processors. So, when you buy a 2.0 GHz Celeron processor you can be certain it'll run reliably in any "room temperature" environment. Just to be safe, manufacturers typically under-rate the potential speed of their processors. In other words, that 2.0 GHz Celeron would likely operate reliably in an environment that was 25% hotter than 'room temperature'. AND it would likely run at a higher speed than 2 GHz, reliably, if the ambient temperature was less than 'room temperature'.

Since most folks don't use a computer in a high-temp environment, most CPUs can operate at a higher speed than they were labeled and marketed. The CPU can be made to run faster by manually changing the multiplier ratio and/or the system bus frequency. This is called overclocking - and it's a wonderful hobby I've participated in for several years.

Overclocking is simply taking advantage of the potential improvement in performance that exists in most any processor. Keep it cooler than the manufacturer anticipated, and chances are excellent that it'll operate at a higher clockspeed than they promised it would...if you force it to, that is.

While you might spend a lot of time fondling and caressing that $430 Pentium4 3.4 - Intel didn't waste 2 seconds on it. It probably came off the same slab of silicon as that $170 P4 2.8GHz...perhaps they were even neighbors. See, CPU manufacturers, while offering a range of speeds in a given platform, usually manufacture them ALL at the same time in the same plant on the same machine. They don't 'un-refine' their process to make 2.8GHz versions after they make enough 3.4's, in other words. The process is continually refined and the processors continually improve. While there's certainly some variance between the quality (called 'yield') of each CPU on that silicon slab - the difference isn't all that much. A smart overclocker knows which versions of a CPU 'model' share the same cores. Buying the least expensive model, like a P4 2.8, can get you the very same core you'll get in the flagship 3.2. Overclocking either one would achieve similar results. So (if your goal was to overclock) you save about $250 if you get the 2.8GHz version.


When you get tired of never having the fastest peecee in the world - look me up.

phase-change refrigerant liquid loop @ 30F removes heat from cascaded (stacked) 130watt & 85watt thermoelectric junctions (TECs) resulting in -50F delivered to the surface of the CPU.

it owns with 70's-era nixie numerical gas-discharge displays. geek factor x 10. Modern temp controller is usually hidden.