Frequently Asked Question 06

Why does cooling change how fast the computer runs?

For most everyday overclocking setups, the cooling solution (air, water, peltier, etc.) is aimed at bringing the temperature down to where it normally would be had the computer NOT been overclocked. Computers generally run hot anyway, and the faster the clock rate, the more heat is generated. So, when you take a top-of-the-line system, that is generally hot to begin with, and then you overclock it, you are make a hot situation even hotter. And, if you don't remove the excess heat, you are sending your computer on the fast track to electronic ruination.

In this scenario, the "cooling" does not change how fast the computer runs, it enables the system to run at its overclocked speed without the heat damaging the sensitive components.

I suspect that most people who ask this question are familiar with all of the Liquid Nitrogen photos that are on the web, showing some unthinkable overclocking speeds (in early 2010 at least) of 7 GHz or more.

First of all, we should mention that those extreme clock speeds are attained for incredibly brief periods of time. In fact, on one discussion board (see the front page article for the complete story) an administrator had this to say about "how stable" the high speed systems were, in general:

Basically, the administrator said, the extremely-overclocked systems were able to report their speeds only for the duration of the benchmark. That meant, once the benchmark was completed, the ability to harness that speed ended. Since some of these benchmarks ran for less than one minute, the overclocked system was basically... not very useful at all for day-to-day use.

Back to the question. When a system is cooled down to extreme temperatures, such as below -100 Celsius (with the Cascade cooling system used by Buckeye, for example) the computer's clock runs faster because the cooling system is able to remove the huge heat load and lower the operating temperature of the CPU. When dropping to extremely cold temperatures, the physical properties of the CPU will change. For the most part, this leads to increased performance, but it can also lead to undesired effects.

When a CPU fails to boot at an extremely low temperature, we refer to this as a cold bug. Today, many manufactures are very interested in what we find out about how CPUs handle under extreme temperatures as a result of our experiments. As we send feedback to them, they either integrate our findings into their subsequent manufacturing process, or they design the CPU itself to run more efficiently.

For example, in 2009 AMD released a line of CPUs, the 955 and 965, that virtually had no cold bug at all. They tested these CPUs using a Liquid Helium cooling solution that dropped the temperature to -230 C (which is -382 F!) The CPUs were able to handle the extreme cold and broke the barrier of 7.0 GHz at that time. AMD has since adjusted their entire manufacturing process to produce CPUs that run better, faster, and cooler, for everyday use.

Previous question Next question