Can't Touch This! A Comparison of 46 CPU Coolers

Laboratory Tests Up Close & Personal - CPUs Maxed At 100%

We had already noticed during our last comparative tests that not many cooler models were a good compromise between the necessary cooling performance and an acceptable noise level. In fact, our readings showed that the majority of the coolers don't cool optimally. The testing platform consisted of an AMD reference board (AMD EVTS-BX-004) that was furnished with an AMD-750 chipset from the same manufacturer. All our readings were based on an AMD Athlon 1000 that worked with a 100 MHz front side bus. We used Windows 98 SE as an operating system platform. In order to keep CPU capacity use at 100% as much as possible, we ran the well-known program Seti, which processes data packets from the Internet. At the same time, we ran Flask Mpeg to create an MPEG-4 film and tested its playback. This ensured that the processor was constantly running at maximum capacity, which in turn ensured that heat flow was maximized.

Cooling Temperature

We carried out all the measurements in our Munich laboratory at an ambient temperature of exactly 24°C, two degrees higher than during the last test as a consequence of the warmer season. We calibrated the results in order to ensure comparability with the last measurements at 22°C.

Noise Level

We carried out our sound measurements in our newly conditioned sound measurement box, which simulates an almost completely soundproof environment.

Fan Speed

We measured the fan speeds of the new 23 coolers using a stroboscope. Each rotor was marked in order to facilitate the measuring of the speed.


The coolers were weighed using a precision weighing machine as is used in chemical laboratories to weigh materials.