A high thermal load allows us to gauge cooling capacity effectively. The Foxconn NF4SLI7AA-8EKRS makes overclocking easy, pushing ultimate heat from our already hot P4 530 Prescott core CPU, since it’s overclocked from 3.0GHz to 3.8GHz at 1.52v.

We used 3DMark2001SE in looping mode for one hour while nVidia’s nTune Monitor tracked temperatures. We estimated full-speed (3000RPM) noise at 34dB, but a check of Zalman’s site covered noise levels only at 2600RPM and 1400RPM. Putting all three numbers on a scale, 34dB seems to fit, so we’ll assume for now that our estimates are fairly close to the actual numbers. Below are our initial results as found in a rather warm 26C room.

Correcting for differences in room temperature allows us to compare previous results from other coolers. We did this by subtracting the difference in room temperature between tests. Making things even more difficult for the mid-sized CNPS8000, we compared it to several huge coolers, Zalman’s Fatal1ty FS-C77, Silverstone’s NT-06, and Thermaltake’s Big Typhoon.

Anyone complaining about our temperature correction method will be pleased to know that a different method--based on percent difference in Kelvin—puts the CNPS8000 temperatures even lower by about one degree. The CNPS8000 simply wins the cooling contest here!

For cooling-to-noise ratios, the CNPS8000 wins at 5v and the Big Typhoon takes a close second. We’ve always estimated the Big Typhoon between 21dB and 24dB, and we're still unsure about how Thermaltake determines the advertised noise on its product. In our tests, Themaltake’s claimed “16dB” product is actually noisier than Zalman’s cooler, which advertises a minimum of 18dB.