Hardware.info makes history of sorts by posting a roundup review of a whopping 103 fans. The authors deserve kudos for the sheer volume of their task, though SPCR laments the relatively low usability of their results. First, the fans are measured for noise from just 10cm (4 inches) away. It is extremely difficult to convert this data into SPL at any meaningful distance; no one is likely to be closer than about 50cm from any computer fan. Secondly, the fans are tested at only 2 voltages, 12V and 7V, two data points in a fan curve that can range continuously from as low as 3V to 12V. Thirdly, the non-linear, mostly poor correlation between cooling (ie, temperature drop) and CFM is never touched upon.
Still, the article is worth a read, and the measured fan data could be useful for some. The 30 [email protected] chosen for "fan efficiency" CFM is not badly chosen; this translates to perhaps 18 [email protected]
SPCR forum discussion