Notes On Our Noise Level Tests

All measurements were taken roughly 23 cm away from the AGP slot, with the motherboard in a horizontal position. For the duration of the measurements, the CPU and power supply fans were deactivated, and no drives were connected to the motherboard, to minimize ambient noise. For cards that use different fan speeds for 2D and 3D modes, we were only able to measure start-up noise level, since other measurements would have required us to reconnect the hard drives (to run an OS). This, in turn, would have made it impossible to take any clear readings. Please note that the values given here cannot be compared with other tests or even noise levels quoted by the manufacturers, but are only consistent within the context of this test!

When judging noise level, the frequencies involved play a major role. For example, a fan that produces only 35 dB but has a "singing" or "whistling" undertone can seem louder than a 45 dB fan that only gives off a normal murmur.

Another point to consider is that some manufacturers use the same fan on different cards with different GPUs. A fan that is virtually silent may suddenly become audible on another card. On a card that produces more heat, the fan will be run at a higher voltage and consequently at a higher speed, which, in turn, means more noise.

Perception of fan noise:

• 25-35 dB: practically silent
• 35-50 dB: inconspicuous to quiet
• 50-55 dB: clearly audible
• 55 dB: loud to unpleasant

At mid-range frequencies and noise levels, the human ear perceives a jump of 10 dB as a doubling (or halving) of volume, hence the big jumps at lower noise levels.