Fan NoiseIn past years, users would accept everything for a little bit more performance. Nowadays, users are more aware of the fan noise and the power consumption of their graphics cards.
In order to properly test the fan noise that a card emits, we use the Bruel & Kjaer 2236 sound-level meter (~$4,000). It has the measurement range and the accuracy we are looking for.
The tested graphics card was installed in a system that was completely cooled passively. That is, passive PSU, passive CPU cooler, and passive cooling on the motherboard and on a solid state drive. Noise results of other cards on this page are measurements of the respective reference design.
This setup allows us to eliminate secondary noise sources and test only the video card. To be more compliant with standards like DIN 45635 (we are not claiming to be fully DIN 45635 certified), the measurement was conducted at 100 cm of distance and at 160 cm off the floor. The ambient background noise level in the room was well below 20 dBA for all measurements. Please note that the dBA scale is not linear but logarithmic. 40 dBA is not twice as loud as 20 dBA. A 3 dBA increase results in double the sound pressure. The human hearing perception is a bit different, and it is generally accepted that a 10 dBA increase doubles the perceived sound level. The 3D load noise levels were tested with a stressful game, not with Furmark.
NVIDIA has been pitching their "quiet", "silent", "inaudible" GTX Titan to us, and we were a bit skeptical of such a well-performing card delivering a low-noise gaming experience.
In idle, we see fantastic results. The card is essentially inaudible, which shows that NVIDIA did an outstanding job optimizing fan speed settings.
The card will be readily audible in most systems under load, but it is not nearly as loud as the HD 7970 GHz Edition. It is also significantly quieter than the GTX 690, which makes the GTX Titan the quietest high-end graphics card available.
NVIDIA's new Boost Clock 2.0 algorithm also provides additional software features that let you customize noise levels by trading performance for noise and vice versa. You basically pick a target temperature that the graphics card will try to match by adjusting boost clock, voltage, and fan speed accordingly. The out-of-the-box default is 80°C, which results in 38 dBA fan noise. We also tested with the temperature target set to 70°C, 90°C, and 94°C as reflected by the light-green bars in the second graph. The driver allows you pick a target as low as 60°C. In our testing, we could not achieve 65°C or below, since the boost clock algorithm will never clock the card below 535 MHz, which means the card can't cool down enough to reach those noise levels.
You still have the option to manually adjust fan speed by picking a fixed speed or changing the fan curve with the right tweaking utilities, which would let you enable a higher temperature target while still enjoying lower noise levels.
Unfortunately, NVIDIA does not provide any fan speed tweaking software of their own.