Cooling Efficiency Tests
Dependence of Cooler Efficiency on Fans Number and Rotation Speed
The first thing that made me a little uneasy was the fact that with Noctua NH-D14 default fans installed in a standard manner the 120 mm fan turned out extremely close to the back of the graphics card. There is barely 15 mm of space between them, which will hardly allow any fresh airflow to get to the fan and the heatsink. That is why we ran the tests in the following two configurations:
As you see, when we moved the 120 mm fan onto the second heatsink array and installed it for air exhaust, the cooling efficiency didn’t change. Note that we use a “cool” Radeon HD 5750 graphics card. Of course, the standard fan configuration on Noctua NH-D14 will be considerably less efficient if the 120 mm fan will be fed the airflow from the scorching hot Radeon HD 4870 X2, for instance. Therefore, we are going to complete our today’s test session with the second fans configuration. For all potential owners of the new Noctua NH-D14 we would strongly recommend to experiment with fans configurations and airflow directions.
The next round of tests was intended to check out the dependence of Noctua NH-D14 heatsink efficiency on the rotation speeds of the fans, their number and type. To accomplish this we tested the cooler in two rotation speed modes with one 140 mm Noctua fan, two default Noctua fans, three fans (Noctua + Thermalright TR-FDB) and two Thermalright TR-FDB fans:
In the latter case the fans rotation speed varied between 600 and 2000 RPM with 200 RPM increments (±20 RPM). The results are summed up on the graph below:
Here we should dwell on several interesting things. Firstly, Noctua NH-D14 is currently the only solution that can cool a CPU overclocked to 4 GHz running Linpack 64-bit with two fans at 600 RPM. Believe me, it is an unprecedented achievement! Secondly, despite relatively large inter-plate space and heatsink optimization for quiet low-speed fans, the cooler efficiency with two Thermalright TR-FDB fans increases almost linearly up to 1800 RPM (by 8°C in the 600-1000 RPM interval, 6°C in the 1000-1800 RPM interval). Thirdly, with two default Noctua fans, NF-P12 and NF-P14, the cooler efficiency is exactly the same as with a pair of Thermalright TR-FDB fans, which means that replacing the default fans with alternative ones doesn’t make any practical sense. If we only leave one 140 mm fan, then even in quiet mode at 870 RPM the cooler can work a small wonder, and by adding the third fan to the two default ones you can lower the peak temperature by 2°C more.