CPU: Intel i5-2500K @3.8GHz
Thermal Paste: Noctua NT-H1
Motherboard: ASUS P67 Sabertooth
RAM: 2x4GB Patriot Viper Xtreme II 1600MHz DDR3
OS HDD: Patriot Pyro 60GB SSD
Secondary HDD: Patriot Wildfire 120GB SSD
OS:Windows 7 Professional 64-bit
For idle testing, we booted up the system, allowed the CPU to stabilized at ~0% usage, and then tested for at least 10 minutes. Then we used RealTemp to record the CPU temperatures once the temperatures were stable. We tested two different conditions of the i5-2500K, the first is completely stock and default settings, and the second is overclocked to 3.8GHz with AUTO voltage. The single temperature recorded was the average between the four cores.
For our load tests, we used Prime95 to stress the CPU to 100%, and let it run that way until the temperatures became stable (allowing 15 minutes minimum), and then used RealTemp to record the temperatures. We tested two different conditions of the i5-2500K, the first is completely stock and default settings, and the second is overclocked to 3.8GHz and AUTO voltage. The single value recorded is the average temperature of the four cores.
We can clearly see that the NH-L12 doesn't push any sort of heat dissipation envelopes, but it was never designed to. The idle temps are within a couple degrees of all of the other coolers on the list, but a heck of a lot quieter overall. Even a full load at 3.8GHz keeps the i5-2500K at a very reasonable 60°C, which is something anyone would be happy to run 24/7, so there is headroom for an impressive 4GHz+ overclock even out of this low-profile cooler. Another trend we noted is that the addition of the bottom 92mm fan leads to little benefit as the load gets higher, as more of the load is dissipated through the heatsink, and not directly from the base. The single 120mm fan pushing from the top only adds a quarter of a degree at full load at 3.8GHz, the same for all intensive purposes.
We tried the pull-only configuration with both the 92mm and 120mm fan, and we also found those to performan identically. The temps shown are using the 120mm fan in a pull, ultra-low profile configuration, but again, we saw less than a degree average difference when using only the 92mm fan as the pull fan. We see performance lag a bit further in the pull-only setup, as expected. However, it still holds its own in an ultra-low profile configuration, and at 3.8GHz and full load keeps it at a healthy 65°C, and again leaves headroom for 4GHz+ CPU speeds at an extremely low profile. This result was better than we anticipated, but Noctua knew what they were doing when they put this package together.
When the fans are forced to 100 percent, a dull "whiiirr" can be slightly heard outside of the case, and any speed less than that it becomes nearly-silent. Using the low noise adapters ensured that the fans are absolutely whisper quiet, and we only saw a performance detriment of 1-3°C on average, which is not bad.