Performance

Test System

All testing is done at a room temperature of 20°C (68°F), with a 1°C margin of error. The coolers are tested with Turbo, EIST, and C1E enabled, which will allow the CPU to clock down to a low 1.6 GHz while idle, or clock up to proper speeds under stock and overclocked conditions. With the use of XMP, the Intel i7 3960X ES chip I used for testing runs at 3.6 GHz under stock load. Overclocked, the chip is set to 4.1 GHz at 1.225 volts. During all these tests, fan speeds are set to run at 100% in the BIOS, with temperatures being recorded by AIDA64.

The idle test will consist of the CPU sitting idle at the desktop for 15 minutes. This will allow for a stable temperature reading that will be recorded at the end of those 15 minutes.

AIDA64 and its CPU stability test represent a typical multithreaded user load. It is run for 15 minutes before the highest reading during the test is recorded and taken as the result. This test lets enthusiasts know what temperatures they can expect to see with games and applications.

Prime95 is the multithreaded stress test I will use to find the cooler's temperatures at maximum load. This is done by using the "In-place large FFTs" setting to truly stress the cooler's ability at keeping temperatures in check. The test is run for 15 minutes, and the highest recorded temperature is used as the result.

Fan noise testing is done at 20%, 50%, and 100% settings, and the dBA level is recorded by a Pyle PSPL25 sound pressure level meter at a distance of 30 cm. Fan RPM results are taken at the same 20%, 50%, and 100% settings.

Idle

At idle, the Noctua NH-L12 comes in near the back of the pack, but users need to remember its size plays a huge role in performance. At stock in dual-fan configuration, the difference between the NH-L12 and competition products is a wash, but when switched to single-fan configuration, a rise of 2°C in idle temps at stock takes place. Overclocking results are omitted for the single-fan configuration because the fan failed. The cooler is not meant to be used on overclocked extreme processors with just one fan.

Typical Load

During the typical load test, the Noctua NH-L12 pulls up at the back of the pack, 4°C behind the nearest competitor. In single-fan configuration, it comes in 6°C behind the typical dual-fan setup. Looking at overclocked settings, the NH-L12 again falls back into last place, 9°C behind the Hyper 212 EVO.

Max Load

At stock settings, the Noctua NH-L12 performs admirably on this test. It does come in near the back of the pack, but the cooling performance it offers for its size is great, especially for single-fan performance. Yes, it gets hot, but 79°C on a 3960X is not something to make fun of, since the cooler is just 66 mm tall in its single-fan configuration. Things do fall apart with the overclocked CPU. The NH-L12 was never meant to be used in such way, but ties the Coolermaster Hyper 212 EVO on the 4.1 GHz CPU overclock. However, it should be noted that 85°C is only 1°C away from hitting the thermal throttle point of this CPU.

Fan Noise

As can been seen by the fan noise and speed charts, the Noctua NH-L12 is the quietest cooler I have tested to date. This, coupled with its size, makes its performance truly astounding. The fan speeds while achieving these low sound levels are surprising. 1720 RPM on the 92 mm fan is what causes the brunt of the noise - to the point of no change in noise levels with a single-fan configuration. This reinforced what I had already suspected: the 92 mm fan is the biggest source of noise on the Noctua NH-L12. Hence, I would recommend running the cooler with both fans unless you want to install the cooler into a low-profile case that might benefit from a 66 mm high cooler.