Noctua AMD Threadripper Air Cooler Roundup

We take all three Noctua air coolers built for AMD Threadripper CPUs and put these to the test on both the 2950X and 2990WX. We run the NH-U14S, NH-U12S, and NH-U9 through the paces with both Precision Boost 2 and Precision Boost Overdrive using multiple fan configurations and compare these to the Silver Arrow and Wraith Ripper too.

What Does Noctua Mean?

The little owl, Athene noctua, is the symbol of the Greek goddess Athene, who represents wisdom, science and strategy. She has a marked preference for the most clever among the Greek heroes, like Odysseus, who endure their adventures by the use of savvy consideration instead of mere force.

Article Image

The owl has always symbolised intelligence and prudence, and with its manlike face, it is the emblem of attentiveness and communication. We at Noctua imitate its attributes of calm and accurate observation in the dark, its soundless flight as well as the effective, economical use of its powers and precise attack.

The name Noctua is known by every computer hardware enthusiast on the planet worth his salt. Noctua is known for designing and building excellent CPU air coolers, but above all its coolers are widely known for being near-silent in most applications. As noted above, its signature logo owl is known for its silent flight and sneaking up on its prey. (However the one that sits on top of my house every night at around 3am is loud as hell and has a habit of waking me up, admittedly however, I never hear her coming or going.) When Noctua offered us up every AMD Threadripper cooler it produces, we immediately took it up on the offer. AMD's Threadripper and silent air cooling seem to be far from compatible to us, so we wanted to see all of these in action.

Noctua AMD Threadripper Air Cooler Roundup

Noctua has found a winning formula for CPU air coolers over the years. We have three Noctua coolers up for review today that extend that tradition; the NH-U9 TR4-SP3, the NH-U12S TR4-SP3, and the NH-US14S TR4-SP3. As is par for the course, these come equipped with Noctua's brown and other color brown fans.

Article Image

As you might have also guessed, the number noted in the first half of the part number refers to the fan size along with an appropriately sized heatsink. Also, as you might have guessed these coolers are only for use on the TR4 or SP3 sockets used with AMD Threadripper and EPYC CPUs. When we mention "only," that is exactly what we mean.

Article Image Article Image

Please note that the NH-U9 TR4-SP3 (and other TR4-SP3 coolers) is a dedicated version for the AMD TR4/SP3 socket. Due to its customised larger base, it cannot be used on any other sockets and is incompatible with Noctua’s standard mounting kits.

Article Image

What we have learned in the past is that coolers built specifically for Threadripper processors tend to work a lot better when it comes to thermal performance. Of course that is exactly what we want to find out.

Test Scenario and Criteria

For all our testing here today we are using MSI's MEG X399 Creation motherboard and the Threadripper 2990WX and 2950X. You can read our review of those CPUs here, with installation outlined here, and 2950X overclocking covered in-depth here.

We changed up the way we are testing the Threadripper CPUs in our last Threadripper Air Coolers-Silver Arrow TR4 vs. Wraith Ripper. Given the fact that we think that many Threadripper owners will be using Precision Boost 2 and Precision Boost Overdrive to control their CPU clocks, we wanted to use these dynamic clock controls in our review. If you are not familiar with how PB2 and PBO work, we would highly suggest you give our AMD Ryzen Threadripper 2950X Overclocking article a read. In that article we show just how much cooling can and will impact your Threadripper clocks using a chiller system, and of course traditional water and air cooling.

For the sake of comparison, we have pulled all the data from the Silver Arrow vs Wraith Ripper review and used those temperatures in our graphs. What is new here, in case you have not read that article, is that we have included CPU Package Power as well as average CPU clocks after the system has been heat-loaded using Prime95. This is very much a worst case scenario in terms of workload. Since the system is fully heat-loaded when we collect data, the package power will be at its peak, and the clocks will be at the lowest during the test. Conversely, we are running these cooler tests on an open test bench which will give us a best case scenario. To somewhat hedge against that, we have brought the ambient temperature up to ~78F/26C. The reason that we have done this testing on an open bench is simply because of the workload that is required to install most of these coolers. Many "require" having the video card removed, shuffling multiple fans on and off, moving hardware and wiring, etc. Quite simply we just did not have the bandwidth to get this done in any reasonable timeframe if we were to test inside a case. Also, keeping the ambient temperature static as we reach the end of summer here in Texas is not tremendously easy either as the sun rises and falls, so we have to get work done when we know we can keep our temperature window inside 1 degree F.