Differences Threadripper Cooling Comparison and Noctua CPU Cooler. We’re going to check out three different CPU coolers from Noctua for AMD’s TR4 Threadripper socket, and see how they perform both in terms of temperatures, but also overall system volume at both stock clock speeds and while overclocked.
I originally bought the largest model available, the NH-U14S for my 16 core 1950X CPU, but then I wondered if you could save a little money and get a smaller and cheaper cooler with perhaps similar performance. I reached out to Noctua and asked if they’d be willing to provide their other two Threadripper coolers that also cover the full TR4 socket, the NH-U12S and NH-U9, and they were happy to help out so huge thanks to them for that. Let’s first discuss the main differences between these three coolers so we know what we’re comparing.
The first obvious factor is the size of the heatsink, where the U14S that I purchased first is the largest of the lot, followed by the U12S and then the U9 being the smallest. As you can see here, the U14S actually covers the first PCIe slot of my ASUS Zenith Extreme motherboard by default, however it does have the ability to let you offset the heatsink by 3 or 6mm, I didn’t have any coverage issues with the smaller coolers.
The U14S comes with a 140mm fan and weighs 1030g, the U12S comes with a 120mm fan and weighs 870g, while the U9 comes with two smaller 92mm fans and actually weighs slightly more at 895g owing to that extra fan. What makes these coolers special is that they have full coverage of the Threadripper IHS, which should provide better cooling compared to using other coolers with the asetek bracket. I’m going to be testing these 3 coolers with my AMD 16 core 1950X Threadripper CPU at idle and under full load, both with stock speeds and overclocked to 3.9GHz. I simulated full system load using Aida64, and this was chosen as from my research it appears to be a more realistic workload when compared to Prime95, and I was after something closer to real world than absolute insanity for this test.
All tests were completed with an ambient room temperature of 20 degrees celsius, and I waited for over half an hour for temperatures in each test to stabilise. First we’ll discuss the idle temperatures. As expected we can see that the larger U14S cooler performs the best at idle, both at stock speeds or overclocked, however there’s only a small difference of 1 to 2 degrees from one cooler to the next down the line. While running at stock speeds under full load there’s a little more difference, however the results are still fairly close together. Only with the 3.9GHz overclock across all 16 cores do we start to see a larger difference, with the U14S performing almost 8 degrees celsius better than the U12S.
While overclocked the U9 was only a little worse off than the U12S which I found interesting, based on its size I thought it would be worse off, it looks like the extra fan is helping it out. It is possible to add a second fan to the U14S and U12S, however that was not tested here as I don’t have additional fans, I’ve tested with the fans that you’d get if you bought these coolers yourself. Cooling performance isn’t the only concern, let’s next take a look at how loud the system got with each cooler.
For this test I’m measuring the overall system noise rather than just the CPU fans, and the case fans should ramp up and down based on the CPU temperatures automatically so although this test depends on the fans of my case, hopefully it’s a fairly real world example of what you could expect to see, or I guess hear in this case. For reference the case I’m testing in is the Thermaltake View 71, it’s got quite a lot of openings for air between the tempered glass and the case. We can see at stock speeds the idle volumes are very close together, with a little more variance when overclocked at idle.
Interestingly, I found that the larger coolers were slightly louder at idle, but quieter under full load. Overall I’m pretty impressed with the system volumes, I’ve run much lower powered laptops that easily reach the 55 decibel mark, and 16 overclocked cores is quite a lot more to cool. Now it’s important to call out that there are other Threadripper chips with less cores that probably run cooler than what I’ve got here. In this particular example I’ve got the 16 core 1950X which will likely produce more heat compared to the 12 core 1920X, or 8 core 1900X CPUs.
I don’t currently have those lower tier Threadripper CPUs to compare against, so for now we’ll just be getting an idea of the cooling differences with the top end chip, which I think is actually better as it should provide more of a worse case scenario that these coolers would be dealing with. Based on these tests it seems that running at stock speeds any of these three Noctua coolers seem fine with the 1950X at both idle and full load, only when overclocking is introduced do things start to get a little worse off for our smaller coolers. If you’re planning on buying Threadripper and overclocking it, I’d definitely recommend the larger U14S cooler over the others if you have the room to physically fit it inside your case. I was originally planning on using the air cooler as a temporary solution until there were more all in one cooler options available that cover the full Threadripper socket, but after my testing I’m more than happy with the results of the Noctua coolers and I’ll be sticking with the U14S for some time to come.
Let me know down in the comments which cooling solution you’ll be using for Threadripper, and leave a share on the post if you found the information useful. Thanks for reading, and don’t forget to subscribe for future tech reviews like this one.