Before getting into the testing I did want to include the comparison numbers for heatsink height. This is important to keep in mind with SFF coolers especially because if you only looked at the performance numbers a few of the largest coolers would look amazing but may not fit. So with the L9a-AM4, for example, it is important to remember that it is only 37mm thick just like the other L9 coolers, less than anything else tested. The L12S, on the other hand, is 70mm so I am especially interested in how it compares with the Phanteks PH-TC12LS and the Wraith Spire from AMD. The Phanteks has better tall memory clearance, the L12S is considered “depends on the orientation” but even the best case still has some height limitation.
So my temperature testing is split into two sets of testing, one using the FPU test load in AIDA64 as it is the hottest you can get a CPU and then a second test run using the CPU setting that is a better representation of actual real-world high CPU load. The FPU test is especially complicated because it tends to overload all but the best heatsinks, especially when testing with the higher end CPUs. Because of that, I have to include the clock speeds as well because once the Ryzen CPUs hit 75c they stick at that temperature but downclock as needed to keep from getting any hotter. As you can see I tested with the 1800X, 1600X, and 1200 to get a range of possible CPUs and I also test using the stock fan profiles as well as a 100% fan speed to see how much headroom is left. It’s a LOT of testing, even for just two heatsinks but it does help get a better picture of how things perform and it shows you what CPU you might be able to comfortably run on it.
In the FPU test the NH-L9a-AM4, like the previous model, was hit hard when using the 1800X and the 1600X. Referring back to the clock speeds though you can see that there is an improvement, even if it isn’t a large one. Then with the Ryzen 3 1200, we can finally see that the gap is basically one extra degree. I suspect that If you wedged a 120mm fan over the top that would cool that extra heatsink sticking out on both ends things might run cooler. But it is still not a bad cooler considering the height. The Cryorig C7 performed better with the 1600X but is also taller, so it really depends on how much clearance you have. The NH-L12S, on the other hand, handled all three CPUs better here. None of them reached their thermal limit and it was actually the third best performing heatsink here with the Noctua D9L and Wraith Max performing better in a few tests. The 120mm AIO still performed a lot better, but if you need air we have a contender.
Now as you can see the CPU workload in AID64 is a lot less demanding and a better representation of the performance you should expect. That said I typically shop focused on FPU numbers so I never have to worry about overheating with some random fluke load. Anyhow the L9a-AM4 looks like a better option here with even good performance with the 1800X and of course the 1200 runs ice cold. The NH-L12S once again comes in up near the top of the charts. The H60 AIO is still the best option if you can fit one and the D9L is the hidden gem as well as AMD's Wraith Max cooler. But if you want a bottom facing fan design that will also cool other components in an SFF case this is a great option. It outperformed the Phanteks and Wraith Spire coolers that were both similar in height.