Cooling performance
Now I’ve been writing about hardware for nearly 10 years and I just have to say that this project ballooned into a crazy amount of testing. When I started this roundup I was just planning on testing everything on one CPU at stock and 100% fan speeds. I decided later to go with three different CPUs to get a better look at what options might be best for each cooler.
I then spent two 13 hour days testing only to figure out that my results were wrong. I was using the wrong temperature reading for one, but I also noticed that with a lot of the coolers my results were all the same. As it turns out with the Ryzen 7 and Ryzen 5 CPUs I was actually pushing the CPUs all the way up close to the TDP and they were throttling down. Because of this, I decided to toss everything and start over. This time documenting the clock speeds as well and using the correct temperature readout as well. But to go along with that I also recognized that the FPU stress test in AIDA64, while being the best way to completely heat up a CPU, doesn’t represent a normal CPU load as well. So I doubled up my testing again and added the CPU stress test setting as well. Huge thanks to AID64 for working with me through email and later over skype as well to figure out why my results were off and to help refine my testing.
So when it was all said and done, an article that I originally planned on having done in less than a week now needed something like 80 hours of testing, not counting the time needed to cool down between tests and the time it takes to swap between CPUs and coolers. Not to mention taking photos, editing photos, and writing everything. I’m only talking about all of this because this is how things snowball lol. Everyone should be warned!
So now that you know why I did each of these tests, let's take a look at the results we ended up with. So the Ryzen 7 and Ryzen 5 CPUs both had the 20c offset and from what I can tell a 75c/95c temperature is the TDP. So they did throttle down to stay below that. That is why you see a lot of the 75c results, especially with the 1800X. These are desktop CPUs that I’m attempting to run using extremely small heatsinks. So in the FPU results be sure to note the clock speeds to separate results from different capped results. Also huge props to AMD for the way the Ryzen CPUs handle the heat, they are able to pull a lot to keep thermals down.
So in order to best take all of this information in it is best to remember that the coolers tested do range in size. Specifically, you have some like the Noctua L9a that are only 37mm competing with coolers up to 110mm tall so there isn’t any one answer to what is best. You need to figure out how much room you have in your case then you can see what the best performing cooler is at that height
|
|
Noctua L9a |
37 mm |
ID Cooling IS-40V3 |
45 mm |
Cryorig C7 |
47 mm |
AMD Wraith Stealth |
54 mm |
ID Cooling IS-60 |
55 mm |
ID Cooling IS-50 |
57 mm |
Noctua L9x65 SE-AM4 |
65 mm |
AMD Wraith Spire |
71 mm |
Phanteks PH-TC12LS |
74 mm |
AMD Wraith Max |
85 mm |
Noctua D9L |
110 mm |
Corsair H60 |
Pump – 30mm Radiator/fan - 52mm thick |
One thing is clear right away, the Ryzen 3 1200 can run with any of the coolers tested and in every case the after market coolers are an improvement over the stock Wraith Stealth cooler that it comes with. So most of my focus is on the 1600X and 1800X that push the limits of the coolers more. With the 1600X I was surprised that the Wraith Max was actually the best performing air cooler. With this demanding test you really can only go with the Wraith Spire, Wraith Max, C7, IS-60, D9L and the PH-TC12LS if you don’t want to be thermally limited. Of those, the C7 was the shortest and the IS-60 had the best cooling while still being relatively short.
For the Ryzen 7 1800X things were even hotter as expected. The Wraith MAX once again impressed me being the coolest of the air coolers tested with the stock fan settings. The Noctua D9L was the best air cooler with the fan cranked up but it was the tallest cooler tested. Of course the water cooling, even being just a 120mm kit was still better than the rest but that is only if you can fit it in your case. Overall it is larger than all of the other coolers by a large margin and the tubing was long even in our large ITX case.
The main thing to take from this though is that if you plan on slamming the CPU as hard as the FPU stress test does you might want to stick with a 1600X unless you have room for a larger cooler.
AIDA64 FPU Stress Test |
AMD Ryzen 7 1800X |
AMD Ryzen 5 1600X |
AMD Ryzen 3 1200 |
|||
|
Stock |
100% |
Stock |
100% |
Stock |
100% |
AMD Wraith Stealth |
75c 3292MHz |
75c 3391MHz |
75c 3491MHz |
75c 3541MHz |
47c 3092MHz |
45c 3092MHz |
AMD Wraith Spire |
75c 3541MHz |
75c 3591MHz |
74c 3691MHz |
71c 3691MHz |
44c 3092MHz |
39c 3092MHz |
AMD Wraith Max |
70c 3691.2MHz |
68c 3691MHz |
64c 3691MHz |
61c 3691MHz |
39c 3092MHz |
37c 3092MHz |
Corsair H60 |
56c 3691MHz |
54c |
54c |
54c |
33c |
32c |
Cryorig C7 |
75c 3491MHz |
75c 3541MHz |
74c 3691MHz |
72c |
42c 3092MHz |
40c 3092MHz |
ID Cooling IS-40V3 |
75c 3516MHz |
75c |
75c |
75c |
45c |
42c |
ID Cooling IS-50 |
75c |
75c |
75c 3641MHz |
75c |
43c 3092MHz |
42c |
ID Cooling IS-60 |
75c |
75c |
71c |
70c |
40c 3092MHz |
39c |
Noctua L9a |
75c 3466MHz |
75c |
75c |
75c |
44c 3092MHz |
40c |
Noctua L9x65 SE-AM4 |
75c |
75c |
75c |
70c |
43c 3092MHz |
39c 3092MHz |
Noctua D9L |
72c 3691MHz |
65c 3691MHz |
65c 3961MHz |
59c 3691MHz |
39c 3092MHz |
36c 3092MHz |
Phanteks PH-TC12LS |
75c 3616MHz |
75c 3641MHz |
70c 3961MHz |
67c 3691MHz |
40c 3092MHz |
38c 3092MHz |
In the second batch of testing with the AIDA64 CPU setting in their stress test the temperature numbers all look a lot more manageable. This test is closer to what you will most likely be seeing when gaming or using your PC in the real world. As you can see we finally weren’t running the CPUs all the way up to TDP so throttling wasn’t an issue this time around. The Ryzen 3 1200 was once again a non-issue with any of the coolers. The smallest cooler the Noctua L9A still ran cool with this CPU.
The Ryzen 5 1600X started to heat up a little more but again even the L9A could handle this while seeing temperatures I would be happy with. Going up to the larger coolers did drop the temps down into the 40s but whatever fits in your build is going to work at this range and load.
Then with the Ryzen 7 1800X, I was surprised to see that this load didn’t really push the 1800X much past the 1600X. The L9A was still able to handle things shockingly with the hottest aftermarket cooler being the IS-50. The Wraith MAX was solid as well, as expected but I’m really digging the Phanteks PH-TC12LS. It is taller but not as tall as the D9Lor the Wraith MAX yet it performed really well here.
AIDA64 CPU Stress Test |
AMD Ryzen 7 1800X |
AMD Ryzen 5 1600X |
AMD Ryzen 3 1200 |
|||
|
Stock |
100% |
Stock |
100% |
Stock |
100% |
AMD Wraith Stealth |
69 |
67 |
65 |
57 |
41 |
36 |
AMD Wraith Spire |
62 |
54 |
57 |
50 |
38 |
34 |
AMD Wraith Max |
53 |
47 |
51 |
44 |
35 |
33 |
Corsair H60 |
44 |
39 |
42 |
39 |
30 |
28 |
Cryorig C7 |
60 |
55 |
56 |
50 |
37 |
34 |
ID Cooling IS-40V3 |
63 |
57 |
58 |
53 |
39 |
36 |
ID Cooling IS-50 |
62 |
60 |
56 |
54 |
37 |
36 |
ID Cooling IS-60 |
56 |
54 |
53 |
50 |
34 |
33 |
Noctua L9a |
63 |
57 |
63 |
56 |
42 |
39 |
Noctua L9x65 SE-AM4 |
60 |
54 |
58 |
51 |
39 |
34 |
Noctua D9L |
54 |
45 |
50 |
43 |
34 |
31 |
Phanteks PH-TC12LS |
57 |
51 |
54 |
48 |
36 |
33 |