Noctua NH-L9x65 and NH-L12S Faceoff (Benchmark inside)

stratuh · January 6, 2019

Last week I bought a Noctua NH-L9x65 cooler for my Ryzen 2600. I sort of wanted to fit a 120mm fan on it, and after trying to make my own fan clip with steel wire, I gave up and decided to pick up a Noctua NH-L12S to replace it.

Since I have two low profile Noctua coolers in my possession at the moment, I thought why not benchmark them against each other? I don't have equipment to test sound level, but using Aida64 I can compare their idle temperatures and their temperatures under a synthetic load. I'm doing this test partly out of curiosity but also due to the lack of comparisons I could find online. I think I found two or three, neither of which use my CPU, and I wanted to see for myself how they fare head-to-head. System configuration (or what I think is important for the tests):

Ryzen 5 2600
2x8gb DDR4-3000 Corsair Vengeance RGB Pro
Gigabyte X470 Aorus Ultimate Gaming
Silverstone FT02

The FT02 has three 180mm Silverstone AP181 intake fans. I've used them on their "low setting" of 700rpm. The FT02 also has one 120mm exhaust fan (don't know model) which is controlled by the motherboard, although the FT02 spec sheet says 1200rpm. These are the only case fans and admittedly, after five plus years of use, they're not pristine or perform as you'd expect brand new fans.

The two coolers are tested. The L9x65 is tested with the stock NF-A9x14 and the L12S is tested with the stock NF-A12x15 (Fan A, 1850rpm) and separately with an NF-P12 (Fan B, 1700rpm). Both Fan A and Fan B are atop the L12S in a push configuration. All fans plugged into the CPU_FAN header on the motherboard and no speed reducer cables are used. Gigabyte's Smart 5 fan curving is used with the below values:

30% fan speed at 25c
40% at 35c
55% at 45c
75% at 55c
100% at 65c

So two coolers, one of which have two fan configs gives three tests. I've also done the same tests with a slight overclock, giving six tests total. The stock 2600 speed is 3.4ghz and my overclock is at 3.7ghz. AMD C&Q is disabled in BIOS.

I used Aida64 benching CPU, FPU, cache, and system memory together to give an idea of a strong 100% load. Tests were run for 20 minutes and I report the idle temperature, the average temperature, and the maximum temperature. The maximum temperature is there because sometimes a test may have spiked or landed at a certain temperature briefly before normalizing. The average temperature is indicative of the CPU temperature. I don't know ambient temperature but my thermostat in an adjacent room says 21.5c. In no test did I have throttling issues.

The statistics screen of Aida64 gives me the summary for each test:

L9x65 stock
L9x65 OC
L12S stock w/ Fan A
L12S OC w/ Fan A
L12S stock w/ Fan B
L12S OC w/ Fan B

And put into a chart:

MSRP for the L9x65 and L12S respectively in Canada are $59.99 and $79.99. For $20 more, I'm able to drop a good few degrees using the stock fan and an additional degree or two using a different 120mm fan. Something interesting is changing the fan from a 15mm thick one to a 25mm one did little to the temperature, at least nothing outside the margin of error. The maximum temperature for the OC tests are just spikes in the temperature I'm not sure what cause. Still, on the whole, the difference in average temperature is minimal. Clearly if you have a 25mm thick 120mm fan (or the NF-P12 specifically) lying around, definitely use it, but it's not a dramatic change.

One thing I wish I could do is compare against the stock AMD Wraith cooler the CPU comes with. I personally don't like putting these coolers on because they're hard to take off for me, so unfortunately those results are absent. Something not graphed but apparent looking at the tests is the bigger 120mm fan cooled GPU and motherboard. There's a 2-3 degree difference using the 15mm thick fan versus the 25mm thick one. I'm mostly concerned with CPU temps so take from that what you will.

You can make your own conclusion from this: if you're on a budget, the L9x65 is a splendid cooler, but if you can spend a touch more, the L12S does bring the temps down a bit. Both are low profile so perfect for small form factor or slimline builds.

WoodenMarker · January 8, 2019

9 minutes ago, Phaytuh said:

Great work! It's not often that I see someone put so much effort into sharing what they've learned in their own PC adventures. I didn't know how little a difference there was between he 15 and 25mm fan but that's good to know.

Yep, the NF-A12x15 is a pretty impressive fan.

https://www.overclockers.com/noctua-nf-a12x15-fan-review/

https://www.overclock3d.net/reviews/cases_cooling/noctua_nf-12x15_slim_fan_performance_review/3

http://tech-legend.com/reviews/noctua-nf-a12x15-pwm/

stratuh · January 8, 2019

53 minutes ago, Phaytuh said:

Great work! It's not often that I see someone put so much effort into sharing what they've learned in their own PC adventures. I didn't know how little a difference there was between he 15 and 25mm fan but that's good to know.

It's difficult to find a common ground. They're different RPM, different CFM, even different lines (the NF-P12 is their Redux line). Still, I thought a 25mm thick fan would do more than it did, with the only change really noticeable is with an overclock where it performs a few degrees better. The thicker fan also cooled the motherboard and GPU a touch, why or how I'm not sure. It could be error territory and can be ignored, but I can reproduce it in another case (Corsair 400C).

I think I just found conflicting reports on other sites benching the two coolers which prompted this test. The L12S is $20 more expensive with a bigger fan but smaller heatsink. One site said they performed more or less equal, I even found one say it performed worse. There just wasn't enough data so why not contribute my own?

If you want to hear something wild: I originally did this test in my FT02 with the three 180mm 700rpm intake fans. These push a crapload of air. I redid a couple of the tests migrating to a Corsair 400C with two NF-P14 140mm intake fans standing in for the stock Corsair ones. You'd think 3x180mm fans would be better, and Silverstone is definitely a respectable brand for fans, but the Noctua held its own with almost equivalent CPU temps. I benched for about 1.5hr and 50c was my average. Noctua is just magic.