Best air cooler AM4?

doyll said:
While competitive with most twin towers it doesn't beat them.
NH-U12A prices of £109.99 and £124.99 are among the highest cooler prices out there.
As for quality, there are many other coolers just as well made.
Again, only thing Noctua has others don't is free mount updates, but I'm not paying half again to twice as much for a cooler just to be able to get mount update in 3-5 years. I'll save £55.00-80.00 for other things and in 3-5 years buy a new cooler for £40-60. Then sell old system as complete working unit to offset some of expense. ;)
Click to expand...
That's like comparing a laptop with a desktop
The laptop that performs similarly to a desktop will be more expensive - but still a sign of better engineering. There is no other single tower cooler that gets anywhere near the U12A, that's freaking impressive.

Regarding quality, have you actually had it side by side with a cheaper cooler to compare build quality? Cause I have compared it with several air coolers and the U12A feels much much higher quality.

Of course if you are on a budget you don't go for the U12A, but the same way if you are on a budget you don't go for an sff pc. Bulkier pcs tend to cost less for similar performance.
 
Last edited:
I use the U14s with my 3900x and it's great. I swapped the bundled fan for 2xAkasa Viper R which are more powerful at max / quieter for same (lower) performance of the noctua ones. Use a noctua PWM reducer taken outside the case to manually swap the PWM cap between quiet general running at 70% limit and flat out max power 100% limit for when I need it.
 
PA120 is a 120mm fan 4x 6mm heatpipe single tower cooler.
Maybe you are thinking of a different cooler?
PS120 SE is 2x 120mm fans w/ 7x 6mm heatpipe twin tower cooler.
NH-U12A is 2x 120mm fans on 7x 6mm heatpipe single tower.

But what is testing procedure used for your "real world workload test"?
Are they even recording air temp entering cooler when CPU temp is recorded?
Or are they just switching cooler in their personal system?

Just because someone says something or does a so called "review" or claims to have "tested" such-and-such coolers doesn't make their statement accurate or truthful. These peeps / sites are always trying to increase viewing / increase market numbers. But that doesn't mean their testing is even close to accurate or truthful.

What is your definition of "within margin of error"?
What is testing procedure? If it's not written down and followed doing every cooler test exactly the same, then tests are only a bunch of mumbo-gumbo.

Accurately testing coolers requires recording air temp entering cooler at same time CPU temp is recorded. Very few reviewers / testers do this. They only record CPU temp. Room ambient is sometimes used, but that's not actual temperature of air entering cooler. So we have reviewers claiming "margin of error of less than 1c" from tests ran in a room with air temperature varying from something like 19c to 25c claiming best cooler is 1c better than another. :rolleyes: Reality is their margin of error is not even known. :D

There are only a few sites that do anything even close to accurate testing.

GamerNexus and AnandTech are a couple.

AnandTech NH-U12A test results:
111380.png


111375.png


Second from bottom hottest cooler and noisiest of coolers in these tests. Accurate test results are very different from what you posted.

Expectedly, the performance chart of the Noctua NH-U12A lands right above that of its larger counterpart, the Noctua NH-U14S. The difference is very small (practically negligible), confirming Noctua’s claims that the NH-U12A can compete head-to-head with performance-oriented 140 mm fan coolers. For example, NH-U12A's performance advantage over a popular and significantly lower-priced tower cooler, the Cooler Master Evo 212, is momentous, partially justifying the retail price difference between the two products.
Click to expand...
Click to expand...
Their results are not at all like what you posted.
 
Following some laborious testing on a batch of fans the Akasa Viper R (which have a better spec in all categories than the noctua fans) perform better than the noctua fans. Testing used anenometer and dB meter to plot air flow with distance from fan and noise.

For a set dBA the Viper run quieter and shifted more air. At max speed the Viper is running at 1600rpm vs 1500rpm for noctua but shifting a lot more air, faster. The Noctua at 1500rpm is noisy. Lower pitched than the Akasa but noisy. It's why they limit the retail fan version to 1200rpm instead. And you only get one 1500rpm model in the box with the cooler. BTW if testing is just using the one fan it will not show the full potential of the cooler.

Anyway sticking those two Viper R fans on the U14s improves performance above what the Noctua fans can provide.
 
Last edited:
doyll said:
PA120 is a 120mm fan 4x 6mm heatpipe single tower cooler.
Maybe you are thinking of a different cooler?
PS120 SE is 2x 120mm fans w/ 7x 6mm heatpipe twin tower cooler.
NH-U12A is 2x 120mm fans on 7x 6mm heatpipe single tower.

But what is testing procedure used for your "real world workload test"?
Are they even recording air temp entering cooler when CPU temp is recorded?
Or are they just switching cooler in their personal system?

Just because someone says something or does a so called "review" or claims to have "tested" such-and-such coolers doesn't make their statement accurate or truthful. These peeps / sites are always trying to increase viewing / increase market numbers. But that doesn't mean their testing is even close to accurate or truthful.

What is your definition of "within margin of error"?
What is testing procedure? If it's not written down and followed doing every cooler test exactly the same, then tests are only a bunch of mumbo-gumbo.

Accurately testing coolers requires recording air temp entering cooler at same time CPU temp is recorded. Very few reviewers / testers do this. They only record CPU temp. Room ambient is sometimes used, but that's not actual temperature of air entering cooler. So we have reviewers claiming "margin of error of less than 1c" from tests ran in a room with air temperature varying from something like 19c to 25c claiming best cooler is 1c better than another. :rolleyes: Reality is their margin of error is not even known. :D

There are only a few sites that do anything even close to accurate testing.

GamerNexus and AnandTech are a couple.

AnandTech NH-U12A test results:
111380.png


111375.png


Second from bottom hottest cooler and noisiest of coolers in these tests. Accurate test results are very different from what you posted.


Their results are not at all like what you posted.
Click to expand...
Anandtechs tests are obviously flawed. Shadow rock slim better than the u12a. LOLK
 
Last edited:
Bencher said:
Anandtechs tests are obviously flawed. Shadow rock slim better than the u12a. LOLK
Click to expand...
rare said:
Agreed - their test rests are clearly nonsense.
Click to expand...
Peeps saying AnandTech test procedure is "flawed" and |"clearly nonsense" likely doesn't understand what a test procedure is or how to determine if it is a flawed test procedure or not.

AnandTech cooler test procedure does a much better job of accurately eliminating possible variables for more accurate test results. But cooler base of cooler being tested need to mate up with test station IHS. If it does not mate well to their IHS than heat will not transfer from IHS to cooler base causing test results to be higher temperature than coolers that to mate well.

I don't know why AnandTech NH-U12A test sample did as poorly as it did. It's quite possible that NH-U12A base plate was not as flat as other coolers tested .. meaning it didn't mate up well to test station IHS and thus did not transfer heat away from IHS resulting in it's higher temps.

Below review by EnosTech test results show quite a large gap between NH-U12A and NH-D15 results.
www.enostech.com

Noctua NH-U12A chromax.black and Chromax Covers Review - EnosTech.com

The Noctua NH-U12A chromax.black is a single tower CPU cooler that comes in black and is packing two fans for maximum performance.
www.enostech.com www.enostech.com
 
Last edited:
doyll said:
Peeps saying AnandTech test procedure is "flawed" and |"clearly nonsense" likely doesn't understand what a test procedure is or how to determine if it is a flawed test procedure or not.

AnandTech cooler test procedure does a much better job of accurately eliminating possible variables for more accurate test results. But cooler base of cooler being tested need to mate up with test station IHS. If it does not mate well to their IHS than heat will not transfer from IHS to cooler base causing test results to be higher temperature than coolers that to mate well.

I don't know why AnandTech NH-U12A test sample did as poorly as it did. It's quite possible that NH-U12A base plate was not as flat as other coolers tested .. meaning it didn't mate up well to test station IHS and thus did not transfer heat away from IHS resulting in it's higher temps.

Below review by EnosTech test results show quite a large gap between NH-U12A and NH-D15 results.
www.enostech.com

Noctua NH-U12A chromax.black and Chromax Covers Review - EnosTech.com

The Noctua NH-U12A chromax.black is a single tower CPU cooler that comes in black and is packing two fans for maximum performance.
www.enostech.com www.enostech.com
Click to expand...
There is what, 4c between the d15 and the U12A? So it's obvious that something is wrong with anandtechs test.
 
;
Bencher said:
There is what, 4c between the d15 and the U12A? So it's obvious that something is wrong with anandtechs test.
Click to expand...
I will agree something is wrong.
But it is not AnandTech test station or procedure.
For any cooler test system to work heat has to first move from test station IHS into cooler base, then into heatpipes, to fins, and finally into airflow from cooler fans thru cooler.

I assume you know AnandTech graph can be set to 60w, 100w, 150w, 200w, 250w and 300w so reader can see how cooler performs at different wattages.
AnandTech test station and procedure are very done.
AnandTech heat source wattage is more accurate than using an actual CPU.

GamerNexus is now cooler test station built similar to AnandTech to eliminate differences in CPU output heat.

You might find the 2020 GamerNexus article titled "Why Most Cooler Test Are Flawed: CPU Cooler Testing Mechodolgy" helpful:
www.gamersnexus.net

Why Most Cooler Tests Are Flawed: CPU Cooler Testing Methodology | GamersNexus

Coolers Why Most Cooler Tests Are Flawed: CPU Cooler Testing Methodology March 5, 2020 Last Updated: 2023-11-03 Diving into GN's methods for testing CPU coolers (2020 - 2023 Test Methodology) The Highlights GN's cooler testing has evolved over a decade and thousands of testsNormally, automatic...
www.gamersnexus.net www.gamersnexus.net

It's like I said, maybe the NH-U12A AnandTech tested was not moving heat from test station into cooler base because cooler base was defective / had too much convex curvature in it. That is assuming base to heatpipe, heatpipe evap to condense, and heatpipe to fins were all transferring heat as they are designed to do, same as cooler base is supposed to be designed to mate well to IHS.

When I was doing cooler testing I found cooler bases are not all the same flatness / convex curvature, even when comparing same cooler models. This difference in flatness / curvature greatly effects heat transfer from IHS into cooler thus creating a very wide range of cooling results. Only way to resolve this is to lap cooler bases before testing so all are same flatness, and this is not realistic because very few users lap cooler base and IHS.
 
doyll said:
;

I will agree something is wrong.
But it is not AnandTech test station or procedure.
For any cooler test system to work heat has to first move from test station IHS into cooler base, then into heatpipes, to fins, and finally into airflow from cooler fans thru cooler.

I assume you know AnandTech graph can be set to 60w, 100w, 150w, 200w, 250w and 300w so reader can see how cooler performs at different wattages.
AnandTech test station and procedure are very done.
AnandTech heat source wattage is more accurate than using an actual CPU.

GamerNexus is now cooler test station built similar to AnandTech to eliminate differences in CPU output heat.

You might find the 2020 GamerNexus article titled "Why Most Cooler Test Are Flawed: CPU Cooler Testing Mechodolgy" helpful:
www.gamersnexus.net

Why Most Cooler Tests Are Flawed: CPU Cooler Testing Methodology | GamersNexus

Coolers Why Most Cooler Tests Are Flawed: CPU Cooler Testing Methodology March 5, 2020 Last Updated: 2023-11-03 Diving into GN's methods for testing CPU coolers (2020 - 2023 Test Methodology) The Highlights GN's cooler testing has evolved over a decade and thousands of testsNormally, automatic...
www.gamersnexus.net www.gamersnexus.net

It's like I said, maybe the NH-U12A AnandTech tested was not moving heat from test station into cooler base because cooler base was defective / had too much convex curvature in it. That is assuming base to heatpipe, heatpipe evap to condense, and heatpipe to fins were all transferring heat as they are designed to do, same as cooler base is supposed to be designed to mate well to IHS.

When I was doing cooler testing I found cooler bases are not all the same flatness / convex curvature, even when comparing same cooler models. This difference in flatness / curvature greatly effects heat transfer from IHS into cooler thus creating a very wide range of cooling results. Only way to resolve this is to lap cooler bases before testing so all are same flatness, and this is not realistic because very few users lap cooler base and IHS.
Click to expand...
How is a tester better than an actual cpu? Usually cpu ihs is not flat. That's why for example my U12A never had issues with 12 / 13th gen and the supposed bending. I tried the bracket and it didn't do anything, probably because the U12A is not flat but convex.

A cooler review has to be taken into context _cause different coolers behave differently depending on the cpu being used. You can't have an end all be all test.

On average the U12A is the only single tower 120mm cooler that's up there competing or even beating dual tower coolers. I don't understand why people can't accept how impressive that is.
 
Bencher said:
How is a tester better than an actual cpu? Usually cpu ihs is not flat. That's why for example my U12A never had issues with 12 / 13th gen and the supposed bending. I tried the bracket and it didn't do anything, probably because the U12A is not flat but convex.

A cooler review has to be taken into context _cause different coolers behave differently depending on the cpu being used. You can't have an end all be all test.

On average the U12A is the only single tower 120mm cooler that's up there competing or even beating dual tower coolers. I don't understand why people can't accept how impressive that is.
Click to expand...
What I find most impressive about U12A is it's price. To be fair, it has 2 of the most expensive fans on the market on it as well. Noctua coolers are no better than other companies but are definitely the most expensive out. Only other extreme priced cooler I can think of is IceGiant ProSiphon, and it doesn't use heatpipes.

There is another cooler similar to NH-U12A at much lower price. Thermalright Ultra 120 Extreme Rev. 4 (U120EX Rev.4) priced $43-53 depending on color. Comes in silver w/ single fan as well as white & black 2x fan models.
 
doyll said:
What I find most impressive about U12A is it's price. To be fair, it has 2 of the most expensive fans on the market on it as well. Noctua coolers are no better than other companies but are definitely the most expensive out. Only other extreme priced cooler I can think of is IceGiant ProSiphon, and it doesn't use heatpipes.

There is another cooler similar to NH-U12A at much lower price. Thermalright Ultra 120 Extreme Rev. 4 (U120EX Rev.4) priced $43-53 depending on color. Comes in silver w/ single fan as well as white & black 2x fan models.
Click to expand...
So you dont find this impressive? That's 330 watts...

 
Bencher said:
So you dont find this impressive? That's 330 watts...

Click to expand...
No, link you posted does not impress me at all. :D
Why should it?

NH-U12A with it's 7x 6mm heatpipes is most heatpipes in such a small fin pack .. and NF-A12x25 fans are some of best on the market. But I find it hard to believe it can cool 330watts .. maybe with 21c air entering cooler and fans at full speed. But I wouldn't want to be in same room with it trying to keep 330watt CPU cool. ;)
 
doyll said:
No, link you posted does not impress me at all. :D
Why should it?

NH-U12A with it's 7x 6mm heatpipes is most heatpipes in such a small fin pack .. and NF-A12x25 fans are some of best on the market. But I find it hard to believe it can cool 330watts .. maybe with 21c air entering cooler and fans at full speed. But I wouldn't want to be in same room with it trying to keep 330watt CPU cool. ;)
Click to expand...
Ambient was higher than 21, you can tell by the minimum temperature. Cooler is not very noisy even at full fan speeds, so im not sure what the heck are you talking about. PA120 is much much noisier at full speeds.
 
How can you tell ambient by "minimum temperature"? If you ae referring to CPU "minimum temperature", please explain how it gives us air temp entering cooler is. I know CPU temp goes up and down with changes in air temp entering cooler, but it does not give us the air temp. Air temp entering cooler is lower than CPU temp, and changes air temp entering cooler effect CPU temp at very close to 1:1 ratio.
 
Bencher said:
PA120 is much much noisier at full speeds.
Click to expand...
This isn't true, Gamers Nexus review showed that Hardware Canucks noise testing was erroneous or they had a faulty unit. PA120 is likely quite a bit quieter than the U12A at full speed. Gamers Nexus doesn't test the U12A, but at 20" distance they have the PA120 at 42.3 dBA and the NH-D15 at 43.9 dBA. Hardware Canucks show the U12A is louder than the NH-D15.

I'd trust Gamers Nexus much more than Hardware Canucks as it's a technical in-depth sort of channel with excellent attention to detail. If you look at the noise floor, Gamers Nexus is at 26 dBA compared to 35.3 dBA for Hardware Canucks, so they were definitely not testing in a sound proofed room.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom