Air Cooling -- Myths and setup tips for the novice performance / gaming builder

[Space Lynx]

Has the air cooling community decided if the U12A or the D15S/D15, between the two, which is the best? I have watched several reviews, and it seems to go either way. With the U12A winning in the last two reviews I watched.

my guess is with a slight fan curve bump on two of the best fans ever made (U12A) it might even match some AIO coolers... all the reviews I have seen compare stock fans only or 100% max only. I don't think the 100% is realistic... I think a slight fan curve bump is though...

 

[Mussels]

Probably close enough that it comes down to the fan speeds you use

 

[Bill_Bright]

I mean, GN has demonstrated 10* differences in CPU temps but for your review:

Did you actually read your own source and take a moment to study what the article says about "obstacles", or more specifically, did you note the different results between the different types of obstacles?

Why is that important here? Because typically, cases that do NOT have filters do NOT have huge gaping, wide open, unrestricted holes through which they push or pull the air through.

Is there a difference when the fan has no obstacles in its path compared to obstacles? Of course! However, as your article noted,

fans do not usually blow into an empty space, but have a filter, grille or radiator in front of or behind them

I assume there are exceptions but I cannot think of any case I have seen designed for use without filters that didn't have some sort of grill similar to the hexagonal grill noted in your article. These grills are there to prevent lawsuits - by keeping little, prying fingers out of the paths of spinning fan blades. They offer little to no dust filtering capability - at least not when clean.

Now, if we note the 31 dBA graphs for the Noctua, as examples, and study the results, we see that fan moves 46.44 m^3/h (cubic meters per hour) of air into empty space. But again, when in use, case fans do NOT blow into empty space. When mounted in a case, they blow through (or draw through) different obstacles. With the hexagonal grill, that number drops significantly to 31.05 m^3/h. But with a decent nylon filter, that value is only a little less at 28.2 m^3/h.

For sure, there is a difference between no filter (31.05 m^3/h) and filter (28.2 m^3/h). But is that really "significant" as you claimed? Umm, no. Just for visualization purposes, a 10 x 12 x 8 foot room is 960 cubic feet or 27.18 m^3/h.

Now for sure, the difference between empty space (no obstacles) and the grill definitely is significant. But I know of no case or user who runs with spinning fan blades exposed to little fingers like that. And if you think the difference between a typical hexagonal grill and the nylon filter is "significant", then you are entitled to your opinion. But I don't see that as significant.

I think we also have to be realistic and use some common sense here, and set some guidelines. First, I am talking about my experiences and I always buy quality cases from reputable makers for all my personal and our client builds here. These quality cases use quality fans and quality filters, and they are marketed touting their sound suppression and cooling options. Example, Fractal Design cases.

I could be wrong but I believe most readers here who buy their own cases, look for those (at least the cooling) qualities too. No case maker worth their salt is going to design, build, and market their cases in that manner, only to have them fail miserably at being capable of providing adequate cooling.

Do filters impact air flow? Of course! But in a quality case that uses quality fans properly configured for good air flow, the difference between using filters and non-filtered grills is not that significant. If you need to boost the fan speed a little to compensate for an increase in temps, a quality fan can easily do that while keeping any increase in fan noise negligible. And if you need to add another quality fan to increase air flow, a quality, properly chosen case, will allow that too.

I note two quality fans running at slow (quiet) speeds can move a lot more air than one fan running at high speeds.

If the case you selected and bought does provide adequate cooling quietly, and/or does not support adding another quiet fan, YOU FAILED to do your homework and bought the wrong case for your needs! That is not the case's or filter's fault!

And for the record, I don't say this much but, I really hate fan noise.

 

[claes]

I assume there are exceptions but I cannot think of any case I have seen designed for use without filters that didn't have some sort of grill similar to the hexagonal grill noted in your article. These grills are there to prevent lawsuits - by keeping little, prying fingers out of the paths of spinning fan blades. They offer little to no dust filtering capability - at least not when clean.

Now, if we note the 31 dBA graphs for the Noctua, as examples, and study the results, we see that fan moves 46.44 m^3/h (cubic meters per hour) of air into empty space. But again, when in use, case fans do NOT blow into empty space. When mounted in a case, they blow through (or draw through) different obstacles. With the hexagonal grill, that number drops significantly to 31.05 m^3/h. But with a decent nylon filter, that value is only a little less at 28.2 m^3/h.

For sure, there is a difference between no filter (31.05 m^3/h) and filter (28.2 m^3/h). But is that really "significant" as you claimed? Umm, no. Just for visualization purposes, a 10 x 12 x 8 foot room is 960 cubic feet or 27.18 m^3/h.

Between the condescension and your laziness I just skimmed past here. You took the results from the noise floor and compared them to one another, which represents a best case scenario (idle), rather than using 100% fan speeds or a higher noise level, ie load, where cooling matters.

A nylon filter is a 20% loss, hexagonal holes are a 23% loss. I must be an idiot for thinking that’s insignificant. Now imagine you have a hex mesh front, and a filter, and whatever other instructions are part of the chassis fan mount. Nbd I guess.

I agree about the quality of the chassis being most significant in impacting a fan’s performance, but most people aren’t running their fans at 800rpm under load, and we’re talking about the best standard 25mm 120mm on the market, with an excellent PQ curve designed to overcome standard impedance.

Which all goes back to the initial question — obstructions in chassis affect how much intake you need to overwhelm the less impeded exhaust fans (typically only hex stamps) to maintain positive pressure.

Edit: to be sure, I hate noise, too, which is why my fans don’t run at idle and don’t surpass 800rpm on load. But I’m in a minority.

Also, the vast majority of “airflow” cases these days come with a stamped mesh front and a nylon filter. I think you are missing out on market changes here.

 

[Bill_Bright]

Between the condescension and your laziness

It is sad when you see someone who has such low self-esteem who, in an attempt to make them look superior, are compelled to lash out with puerile personal insults when another poses a different or opposing stance, instead of maturely, and simply debating the technical facts. I pity them.

I just skimmed past here.

But that's not condescension, right? Well, you should have read my post - carefully - because clearly, you missed several points I made. To include where I said filters do indeed impact air flow. There was never any denying that.

I also said you are entitled to your opinion - but apparently, it's condescending and lazy for others to express theirs when different from yours.

You took the results from the noise floor and compared them to one another, which represents a best case scenario (idle), rather than using 100% fan speeds or a higher noise level, ie load, where cooling matters.

Gee whiz. You accuse me of using one extreme (best case, idle), then you go all the way to the opposite extreme with 100% fan noise! How is that better? That is just as bad, if not worse because NO fan should need to run at 100% (except, maybe in a short burst) because then, they have no cooling capacity left!

If case fans need to run at 100%, YOU FAILED to set up your case cooling properly!

The reality is, most computers run much closer to idle than maxed out most of the time.

And I never said "insignificant". That is your word. But IMO, 3% is not "that significant".

 

[claes]

Check 1000-1200 rpm, ie 33dB-39dB
1000rpm = 33dB
1139rpm = 36dB
1286 = 39dB

Unobstructed:
1000: 54
1139: 60
1286: 67

Nylon:
37 = 32% loss
45 = 25% loss
51 = 24% loss

Hex:
40 = 26% loss
44 = 26% loss
49 = 26% loss

If we’re to use your noise floor example:
787rpm @ 31dB
46 unobstructed
28 nylon, 39% loss
31 hex, 33% loss

The only reason hex is not as much of a problem at 820rpm as nylon is because there’s very little pressure at such a low rpm. When you actually need cooling a nylon filter is not so bad.

Again, the premise of this argument was how to maintain positive pressure, not whatever tangent you’re on about about the meaning of “significant.” Most of the time your intake fans are facing greater obstructions than your exhaust, and so you need to account for those obstructions in addition to the rated CFM.

Have a nice day

 

[Count von Schwalbe]

If we’re to use your noise floor example:
787rpm @ 31dB
46 unobstructed
28 nylon, 39% loss
31 hex, 33% loss

What units are these? 46 CFM?

If so, how are you measuring? I would like to try this myself.

 

[claes]

Please see the hwcooling.net link in my previous post. I was being lazy — they measure in m^3/h and not CFM, I was just referring to the previous discussion about positive/negative pressure and fan area and so on.

TBF, Bill and I are in agreement — fans generally face obstructions, and we both think you need to account for them. I’m just pointing out that it’s not as simple to determine case pressure as intake - exhaust, but (intake/impedance) - (exhaust/impedence). I don’t know where Bill’s disagreement lays as they seem to agree with the premise.

Edit: I get it now — Bill is comparing a typical room to a 50L chassis with 300-1200W of heat inside it. I think the “significance” speaks for itself here.

 

[sneekypeet]

If so, how are you measuring? I would like to try this myself.

You can grab an anemometer at Amazon.

 

[Tech_fanatic]

Aircooling has all ways been my favoridt cooling type. Cheaper, most reliable and easy to/less maintinense needed compared water cooling. The downside of aircooling is that not every one like a big ass dual tower cooler in there system and aircooling can dissapate less heat than a proper custom water loop, limit overclock in some cases.

With the rate CPU´s goes in form of power consumption. Aircooling and high-end cpu´s like 13900K and 7950X really pushing the limit for what even the best air cooler can handle with stock power consumption of all ready 300 watt and 230 watt respectivly. Even the zen 4 6 core now has a power of 105 watt compared to my 5600X of 65 watt. Zen 3 is easy to aircool while zen 4 is a nightmare. Specially for the 170 watt rated parts. 5950X rated for 105 watt (in realioty it uses stock 141 watt) while 7950X is 170 watt (in reallity it more like 230 watt). The higher clock speed the latest gen of CPU´s are capable of hitting, has a back side to that. More heat and power consumption that needs even better cooling. Intel and AMD pushes there cpu all ready at stock now to the limit of what air cooling can handle.

My 5950X cooled by a Noctua NH-D15 hits stock single core up to 72C while multicore load is 58 C. Witjh PBO single core load is the same while multi core load raises to 76 C (that is at 200 watt my motherboard allows as max). Manuel all core OC to 4.65 GHz at 1.375 volts i hit 86 C and is really the limit for my cooling before throttle temp is reached at 90 C.

What i try to say is that aircooling is apselutely sifficiant for Zen 3 based CPU´s, while Zen 4 and alder lake/meteor lake is really pushing aircooling to the limit because of the increased core clock and power consumption. This unfortunaly can mean this is my last aircooled high-end build do to cpu are pushed so hard all ready at stock. I can off cause limit power, but how fun is it to buy an exspensive cpu to just limit it´s capabillity.

I will say a good big dual tower air cooler is just as good as at least a 240 MM AIO, maybe even 360 MM in some cases.

My system is at least a prove of that a high-end system dosent have to be water cooled to perform it´s best. My system is completely aircooled and i will not call my system low end by any means.

My system (se all spec in my profil):

Thats's a beast of a dual cpu/mb/gfx system in one case (if I've ever seen one). Air-cooling two cpu/mb/gfx combos in one case is one heck of a thing that you have done here. Really awesome

 

[Dirt Chip]

Wonderful post and thank you @An0maly_76

I'm going with 13900k and nocuta nh-u12a cooling on fractal define 7 case. I know it might be too worm with the stock 2*140+1*140 and after I will do some testing I'm planning to add more 140mm fans to get better airflow.
The case is sitting on the floor so dust filters are on and I'm aming at a very quiet rig.
I have the aero z690 that come with acoustic sensor, so noise monitoring will be easy and accurate.
It will be an interesting experiment.

 

[Tomgang]

Thats's a beast of a dual cpu/mb/gfx system in one case (if I've ever seen one). Air-cooling two cpu/mb/gfx combos in one case is one heck of a thing that you have done here. Really awesome

Thanks. It was from the beginning of this build it was ment to be aircooled. I chose zen 3 for this as it's easy to aircool and I wanted my system to be different from other dual systems. All dual systems I have seen, had one in common. They were all ways water cooled. Either aio or custom loop. So a dual system is rare, but a aircooled dual system is even more rare and I like it like that.

Aircooling this system also puts on some limitations. Specially for the mini-itx. The cooler size for one. But 5600X is a cpu relatively easy to cool at stock, even with a low profile cooler. It limits the oc, but allows a little bit. Also gpu size is a thing to consider, so it doesn't choke the cpu cooler fan. So I am limited to low profile gpu's. So very limited there and the fastest card there is the rtx a2000 That has gaming performance between rtx 3050 and rtx 3060 none TI. All that performance at only 70 watt. So it's powered by pcie alone. Also needed as my psu only has 4 x 8 pins connectors and since rtx 4090 takes them all. A2000 really was the only right choice. Else it would have been gtx 1650 with gddr5 or rx 6400 as alternatives. There has been a lot of things to consider when I designed this build and also I had to make some work arounds my self and modify things. This system is powered by only 1 psu, so you can maybe guess that requires some alternative thinking. But when it was solved. It has worked apselutely great.

If you want to know more or want more details of my system. I suggest you tjeck out my project build log after others requested I made one.

Tomgang´s project 2 systems in 1 case. All air cooled.

I was asked to make a build log for my new project 2 systems in 1 case. You can see all parts that goes in to this build in the link: https://www.techpowerup.com/forums/threads/the-show-off-your-tech-related-purchase-thread.225885/post-4537865 The first image of the mini-itx system up and...

www.techpowerup.com

 

[Dirt Chip]

In general, water can keep you on acceptable temp longer than air if you didn't build the setup right (also when all is OK but).
But that dosent mean that the answer is 'of coures get water' because if you didnt do the setup right your also losing cooling performance with water.
It just enable you to postpone the problem effects- not eliminating them. So you end up paying more on water, while exposing yourself to new set of problems as discribed later with evidance, just because you dont want to deal with proper arrengment of your case to get good airflow. If you disregard airflow in your case and companstet with overkill AIO cooling, you are more prone to also disregarding other factors concerning cooling (you just don`t want to deal with tham also) so you may avoid one problem but while so create an equal or bigger one.
Also, water don`t come with total freedom of placement- you are restricted with radiator placement relative to the pump. In a wrong setup you end with noise for start and later on with damaged pump.
All in all, if I can get the desired cooling preformance with air I see no reson to go water when factoring all the possible problems of leakage, pump malfunction and debree build up that you neet a lot of effort toclean it even possible.
I think that above all, PR, the need to sell and beauty look of the rig play a big roll when the average user decide to go with water by default.

I don`t oppose to water cooling in principle and I might do on myself if I will ever go with a high-end Threadripper or equal highly multi threaded system, but in only small number of cases you are really need to go water. This number is far smaller than what we see in the market and is pummep by the need to sell above all. It's not the consumer preferred choice ifcalculated objectively.

Plus- one area where I see the opposit situastion- that is choosing air when water is far better- is with the +350w GPU`s.
For some reson many go with water on a 100-200w CPU but go with air on 400-600w GPU. Makes no sense. The is even more pronunce when factoring the obscene size of current air cooled 4090 gpu`s. Water will save tone of space, will enabel much more clear and unobstructed airflow route, will enable more safe (read un-bent) and clean cable routing and in not few cases will save you from needing to upgrade to a bigger case that can accommodate those heumangus, overly sized, monsters.

So, invest you budget in water where you really need it and not only where it looks cool and in accordance to popular current fashion.

 

[Mussels]

Ah damn i made a huge post in the zen thread that belongs here too

This was on how people make the common mistake of simplifying what cooler they're getting "CM 212" and going by the reputation and results of an ENTIRELY DIFFERENT PRODUCT.

Below are the half a dozen baseplates the 212 variants have used over the years, and the pure heatpipe versions suck horribly on modern high heat density CPU's since you can get unlucky with a big air pocket or stripe of Alu right over the top of your CPU cores

(I left this open in my browser for like 8 hours, lol)

The Hyper 212 from 2007:
A solid slab of copper under the heatpipes.
It all goes downhill from here.
View attachment 269765

212 Evo: Copper slab gone, thinned back to just the heatpipes. That loss of solid copper hurts.
View attachment 269758

212x: This is the globally popular one that stayed with the close heatpipes that were effectively solid copper contact.
View attachment 269759



212 LED(white/turbo etc)
Flat in this image at least, but the spaces begin
View attachment 269760.View attachment 269761


Evo V2: This even comes in an LGA1700 version
More spaced out, but a terrible choice for multi chip CPUs
You can see the height difference between alu and copper
View attachment 269757



Black edition: (RGB variants the same) - It may have nickel plating on the bottom vs plain Alu, at least.
Gaps here exist but are very minimal
View attachment 269756View attachment 269763

212 RGB black: They stop showing actual images and give you renders only
View attachment 269762


They sold a bunch of shitty products with the name of the original, relying on customer confusion to buy the wrong one

I skipped a good 10 variants just because they're all the same

 

[Bill_Bright]

Half dozen? It would be easy to keep track of a half dozen. I wish it were that easy.

Cooler Master Hyper 212 EVO - Bing images

 

[eidairaman1]

ATX cases with mounts fwd, aft, top, bottom.

Intakes are fwd and bottom, exhausts are aft and top.

Use High static pressure fans as intakes when there are obstructions, use high flow fans as exhausts.

 

[Mussels]

Half dozen? It would be easy to keep track of a half dozen. I wish it were that easy.

Cooler Master Hyper 212 EVO - Bing images

half a dozen baseplates, the fans and colours vary a lot more

 

[Bill_Bright]

Yeah, because clearly, blue cools better than red!

But yeah, you are right about fans. And not just size but the number of fans too.

 

[Pictus]

Wonderful post and thank you @An0maly_76

I'm going with 13900k and nocuta nh-u12a cooling on fractal define 7 case. I know it might be too worm with the stock 2*140+1*140 and after I will do some testing I'm planning to add more 140mm fans to get better airflow.
The case is sitting on the floor so dust filters are on and I'm aming at a very quiet rig.
I have the aero z690 that come with acoustic sensor, so noise monitoring will be easy and accurate.
It will be an interesting experiment.

The Noctua NH-D15 has more mass and surface to exchange the heat, works better with high wattage CPUs.

Noctua NH-U12A Review

Noctua delivers top-tier air cooling performance in a compact design with the release of the NH-U12A. Sporting two of their latest NF-A12x25 PWM fans, what this cooler lacks in regards to RGB LEDs, it makes up for with brown and tan Noctua styling and a premium build quality that is second to none.

www.techpowerup.com

https://tpucdn.com/review/noctua-nh-u12a/images/temp_oc_aida64_fpu.png

To avoid the CPU to bent

A case with good airflow, at the 12:52 look at the Noise-Normalized Thermals.

IF you want better 140mm fans

IF you want better 120mm fans

 

[Assimilator]

What most air-cooling aficionados overlook is that liquid cooling allows gives you a lot more control over your entire cooling setup, by virtue of the fact that the radiator can be located far away from the source.

I personally prefer liquid simply because I've never been comfortable with hanging big heavy blocks of metal off a motherboard.

 

[Pictus]

What most air-cooling aficionados overlook is that liquid cooling allows gives you a lot more control over your entire cooling setup, by virtue of the fact that the radiator can be located far away from the source.

For power hungry CPUs I prefer custom water cooler and place the radiator/pump into another room or outside.

I personally prefer liquid simply because I've never been comfortable with hanging big heavy blocks of metal off a motherboard.

That is why my case(CM HAF XB EVO) is horizontal.

 

[freeagent]

I like the cost, the simplicity. If I want to change my cpu or GPU, or anything for that matter there are no loops to drain, no big disassembly, no planning, just do. I can see the appeal of a custom loop, but I see no appeal in AIO.

 

[sLowEnd]

I personally prefer liquid simply because I've never been comfortable with hanging big heavy blocks of metal off a motherboard.

Some pretty heavy copper coolers like the TRUE Copper (1900g *without* fan) worked alright, and as far as I know, no popular air cooler these days is anywhere near that heavy. Even big dual towers like the NH-D15 (1320g with fan) or the Deepcool Assassin III (1464g with fan) don't approach that kind of mass. It can be an issue for shipping, but good enough packing material can mitigate risk there too.

 

[freeagent]

Thermalright FC140 is 1000g without fans, it is a beast. I haven’t tried my TRUE from 2007 on my 5900X because TRUE 2022 got decimated. PA120 also gets destroyed by a tuned 5900X at full load, but at stock it’s fine. Those smaller coolers are good for 6/8 cores but lose their grip on the higher core count CPUs if you tune them with generous power limits.

 

[Mussels]

What most air-cooling aficionados overlook is that liquid cooling allows gives you a lot more control over your entire cooling setup, by virtue of the fact that the radiator can be located far away from the source.

I personally prefer liquid simply because I've never been comfortable with hanging big heavy blocks of metal off a motherboard.

This is why I prefer low wattage hardware, TBH.
And when i can't have low wattage hardware, I tweak the shit out of it for max efficiency.