• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

CPU Air Cooler tested with same fans?

Joined
Jun 11, 2019
Messages
744 (0.34/day)
Location
Moscow, Russia
Processor Intel 12600K
Motherboard Gigabyte Z690 Gaming X
Cooling CPU: Noctua NH-D15S; Case: 2xNoctua NF-A14, 1xNF-S12A.
Memory Ballistix Sport LT DDR4 @3600CL16 2*16GB
Video Card(s) Palit RTX 4080
Storage Samsung 970 Pro 512GB + Crucial MX500 500gb + WD Red 6TB
Display(s) Dell S2721qs
Case Phanteks P300A Mesh
Audio Device(s) Behringer UMC204HD
Power Supply Fractal Design Ion+ 560W
Mouse Glorious Model D-
This has been definitely brought up by others before but wouldn't it be interesting to test the best modern air towers: D15g2, Phantom Spirit Evo, MP7, A620Pro - but with the same fans attached? Strap Noctua's newest, Arctic'c cheapo classic and some others and see whose tower is actually the best and how strongly do the fans affect the end results? Noise-normalize one set of results to 35db and another with fans @full speed. If you saw something like this on the internets, please link here.
 
Haven't seen anything to link here concerning using the same fan on all those coolers.

I would, under assumption, think the cooler with the most heatpipes and fin count (most surface area) would work the best given the same fan used on all 4 different coolers.
 
I avoid Thermalrights lower RPM 120mm fans, but anything that revs to 2K is a good balance of pressure and movement, on their AIO.

When running the EVO, I will load it with T30s for 2 reasons. 1, It will accept them with no jank, 2., not many consumer fans will outperform a T30.

On their bigger coolers, like FC140, it works well with the stock fans, as well as T30s.

But AM5 gets better coverage on the 7x 6mm pipe with 5 covering vs 5x 8mm pipes with only 3 covering.

I do not have a new Noc G2, no plans on buying one so there is that.. maybe if I ran an Intel just because of the different package layout, but I am unfamiliar so cannot say.
 
Short answer, no. Longer answer, physics.

Functionally speaking air towers transfer heat almost exclusively by conductive means, rather than convection. Q/t = kA(ΔT/L) This means you can increase heat transfer (Q/t) by material choice (k), the difference in temperature between the heat source and surrounding medium, the area of the fins, or the thickness of the fins.

The practical limitation of this is simple to imagine. The fins are made of a metal, need to have structural rigidity, and cannot be packed too close together lest they start entering each other's fouling layer. If you get to the point where the fouling layer is too close together you get situations where the static pressure cannot overcome internal resistance, and it fails. This is why most liquid coolers, which are a honeycomb of fins, require high static pressures to get flow through (but because the area is huge and the fins are thin you don't have to have an enormous volume of air. You could likewise have one enormous plate, or visually a tube folded on itself of enormous surface area....where flow velocity is going to matter more because the flow inside the tube will functionally move enough heat with that huge area and relatively thin tube diameter. The problem is that each design requires a balanced fan, meant to weight noise, power draw, flow volume, and static pressure to optimize performance.

Basically, you are proposing strapping a Mazda Miata drivetrain to a bunch of sub-compacts and medium sized cars to see what car is best with what engine...when the complex workings mean that whatever fan they have should instead have been specified for them. The only exception being the ultra budget...where a customer specified fan would probably eat into their budget too far (and thus you wind-up with whatever was good enough an contractually cheap). If that's your goal, power to you. That said, what I see is an enormous amount of testing to discover that any competent tower should not experience a huge bump in performance with a new fan, unless it either generates more noise, is cheaper, or it is price optimized.



Good luck on testing this though. 30-100 USD per cooler, another few hundred in fans, and about 3k to get a measured and monitored constant heat source rig able to demonstrate consistent temperature measurement to within fractions of a degree is...a pretty wild situation.
 
Basically, you are proposing strapping a Mazda Miata drivetrain to a bunch of sub-compacts and medium sized cars to see what car is best with what engine...
Thanks for the explanation! That was enlightening.
However, I'm not very smart, so the quoted part is exactly the type of stuff that piques my curiosity, hence the original question, haha.
On a more serious note, what kicks off my thought process is the following: in GPUs and PSUs manufacturers almost universally cheap out on the fans and people get better results temperature- or noise-wise when they swap them, even when they swap to a more premium option of the same diameter, not bigger. So it would at least be interesting to see if anyone cheaped out on the fans enough for the results to be noticeable on the various air towers.
 
This has been definitely brought up by others before but wouldn't it be interesting to test the best modern air towers: D15g2, Phantom Spirit Evo, MP7, A620Pro - but with the same fans attached? Strap Noctua's newest, Arctic'c cheapo classic and some others and see whose tower is actually the best and how strongly do the fans affect the end results? Noise-normalize one set of results to 35db and another with fans @full speed. If you saw something like this on the internets, please link here.
It would be interesting but what would be the purpose?
Daily use? Because the inclusion of a noise level suggests that.
Overclocking/benching? Because the full rpm suggests that.

For daily use swapping fans isn't really budget friendly, unless you're repurposing the original cooler fans and use them in the case or something.
For a particular cooler putting the optimal fans on it, which would offer the best performance for a certain noise level, makes sense but if those fans need to be bought separately that diminishes the value of that cooler especially if the fans are not cheap.
Furthermore, if the difference with different fans is significant vs stock, it makes the product (in its stock form) less desirable as you could probably get similar performance for lower price by buying a different cooler that doesn't need different fans. Of course the final price plays a big part in this. And the more expensive a pimped out air cooler gets the more desirable a stock AIO becomes.

What you're asking about is somewhat similar to what poparamiro from lab501 was doing some 10-15 years ago. But the purpose of his testing was overclocking/benching even if he did test fanless and low rpm, however noise levels were irrelevant to him in those tests.
Some examples:
Oldies but goldies.

I wouldn't bet on something like this getting made today considering the amount of normies/casuals and toobers. Less and less people want to tinker, they want quick solutions: enable XMP, move that slider etc.
 
I have the Original TRUE, Le Grand Macho RT, and True Spirit 140 Power, and they do not work as well on modern platforms. I would even be as so bold as to say performance is not good at all with those old coolers. I was able to run a 3600XT at 4500MHz static clock and voltage semi passive with LGMRT. But Zen 3 is different.
 
I have the Original TRUE, Le Grand Macho RT, and True Spirit 140 Power, and they do not work as well on modern platforms. I would even be as so bold as to say performance is not good at all with those old coolers. I was able to run a 3600XT at 4500MHz static clock and voltage semi passive with LGMRT. But Zen 3 is different.
The examples I've given were not intended to imply that those coolers should be used on modern platforms.
The intent was to highlight the methodology used at that time and how that could be used as a reference for a new methodology using new coolers.
 
Thanks for the explanation! That was enlightening.
However, I'm not very smart, so the quoted part is exactly the type of stuff that piques my curiosity, hence the original question, haha.
On a more serious note, what kicks off my thought process is the following: in GPUs and PSUs manufacturers almost universally cheap out on the fans and people get better results temperature- or noise-wise when they swap them, even when they swap to a more premium option of the same diameter, not bigger. So it would at least be interesting to see if anyone cheaped out on the fans enough for the results to be noticeable on the various air towers.

You are not wrong per se, but there's a solid reasoning.

The internal specification for a GPU or CPU is basically that at ambient temperature T1, with maximum temperature of the device T2, design the cheapest thing that will meet performance numbers. Note that I did say a cost optimized solution is one case where better fans would make a difference. The endgame for the founders or base model is that you get virtually no overclocking headroom, and the thing runs to the hottest temperatures it is designed for (potentially thermally throttling if your room temperature is too high).

Now, pretend you're a third party. Base price is X. If you add in an upcharge of Y, you could add $20 to the price of a thing when Y is $40, and you've just added a huge amount of profit to your bottom line. That doesn't make sense for the base model, which is meant to sell as cheaply as will be reliable, but there are a ton of consumers out there who'll gladly pay another $40 for a card they class as superior because of a few decibels less noise or a few degrees running cooler that the consumer themselves might take a risk and overclock with....or simply be fine doing nothing with to satisfy their perceived increase in longevity.


I'll say this another way....so we can have some fun. The base model is usually the McDonalds cheeseburger of the options. You could get something that's a throwback and maybe pay a little less (think no-name roadside fastfood), but there's also the Burgerking (flame broiled), high end (Five Guys and Shake Shack), or even the sit-down silliness of a Red Robin. They're all at their core a cheeseburger. That said, you can dress up a McDonalds patty with toppings from Five Guys, high end cheese from Shake Shack, and a bun from a local bakery. It comes to a point when you either believe you know better, or have a different set of goals than the people who set things up. I...vote that for the most part.
Unfortunately, it was already said in 2017...that any competently built unit can cool within marginal error of virtually any other. TPU linked air cooler roundup on a German site. My money is that as long as I don't buy an enormous metal mass (the likes of Dark Rock and Macho had too much thermal mass, and were designed for a much higher TDP processor) I'll get within a few degrees of any other competent cooler...and the make it or break it could simply be how competently I get a thermal interface between CPU and cooler.
 
I've had Zotac Gtx 1070 ti mini, and I've destroyed stock fans, slapped noctua industrial 140mm fan, and the card was cooler and quieter, stock fans where funny, and the fans where not melted through enough, cause I've seen the balls that are inserted to fan moulds inside the stock fan (it looked like cheap chinese metals after failure that are from granulate).
About the gpu and checking different fans, look for deshrouding some ppl still slap pc fans to gpu, and it works really great, lower noise and temps.
 
Back
Top