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

[AlwaysHope]

Instead of being pedantic we could understand that the difference is liquid has the ability, by nature, to quickly and constantly move heat away from the source, whereas air coolers (you know what I'm talking about) have to use heatpipes or vapor chambers to do so (guess what, there's liquid in these). Its getting harder and harder with the increasing transistor density of CPUs especially, to do this.

On a liquid system you can add as much surface area as you want through radiators, there are no limits, assuming you have a pump strong enough, or several in series. On an air system, you are limited by the thermal conductivity of the heatpipes/vapor chambers, which decrease with more distance from the heatsource, fundamentally limiting the effective size of any air cooler heatsink. Liquid also has the ability to have a large thermal reservoir, serving as a buffer for spikes in thermal load. There's also the advantage of liquid cooling systems to directly exhaust the heat outside of the case, instead of blowing it on the other components (like RAM), or GPUs exhausting very hot air directly into the CPU cooler.

Additionally, you can position the various elements of a liquid cooling system anywhere in the case, for aesthetics or space efficiency. Air cooling forces you to have your heatsink directly on top of the heat producing element - leading to absurdities like 4 slot GPUs, or huge tower coolers forcing large case sizes.

There is a good reason almost every serious heat producing element in any part of any industry or product (cars, industrial machinery, nuclear reactors etc.) all use liquid cooling. Air can potentially be lower maintenance, but it depends on the design of the loop, air is fundamentally a lower capacity, worse cooling system. The advantages are simplicity, cost, and to a certain extent, lack of maintenance.

You can call it pedantic if you like, but no getting away from the fact that all things technical are pedantic by nature. Every variable has to be considered, if that's pedantic then so be it.
There is also no getting away from the fact that air is still needed to cool the radiator(s) of a given water cooling system.
What happens if you get a situation of having hot air being forced into a radiator to cool the water cooling loop via the fans? it is not going to work very well isn't it? Unless of course you have air conditioning & then we have the issue of electricity costs coming into the total cooling solution & that's another argument in itself.
So in effect the entire situation gets back to how well the ambient air temps in the space or room that the water cooling solution is placed in ultimately influences the water cooling efficiency result.

 

[dgianstefani]

You can call it pedantic if you like, but no getting away from the fact that all things technical are pedantic by nature. Every variable has to be considered, if that's pedantic then so be it.
There is also no getting away from the fact that air is still needed to cool the radiator(s) of a given water cooling system.
What happens if you get a situation of having hot air being forced into a radiator to cool the water cooling loop via the fans? it is not going to work very well isn't it? Unless of course you have air conditioning & then we have the issue of electricity costs coming into the total cooling solution & that's another argument in itself.
So in effect the entire situation gets back to how well the ambient air temps in the space or room that the water cooling solution is placed in ultimately influences the water cooling efficiency result.

It's pedantic because you're stating the obvious as if it's a point. Liquid cooling being "air cooled" because it still uses air and radiators is pedantic. You're not considering a variable by stating this, you're stating a fact that is obvious to everyone.

What happens if you get a situation of having hot air being forced into a radiator to cool the water cooling loop via the fans? it is not going to work very well isn't it? Unless of course you have air conditioning & then we have the issue of electricity costs coming into the total cooling solution & that's another argument in itself.

Hence the advantage of being able to place your radiator wherever you want to draw in cool air and exhaust hot air out of the pc case. If you're simply talking about ambient air temperature, then this is more of a problem for air coolers than it is for liquid coolers. Liquid coolers by nature run much closer to ambient temperatures due to increased efficiency.

So in effect the entire situation gets back to how well the ambient air temps in the space or room that the water cooling solution is placed in ultimately influences the water cooling efficiency result.

Thanks Sherlock.

You can be detailed without being pedantic.

I think the novice builder should focus on having all their fans pointed the right way and plugged in, that's always a good place to start.

 

[AlwaysHope]

It's pedantic because you're stating the obvious as if it's a point. Liquid cooling being "air cooled" because it still uses air and radiators is pedantic. You're not considering a variable by stating this, you're stating a fact that is obvious to everyone.

Hence the advantage of being able to place your radiator wherever you want to draw in cool air and exhaust hot air out of the pc case. If you're simply talking about ambient air temperature, then this is more of a problem for air coolers than it is for liquid coolers. Liquid coolers by nature run much closer to ambient temperatures due to increased efficiency.

Thanks Sherlock.

You can be detailed without being pedantic.

View attachment 268788



I think the novice builder should focus on having all their fans pointed the right way and plugged in, that's always a good place to start.

Sometimes the obvious has to be stated because in this case, H2O cooling adherents overlook this plain fact and waffle on justifying the expense they have invested in a H2O cooling solution. Just FYI, air cooling is used more extensively than you think. Although that link is associated with air cooled engines, the principle doesn't change & too think, one of the most successful sports cards of the 20th century had an air cooled engine.. aka Porsche 911 at least until 1998.

 

[bobbybluz]

As a sidenote to this discussion I've been mounting the radiators for my AIO's externally for years now. If I was home at the moment I'd take pictures to post (I did post a pic a while back of the NZXT H440 I externally mounted an Arctic Liquid Freezer II 360 on top of after some Dremel surgery. My most recent is a LF II 420 externally on top of a Cooler Master Cosmos II) of some of them. External mounting has advantages including a few that physically can't be done with air coolers.

I've discussed this at length with Arctic and not only do they agree with my reasons for doing it they also said it's an excellent idea for the extremely hot running CPU's we have today. Of course case modding skills and the proper tools are needed to do this and it's not for a beginner. I do have one of my Z690 rigs temporarily set up with a LF II 280 sitting vertically on the top of a Corsair 300R case minus the right side panel (I may make a Lexan panel or mod the stock side panel for hose clearance and adding a 140mm fan) that runs amazingly cool with an i5 12600K @5.3GHz and a XFX RX 6800 XT Merc 319. No modding at all was needed to do this one so far.

 

[Mussels]

Try ocing a 5800X and not get it to 90C...

You do know i've made plenty of threads doing exactly that? They're in my signature.

1. Always be sure you know what you're talking about
2. Double check who you're saying it to

standard PBO ranges were around 60C.

I should bench my ITX rig on it's little 240mm AIO now it's got the 5800x

Since you politely asked:

ITX rig
240mm EK AIO with MX4 - no liquid metal on this one
RGB Jellyfish for extra performance
27C ambients at the start of the test before someone pissed off thor and it began bucketing hail at me, about 25C and humid when i stopped

15 minutes of R23 because a massive thunderstorm/hailstorm combo just hit out of nowhere and the house is shaking and the roads are flooding :/

And CPUz to cover the 4.8 boost (No +200) on this one

 

[Ruru]

Try ocing a 5800X and not get it to 90C...

Does someone manually OC their Ryzens? I've been running my 3600 with PBO +200 since I got this.

 

[Mussels]

Does someone manually OC their Ryzens? I've been running my 3600 with PBO +200 since I got this.

I did a 4.6GHz OC on the 5800x for a long time, it's well known that they run a lot colder that way than with boost/PBO - but you clearly lose some single threaded performance doing so (4.85Ghz to 4.6 was pretty minimal, for the lower heat and especially sustained clocks vs the stock 4.4 all core)

 

[Vayra86]

This is all true, but the VF curve being nonlinear means that the added capacity of liquid cooling doesn't do much for actual performance as proven by TPU in the 7950X cooling test. The clincher for me though is that air cooling is vastly more reliable than cooling using electrically conductive liquids.

Exactly, that thermal reservoir is for most intents and purposes completely pointless exactly because of the thermal density of current chips. They just jump to their peaks, no matter what, hell, Zen is now even designed to keep at peak temp. The differentiator there is ambient temp.

Even liquid today is vastly limited by the weakest link: thermal transfer out of the die. Its why people delid.

The advantage of liquid on today's CPUs is quite limited, dynamic boost technologies fix a lot of the problems before cooling comes into play. Sure, you can push a continuous load for longer on water at a higher wattage, and if that's the use case, then certainly liquid offers the advantage.

There is a good reason almost every serious heat producing element in any part of any industry or product (cars, industrial machinery, nuclear reactors etc.) all use liquid cooling. Air can potentially be lower maintenance, but it depends on the design of the loop, air is fundamentally a lower capacity, worse cooling system. The advantages are simplicity, cost, and to a certain extent, lack of maintenance.

Car analogies are always problematic, and they are here too. Most PC's are idling and then drag race to the finish, to go idle again. Cars don't, and cars also have the advantage of 'increased air intake' when they increase their speed. PCs really don't, unless you put an array of Delta fans in there and are ready to go for their limit.

About pedantic or not... this topic is all about that, to be fair When we're in the territory of high performance cooling, anything goes. And that really is the segment where liquid comes into play - in a custom loop - and below that air is much more cost effective without much if any perf sacrifice.

 

[OkieDan]

I like my AIO because I can control fan speed based on liquid temps, which takes a pretty heavy load a few minutes before the fans speed up.

The 4000D Airflow comes with filters BTW.

 

[Vario]

About pedantic or not... this topic is all about that, to be fair When we're in the territory of high performance cooling, anything goes. And that really is the segment where liquid comes into play - in a custom loop - and below that air is much more cost effective without much if any perf sacrifice.

A good point, one would have to spend a few hundred dollars for custom water over an inexpensive air cooler and the gains are so marginal that a reasonable person wouldn't notice them. They would, however, notice if the money for the watercooling was instead spent on the next higher SKU part(s).

 

[Bomby569]

the thing that bugs me most on the air cooling, is it makes working on the pc much harder, getting to those small places, fan connectors or the cpu 8 pin power for example, or even get the gpu out, reaching for the gpu release hatch. Apart from that mount and forget.

 

[ThrashZone]

Hi,
Large chunk of aluminum
I have one d15 in my entertainment center using a test bench it's about the only way I can deal with a air cooler that large and nope memory access is not going to happen still seeing it's on a x99 with sticks on both sides of it maybe I've not tried yet

 

[DeathtoGnomes]

Bill is right about all this but the single most important factor I've noticed is cooling inside the case itself.

I said that too, it cant be said enough, but it can be said in several different ways.

 

[Bill_Bright]

Car analogies are always problematic, and they are here too. Most PC's are idling and then drag race to the finish, to go idle again. Cars don't, and cars also have the advantage of 'increased air intake' when they increase their speed. PCs really don't, unless you put an array of Delta fans in there and are ready to go for their limit.

I think car analogies work very well "most" of the time. And that is because most users are much more familiar with cars so they can relate. Yet electronics are magical to many. Not to mention, some seem to think electronics and computers are totally different - but that's a different issue.

As far as variable fan speeds - I have to disagree with you there. CPU, PSU, GPU and many "system" fans commonly do indeed, vary in speed, depending on need. Not all, of course, but many do.

That said, to your point about car analogies being problematic, in this case, I agree. Liquid cooling is used in cars for several reasons but one big one is the engine block - a massive hunk of metal with a bunch of precision, moving parts that need to be "evenly" cooled to avoid warping, separations, and binding. Engine coolant is "piped", under pressure, throughout the engine block and heads to places no forced air can reach - and that is a critical difference.

Other differences include the fact a car engine is designed to operate at a specific temperature, typically between 195 and 220°F - which is why car engines have thermostats in the cooling system. Plus a big tank of water (the radiator) takes a lot more energy to heat up, and longer to cool down. In that sense, it acts like a capacitor to help avoid rapid changes.

Then there is us humans. We like to be warm in our cars in the winter. So hot water is pumped through the heater core (basically a tiny radiator) inside the cabin through which the blower motor blows air to keep our little piggies warm (and windows clear).

 

[Shrek]

I wrote that the military could dissipate engine heat through air cooling, but it got deleted as 'off topic'

 

[qubit]

@An0maly_76 Great info you've got there, very useful for newbies and perhaps not so newbies.

Note that you should explain that air cooling works as an offset of the room temperature, eg a 20C room could have the CPU at 40C say, while a 30C room, 10 higher at 50C. Therefore, you can't say that the CPU should necessarily idle at around 40C without specifying the room temperature. If the room is already 41C, say (global warming, hot country) then 40C is literally impossible and it's likely to sit at 55-60C with flat out operation potentially overheating.

 

[caroline!]

Air cooling works when you're not running an extreme overclock.

Liquid cooling is better but it's also expensive, in the order of 10 times more expensive than air cooling. I'm obviously not counting gaming AIO liquid coolers because they're the same as an air cooler, just with extra steps, proper liquid requires good parts, and good parts cost money. OR, if you don't mind the looks you can pretty much build a working loop out of a dead AIO and a fish tank pump, I've done it, works wonders with overclocked chips.

 

[ColinB123]

Missing from the OP's Guide, is any mention of the GPU.

In my case (literally), an R5 3600 produces 65w, but the RTX3070 up to 240w and dumps a high percentage of that into the case. The guide doesn't discuss running the GPU at max speed, or even how it can be monitored to drive case fan speeds. I have one set of case fans (in/out) assigned to CPU temperatures, and another pair to GPU temperaures, and try to partition the case so that the airflows are not clashing. Spare temperature sensors can also play a role.

A large tower air-cooler takes up a lot of space in a mid-sized case, and an AIO would free that space up, making for better airflow to support the GPU airflow cooling.

 

[Mussels]

Exactly, that thermal reservoir is for most intents and purposes completely pointless exactly because of the thermal density of current chips. They just jump to their peaks, no matter what, hell, Zen is now even designed to keep at peak temp. The differentiator there is ambient temp.

Even liquid today is vastly limited by the weakest link: thermal transfer out of the die. Its why people delid.

The advantage of liquid on today's CPUs is quite limited, dynamic boost technologies fix a lot of the problems before cooling comes into play. Sure, you can push a continuous load for longer on water at a higher wattage, and if that's the use case, then certainly liquid offers the advantage.


Car analogies are always problematic, and they are here too. Most PC's are idling and then drag race to the finish, to go idle again. Cars don't, and cars also have the advantage of 'increased air intake' when they increase their speed. PCs really don't, unless you put an array of Delta fans in there and are ready to go for their limit.

About pedantic or not... this topic is all about that, to be fair When we're in the territory of high performance cooling, anything goes. And that really is the segment where liquid comes into play - in a custom loop - and below that air is much more cost effective without much if any perf sacrifice.

This is pretty much it
Zen3/4 on the smaller die process are entirely limited by how good the CPU is soldered to the IHS, and then its all about how good your mounting is and how good that contact is
You dont need bigger coolers, you need better ones

And the market is saturated with low quality junk designed for bigger CPU dies, so a lot of coolers simply suck on these CPUs - a giant heatsink or radiator means nothing if the heat never reaches it

Missing from the OP's Guide, is any mention of the GPU.

In my case (literally), an R5 3600 produces 65w, but the RTX3070 up to 240w and dumps a high percentage of that into the case. The guide doesn't discuss running the GPU at max speed, or even how it can be monitored to drive case fan speeds. I have one set of case fans (in/out) assigned to CPU temperatures, and another pair to GPU temperaures, and try to partition the case so that the airflows are not clashing. Spare temperature sensors can also play a role.

A large tower air-cooler takes up a lot of space in a mid-sized case, and an AIO would free that space up, making for better airflow to support the GPU airflow cooling.

Air temperature in cases evens out incredibly fast, a CPU at 100C would cool down to ambient air temps before moving a centimeter off that heatsink

You need to set the system up for air flow, simply slightly more in than out, blowing across the entire system and out. The temps even out so fast that a 400W GPU would barely make any difference to a CPU, and this HAS been tested and proven

3080 FE blowing its heat INTO an air cooler lowered temps and not raised them, because airFLOW is what matters regardless of if it being warmed up or not first

 

[claes]

TBF, as debau8er points out, the FE blows air, including the CPU heat, out of the case in a way axial coolers don’t.

 

[Bill_Bright]

because airFLOW is what matters regardless of if it being warmed up or not first

Exactly! Well, almost.

I think it slightly more accurate to say, "airFLOW is what matters 'most' regardless... ."

I say that only because ambient (room) temperature "may" play a role here - potentially a significant one "IF", for example, you live in the Sonoran Desert in Arizona in July and August and your AC is out.

But I quickly add that same ambient temp affects both air and liquid cooling systems.

You need to set the system up for air flow, simply slightly more in than out

Air temperature in cases evens out incredibly fast, a CPU at 100C would cool down to ambient air temps before moving a centimeter off that heatsink

Yup and yup! I agree 100% on both points. While a CPU can go from cool to overheated in just a few clock cycles (and remember, for many CPUs there are 3 billion+ clock cycles per second) it can cool down almost as quickly, when air flow is properly setup.

 

[ThrashZone]

Hi,
Only thing I don't like is saying something is a "myth" "without all context included that you're refuting" and just adding a 1-7 word sentence as the myth you attempt to argue against.

Then tying it up with a bow with "This simply is not true in 99.99% of cases" as your conclusion

Good cooling tips but a bad incomplete example of a myth.

 

[Shrek]

While a CPU can go from cool to overheated in just a few clock cycles (and remember, for many CPUs there are 3 billion+ clock cycles per second) it can cool down almost as quickly, when air flow is properly setup.

That is a few billionths of a second; a CPU can't go from "cool to overheated" on that time scale.

 

[damric]

Sorry about taking the thread OT, but that jellyfish thing looks awesome @Mussels

 

[Count von Schwalbe]

That is a few billionths of a second; a CPU can't go from "cool to overheated" on that time scale.

Dang it Shrek, don't do this to me. A 13700K* will go from 20C to 100C, or thermally throttling in an air-conditioned room, in 92 ms. This is equivalent to around 472 million clock cycles for the P-cores, and around 361 million E-core clock cycles at full turbo. Note that this is still less time than it takes you to blink once.

*Assuming pure silicon, a die .5mm thick, PL2, and no heat loss

Edited, thanks to Shrek catching my math error.