Today was hot and my house doesn’t have AC so I spent today doing nothing other than laying around and complaining about how it was too hot to do anything.
One of those complaints led to a debate about air cooling vs liquid cooling.
That was pretty productive, for a hot day anyways, because I learned something about the world: there are who people buy into liquid cooling simply because they think it’s cool. By which I mean they’re stupid, and it feels like they are probably the majority. Liquid cooling computers isn’t strictly wrong, there are reasons why it is better, but it took me a while to come to the conclusion, using thought routes that a bunch of people who support liquid cooling certainly didn’t use. Eventually I’ll make the comprehensive essay I’ve dreamed of writing, but for now the concept of Communication vs Conveyance suffices: people can write a lot of things about how something works and how well it works, but other people only read the parts they understand (if they understand at all), and they perpetuate their own story from there. Of all the comparison tests out there, where the liquid cooler wins, there’s people who read “liquid cooling (in all its forms) is better than air cooling (in all its forms)”.
When so many of these things are titled and labelled “air cooling vs liquid cooling”, who can blame them? That’s what you said, right? And the liquid cooler(s) won? Then liquid cooling is better. Hey wait… it’s not better. Wait, what do you mean you weren’t talking about all liquid coolers anywhere and everywhere?
So anyways, I was sitting in my room in front of my computer, complaining to my good friend Laxeris that it was too hot to lay in bed and do nothing, too hot to think, and too hot to play videogames, when he stated the extremely curious idea that liquid cooling is better than air cooling… for the sole reason that it has liquid in it.
Which is absolutely fucking wrong. But how was he supposed to know? I only happen to know because I took classes on thermodynamics and more broadly engineering classes that helped me look at the physics/systems behind the specifics. His knowledge goes as far as the average person’s i.e. journalist’s understanding takes him.
And what do the journalists say?
EK Water Blocks (liquid cooling manufacturer)
“Air Cooling vs. Liquid Cooling”
NZXT (computer case/accessories manufacturer)
“Should You Cool a CPU with Air or Liquid”
PCWorld (magazine about computers)
“Liquid cooling vs. traditional cooling: What you need to know”
Tom’s Hardware (online magazine about computers)
“How to Choose a CPU Cooler”
Corsair (computer peripherals company)
“Liquid Cooling vs. Air Cooling: Which Performs Better?”
These are the top five non-video results from Google searching ‘ air vs liquid cooling ‘.
On the topic of the fundamental idea, the driving physics behind the reason these ideas exist to begin with, throwing any particular implementations aside, this is what I got from each of the results on air cooling vs liquid cooling:
EK Water Blocks:
“Liquids and solid materials are better conductors of heat.”
” […] water transfers heat much more efficiently than air.”
“[…] fans aren’t as efficient as water cooling […]”
(no comment on comparison between physics)
(no comment on physics at all)
Excluding the 4th and 5th results, the latter of which is “just” playing the SEO game at the cost of coherency for the English language, the first three claim liquid cooling is more efficient than air cooling because ‘liquid is a better conductor of heat’. Which is correct. Liquid is indeed a better conductor/transferrer of heat than air.
But do people know what that means?
Here’s an extremely simplified thermodynamics lesson, without symbols and without math, and it’ll still be better than all the “look at all these pretty parts from our associates you might want to buy” garbage.
The two laws of thermodynamics that are of interest to us are, simplified:
- If two non-isolated items have the same temperature, no heat is exchanged.
- Energy is never created or destroyed.
In the zeroeth law, “isolation” means it’s in a vacuum and not touching anything. Such a thing does not exist in the scope of our analysis of a computer; everything is connected at least by air. ” The first law is clear enough after the clarification that heat is a type of energy, basically saying that if something is hot it’s not about to be cool unless it dumps its heat onto something cooler. According to Wikipedia it’s also the first law states that heat only (naturally) passes from hot to cold… I always thought that followed from the zeroeth, but it’s also in common sense.
Understanding the concept of heat capacity is also helpful: it’s the amount of heat that can be dumped into something before it changes a certain amount of temperature. Example, water has high heat capacity and air low: with any given heat source, air will rise in temperature faster than water ‘because it has less capacity for heat’.
Beyond that… heat transfer between solid-solid and solid-liquid is better than solid-gas.
And that’s all you need to know.
So how do air cooling and liquid cooling work?
Air cooling: You have a part under load. A heatsink is directly attached to the part, and a fan is directly attached to the heatsink.
- Heat moves from part to heatsink.
- Heat moves from heatsink to air, fan speeds that up.
- We stop analyzation of the system at the air in your room.
Liquid cooling: You have a part under load. You have a loop of liquid with a coldplate at one end, a radiator at the other, and a pump somewhere inbetween.
- Heat moves from the part to coldplate.
- Liquid moves to radiator, pump speeds that up.
- Heat moves from liquid to radiator.
- Heat moves from radiator to air, fans speed that up.
- We stop analyzation of the system at the air in your room.
That’s right: “liquid cooling” is actually liquid and air cooling. There are a few other super fancy and exorbitantly expensive ways you could potentially do it, but look at any standard consumer liquid cooling system, and you’ll find a fan on it somewhere. That’s because the air as the final “heatsink” is cheap. Air is, as far as human activity is concerned, everywhere, and, as far as mechanations are concerned, basically always cold. So that’s where they dump their heat in the end.
The original question I was debating with Laxeris was whether or not a liquid cooled system would keep the room a computer is in cooler, which somehow also also got into whether or not a liquid cooled system would keep the computer itself cooler. His position is that both the room and the computer would be cooler.
If we assume that everything about the system is completely identical except for the existence of a liquid system, then a liquid and air cooled system will reach the exact same temperatures and exact same effects for the sole reason that the origin point and the endpoint are the same.
Heat is generated by the computer/part at the start, and heat is dumped into the air in the end. Liquid in the middle doesn’t make that heat disappear, it just means some heat can build up there because there happens to be stuff in the middle, “some heat” being “however much heat capacity it has until it reaches the same temperature as the part”. Its supposed efficiency over air, i.e. solid-liquid vs solid-gas, doesn’t matter because its final steps are exactly the same bottleneck, i.e. solid-gas. It’s not as if you have a river in your room and you stuck the radiator/heatsink in that. If you did then yes, all the logic stacks up, it’s a true liquid cooled system, one which has the same 3 steps instead of 5, and it is accurate to say it is superior “because liquid transfers heat better than air”. And it would be. You wouldn’t need to test it to be confident it had superior results.
But unless you have that river, or a gigantic lake, you’re going to run into the same problems in the end, because the end is the same: you’re still dumping heat, from a solid to a gas, in the same way as air-cooling. Better than liquid cooling is solid cooling; let’s put a giant metal block inbetween the part and the heatsink instead. Actually… since it’s the same anyways, why not just put the heatsink directly on top of the part?
Because that’s air cooling.
Not that any of these top results of self-purported “explanations” would tell you that. If anything what they say is deliberately misleading. It’s not apples to apples. Sure solid-to-gas is less efficient than solid-to-liquid. But if that’s true, why does your liquid system end with a fan? It feels like they’re reciting something they heard more than anything else.
Except the EK Water Blocks guys. Those guys knew exactly what they were doing.
“Secondary cooling surface”? Air-cooled systems don’t have a secondary surface. It’s heatsink to air and you’re done. This supposed air cooling system that they’re saying liquid is 2-10 times better than is a system which has the computer heat up air, air which then piped to a heatsink (the “secondary cooling surface”)… which presumably also dumps its heat with a fan into the outside air.
So basically, a liquid cooling system except the tubes are empty.
Or a computer inside a fridge.
What the fuck am I reading? Is this a serious publication by a manufacturing company? They’re inventing fictional competition which would absolutely never exist to promote their product technology?
That’s the other thing which floors me. It’s one thing that Laxeris and similar aren’t educated or aren’t interested past “just tell me which one keeps my temperatures lower and i’ll decide if the cost and hassle is worth it” and misremember things they don’t understand. It’s another thing entirely that people that do know what’s going on lie, hide, and pretend. Or are unimaginably stupid, but Hanlon’s Razor is for idiots.
This is the “explanation” between the two systems by PCWorld. A bunch of words. If you’d never really looked at liquid cooling before, you’d be fooled by this journalist into thinking that a “radiator” is something that’s fundamentally different from a “fans” and “a big […] heat sink”. Quoted earlier in this post and appearing later in that article, he says “fans aren’t as efficient as water cooling“. You think he doesn’t know that liquid cooling systems have fans on them?
After not bothering with Laxeris on the topic anymore because it was going in circles, I thought about why liquid cooling for computers exists at all. I could figure out why they existed for cars; engine blocks are huge and complicated and air can’t properly cool them, but with cases designing in cable management these days that’s not a problem…
I eventually decided that it was because one of my initial premises was not true: liquid cooled systems aren’t simply the same as air-cooled systems with a loop of liquid inbetween. They do have a fundamental advantage, not in that they use “liquid cooling is 2-10 times more” nonsense, but because they are flexible heat pipes that can allow remote and thus larger heatsinks. Look at actual liquid cooling products, their “radiators” are almost all gigantic pans, significantly larger than any heatsink you can mount on any component (There are some which aren’t, and I’ll bet those are strictly worse than air-coolers). There are more details than the cartoon model/diagram, like the existence of the computer case, and given that there’s no huge fundamental difference between air and liquid cooling, it’s these smaller differences which set them apart. The inside of a computer is only so large, but the outside of a computer is larger, and if you want to jury rig a clothes rack outside to support a bunch of radiators and fans, the flexibility of piping liquid makes that possible. Theoretically you could do it with solid metal blocks or heat piping, but they’re not flexible, and they’re exorbitantly expensive in comparison to some plastic tubes. Effectively, without liquid cooling, you must have heatsinks which fit inside a computer case (caseless is bad, trust me on this).
So the real reason, tl;dr, why liquid cooling would be fundamentally better than air?
“Our heatsinks are bigger”.
But that doesn’t sound as cool as “liquid cooling is 2-10 times more effective”.
One would think that such statements would instantly make people question why air coolers aren’t ancient history if a 200~1000% gap exists, but I suppose not. People will believe such a thing, then continue by the tens or hundreds of thousands to read and watch head-to-head comparison tests on this liquid cooler vs that air cooler, with differences coming down to 1~3°C. We need to test it man! It’s the scientific method! If it’s better by 1° then it’s better! And it’d be better because it’s just inherently more effective!
I mean, I suppose it’s not wrong. Liquid cooling is more efficient than air, all things being equal. Which they aren’t. But given so-and-so test results, it does appear that this liquid cooler is better than that air cooler. And it was a fun video. So I’ll buy the liquid cooler, giving that company some business. And I’ll remember to like, comment, and subscribe, giving these YouTubers some business. I’ll receive the liquid cooler in the mail in a few days, making the temperature monitors show lower temperatures, and most importantly of all, I’ll feel satisfied about my purchase, all the interactions which led to it, and that one science-y sounding bit explaining these results, and science is cool, so science is what I’ll tell all my friends: “liquid cooling is 2-10 times more effective.”
I don’t know the textbook example of marketing success, but I imagine this would be it:
“Everyone involved goes home happy having gained more “value” in their lives.”
Everyone, except me.
I feel like I should know a bit about how things work, and I feel that things like these show that there’s a lot of lying going on unseen in the world. I was able to identify this one after quite some work, but what about the ones I can’t identify? Or have failed to identify? How do I defend against things which I not only don’t know of, but don’t know the nature of? I could tell that a fan advertising itself as “25% Quieter” without saying what it was compared to is meaningless. But now I’m not so sure that fans which list their exact noise value in decibels mean much either. Do they have a different definition of “decibel“?
I’m not happy about that. I did not gain any “value” from having thought this problem through. More than a market for lemons, it’s a world for lemons. If I had to use vocabulary I’m not used to to refer to things I’m not familiar with: Is this the cost of refusing postmodernism?
But in the end, we all know what they say about the constant through all your problems:
so maybe, just maybe,
or maybe absolutely certainly,
I’m the one who doesn’t get it.