Water cooling vs air cooling ,which is better?

[crazyeyesreaper]

It all depends on cash spent.

How much effort one wants to put into it.

Typically an AIO and high end air coolers perform about the same. High end air tends to be quieter. AIO tends to be nicer looking.

Typically I lean toward a few select heatsinks for air AIOs well i tend to just pick a few of the quieter offerings and leave it at that.

Niether is actually better than the other. if CPU thermal throttle temp is 100c and custom water gets you 85 with a high OC an AIO / high end air get you 90c your still 10 c below throttle meaning it doesnt matter. At least Intel side the CPUs are capable of high temps with little issue. As such it really depends on how high you plan to push the Overclock and how lucky you are in that regard ie Silicon lottery.

That said something like the CRYORIG R1 Universal offers similar high end performance to most AIOs with a slightly quieter fan profile and 100% memory clearance same as an AIO. as such in reality it all comes down to two things.

1.) Which do you like better Air or water
2.) How much do you want to spend.

 

[John Naylor]

1. Of course the best cooling comes from custom water loops as they ccan be designed to meet specific and targeted needs. Of course you have to define "best" . . . . is that "best temps at any noise level" or best performance / noise ratio ? I use the latter
2. Next up are the Open Loop type AIOs like the Swiftech units and the modular copper jobs from EK (phoenix series) where you pick each componnets just ike a cusrtom loop but you just connect the QD fittings together as all components are pre-filled.
3. In each and every case I have seen, the quality air coolers will outperform any CLC type AIO at comparable pricing with respect to both performance and noise.. But spending more doesn't necessarily bring anything home. The $90 flagship models from Noctua (NH-D15) and Cryrig R1 (Ultimate) are equaled or outperformed by the $37 Scyther Mugen Max and $45 Scythe Fuma
4. Coming in dead last are the CLC type AIOs whereby the cheap aluminum rads and weak pumps have to be offset by extreme speed fans . . . . and the teeny few that manage to top the better air coolers can only do so with unacceptable noise levels. For example the Cryorig A80 at $125 and available just about nowhere, does manage 66C veruys the Fuma's 69C in testing but it is 3 times the cost and almost twice as loud.

 

[Vayra86]

For my new build I went for big air in the end. I just like my stuff hassle free and pump noise does not add to a silent rig.

Silence, hassle free, basically I want to game without noticing there's a PC doing the work.

 

[bug]

Since people think the first thing a clueless overclocker should do is delidding, I'm going to leave this here: https://www.hardocp.com/article/2017/01/19/intel_kaby_lake_i77700k_cpu_delid_relid_results/

Water is better than air, but also way more expensive. Go water only if budget is not a issue, high-end air coolers will still get you 80% of the way.

 

[phanbuey]

Much of it also depends on your case flow. High-end air performs better than AIO on an open-air test bench/testing that is only doing a CPU torture test.

If you put an AIO as a cold-intake into your case, it will perform better than high end air when your internal case air starts getting hot (most silent setups) i.e. when gaming you have both CPU and GPU going - esp if you have a card that radiates heat into the case (non-blower design), the colder air the AIO is using will more than make up the difference vs a high end air cooler. Most silent cases don't move enough air to compensate for the heat a GPU can dump into the system under load - so just something to keep in mind.

 

[ZenZimZaliben]

My delid at same clocks/voltage was 18C drop across all cores. So I would say 10C is the lowest to expect but 15C+ is very possible.

Water Vs Air? Air unless you can afford a custom WC Loop...otherwise a good Air Cooler is pretty much the same as a cheaper WC setup.

 

[Vayra86]

Since people think the first thing a clueless overclocker should do is delidding, I'm going to leave this here: https://www.hardocp.com/article/2017/01/19/intel_kaby_lake_i77700k_cpu_delid_relid_results/

Water is better than air, but also way more expensive. Go water only if budget is not a issue, high-end air coolers will still get you 80% of the way.

This - and then most importantly that 'Bottom Line' in the article is something every advocate of the evil Intel using evil TIM argument seems to handily forget, omit, or just doesn't realize for lack of own experience.

Suffice to say, I won't be delidding my incoming 8600k If it does happen to be a dud, it'll get sold off and I buy another.

 

[ZenZimZaliben]

This - and then most importantly that 'Bottom Line' in the article is something every advocate of the evil Intel using evil TIM argument seems to handily forget, omit, or just doesn't realize for lack of own experience.

Suffice to say, I won't be delidding my incoming 8600k If it does happen to be a dud, it'll get sold off and I buy another.

I don't know or understand these Delid Down-play comments. Ok, not for you great....However it does far more than just lower CPU temps...it gets that heat build up out of the case. Cooler Motherboard, Cooler Chipset and Cooler Case. The overclock performed by HardOCP was just a straight up Multi increase...they did nothing on the Blck. Getting all of that extra heat out of the case is a good thing regardless of OC results. While Intel says it can take the thermal load with no issues...my other components neighboring the cpu socket were much happier to have that heat removed.

 

[EarthDog]

I don't know or understand these Delid Down-play comments. Ok, not for you great....However it does far more than just lower CPU temps...it gets that heat build up out of the case. Cooler Motherboard, Cooler Chipset and Cooler Case. The overclock performed by HardOCP was just a straight up Multi increase...they did nothing on the Blck. Getting all of that extra heat out of the case is a good thing regardless of OC results. While Intel says it can take the thermal load with no issues...my other components neighboring the cpu socket were much happier to have that heat removed.

It still has to deal with the same heatload...as its still putting out XXX Watts regardless of temps. An analogy for you.... which is hotter.. a lighter with a yellow flame or bonfire with yellow flames? A: they are both the same temp. It should be obvious which one has more energy to dissipate.

 

[ZenZimZaliben]

umm.... it still has to deal with the same heatload...as its still putting out XXX Watts regardless of temps. An analogy for you.... which is hotter.. a lighter with a yellow flame or bonfire with a yellow flame? A: they are both the same temp. It should be obvious which one has more energy to dissipate.

Ummmmmm.....I totally get heat transfer...The chip is still generating the same amount of heat..this is true, however it is not being contained in the CPU as long which mean the heat is not radiating to other components. Controlling how the heat is dissipated is the advantage of delid. I extract more heat to the outside of the case than someone with out a delid.

Also, that is a really bad analogy and has nothing to do with the movement and transfer of heat which is all delid and cooling in general are about...

 

[EarthDog]

So a bit less heatsoak through the cpu, but more of it is getting out....through the cpu heatsink which blows in the case.

Perhaps with a radiator situated on top this may be true, but, in my testing it was negligible, the differences you speak of. The heat has to go somewhere and soak aroind the socket isnt a big deal.

Edit: analogy was spot on for the point I was trying to convey (same heat output).

 

[ZenZimZaliben]

So a bit less heatsoak through the cpu, but more of it is getting out....through the cpu heatsink which blows in the case.

Perhaps with a radiator situated on top this may be true, but, in my testing it was negligible, the differences you speak of. The heat has to go somewhere and soak aroind the socket isnt a big deal.

Edit: analogy was spot on for the point I was trying to convey (same heat output).

My loop is external. I have 2 tubes going out of my case (and signal wire) going into another case, about 2 feet away, dedicated for my radiators, pump and res...so it truly is completely removed from the case. The only argument you could have is I am still dumping the heat in my office..except the WC Case is near the window.

 

[bug]

I don't know or understand these Delid Down-play comments. Ok, not for you great....However it does far more than just lower CPU temps...it gets that heat build up out of the case. Cooler Motherboard, Cooler Chipset and Cooler Case. The overclock performed by HardOCP was just a straight up Multi increase...they did nothing on the Blck. Getting all of that extra heat out of the case is a good thing regardless of OC results. While Intel says it can take the thermal load with no issues...my other components neighboring the cpu socket were much happier to have that heat removed.

Newsflash: the included TIM also gets "that heat build up out of the case", otherwise the temperature would constantly go up. So maybe you want to rephrase what you were trying to say, because I don't get it.

 

[EarthDog]

My loop is external. I have 2 tubes going out of my case (and signal wire) going into another case, about 2 feet away, dedicated for my radiators, pump and res...so it truly is completely removed from the case. The only argument you could have is I am still dumping the heat in my office..except the WC Case is near the window.

oh..my fault. I had no idea you were talking about yourself specifically. Seemed you were speaking generically...

Anyway, you have to admit however, your situation with the rads outside of the case is different than the vast majority with air, aio, and even custom water. But, i digress, that isnt the point. Im sure the cpu and socket area may be a couple c cooler in any case, but, thats really negligible in the scheme of things.

 

[ZenZimZaliben]

Newsflash: the included TIM also gets "that heat build up out of the case", otherwise the temperature would constantly go up. So maybe you want to rephrase what you were trying to say, because I don't get it.

Ok...Using a delid allows better surface contact between the Die and the IHS, this enables the heat to flow more efficiently through the IHS into my CPU block where it is removed from the system. Meaning I am able to more efficiently remove the heat from the CPU to the outside environment. The delid is a 2 sided attack. 1. you remove heat faster. 2. by removing heat faster you hope you will be able to push the overclock higher or use lower voltage...I say "hope" because even if I am not thermally locked I might be locked on the voltage side. Either way more control over the heat flow in your system is never a bad thing.

 

[bug]

Ok...Using a delid allows better surface contact between the Die and the IHS, this enables the heat to flow more efficiently through the IHS into my CPU block where it is removed from the system. Meaning I am able to more efficiently remove the heat from the CPU to the outside environment. The delid is a 2 sided attack. 1. you remove heat faster. 2. by removing heat faster you hope you will be able to push the overclock higher or use lower voltage...I say "hope" because even if I am not thermally locked I might be locked on the voltage side. Either way more control over the heat flow in your system is never a bad thing.

Yes and after you do all that, if you don't chip your chip, it will still overclock pretty much as it did before. But now it's cooler.

 

[EarthDog]

If one is able to use less voltage from 10-15c difference...they didnt test properly and wasnt using the lowest voltage already. Id only buy this with voltage read points and a dmm anyway, not software readings.

Typically, an additional overclock of 100mhz or so is what most will get out of it, if that. More than that is gravy/luck.

If you are voltage locked, chances are you are thermally locked anyway considering voltages needed to be voltage locked in the first place with these high core count and high tdp chips will overrun 3x120mm plus worth of rad...7700k isnt much different.

...or are you talking specfically about your situation again with 4x140mm of radiator for cpu only. What is your stable load voltage at 4.95ghz?

 

[ZenZimZaliben]

...or are you talking specfically about your situation again with 4x140mm of radiator for cpu only. What is your stable load voltage at 4.95ghz?

Well I can really only speak 100% factually about my own system. That said I was able to testbench at 4.8ghz (48x Multi) around 1.4v with temps in the high 80's and I couldn't go any further without chip protecting itself . Prior to delid my daily clock was 4.7 and 1.35v in the high 70's. Now I can run daily at 1.385v @ 4.95Ghz (45x multi with BLCK 110) with temps in the low-mid 70's and chip is happy. So the delid did only raise my OC by 195Mhz..but it allowed me to be stable at low enough voltages that I can be happy with it for a daily driver. I have tried 5Ghz but apparently there is a massive voltage wall there and am not comfortable running it around 1.45v even to test bench with. I am on my work pc so all of these numbers are from memory, but pretty dang close I think..can verify tonight.

My GPU is also on that loop.

Yes and after you do all that, if you don't chip your chip, it will still overclock pretty much as it did before. But now it's cooler.

Yes I guess if you are a clumsy fool then you shouldn't attempt this..or much else in life. Otherwise if you can spread mayonnaise on a piece of bread without cutting the bread or your finger off I would say you can handle a delid...it really is that easy.

 

[bug]

If one is able to use less voltage from 10-15c difference...they didnt test properly and wasnt using the lowest voltage already. Id only buy this with voltage read points and a dmm anyway, not software readings.

Typically, an additional overclock of 100mhz or so is what most will get out of it, if that. More than that is gravy/luck.

If you are voltage locked, chances are you are thermally locked anyway considering voltages needed to be voltage locked in the first place with these high core count and high tdp chips will overrun 3x120mm plus worth of rad...7700k isnt much different.

...or are you talking specfically about your situation again with 4x140mm of radiator for cpu only. What is your stable load voltage at 4.95ghz?

So 100Mhz when already running 4GHz+. A whopping 2.5%.

 

[Vayra86]

I don't know or understand these Delid Down-play comments. Ok, not for you great....However it does far more than just lower CPU temps...it gets that heat build up out of the case. Cooler Motherboard, Cooler Chipset and Cooler Case. The overclock performed by HardOCP was just a straight up Multi increase...they did nothing on the Blck. Getting all of that extra heat out of the case is a good thing regardless of OC results. While Intel says it can take the thermal load with no issues...my other components neighboring the cpu socket were much happier to have that heat removed.

The down playing comes from the fact that the performance gains of a delid are minimal, while you do void warranty, and it is a(nother) time, cost, effort, and risk factor. Its a payoff for 100-200mhz you must be willing to make, and for that meagre real-world gain, I'm not willing to - yet.

The pros of a delid are overstated in a way - yes you gain a great temperature drop. And a minimal - if any - additional clockspeed. I'm just calling things as they are.

Another thing you have to keep in mind is the 8600k is a lot less of a hotheaded chip due to absence of HT. It can run 1.3V on air - and probably a bit more - well within safe ranges.

 

[ZenZimZaliben]

The down playing comes from the fact that the performance gains of a delid are minimal, while you do void warranty, and it is a(nother) time, cost, effort, and risk factor. Its a payoff for 100-200mhz you must be willing to make, and for that meagre real-world gain, I'm not willing to - yet.

The pros of a delid are overstated in a way - yes you gain a great temperature drop. And a minimal - if any - additional clockspeed. I'm just calling things as they are.

Another thing you have to keep in mind is the 8600k is a lot less of a hotheaded chip due to absence of HT. It can run 1.3V on air - and probably a bit more - well within safe ranges.

I am not disputing the OC potential. It did take me from what I was benching, not 24/7 stable, and allowed me to overclock 200mhz beyond that point. From a 24/7 stability standpoint I gained nearly 300Mhz with about the same voltage levels. YMMV. IMO I don't think that was a worthless nearly free upgrade. 80% of the delid was about not seeing temps in that range even if they are acceptable.

 

[EarthDog]

Weird....

Doesnt like temps in the acceptable range and wants it lower...obliterates the warranty in the process. Seems like cutting off your nose to spite your face.

 

[ZenZimZaliben]

Weird....

Doesnt like temps in the acceptable range and wants it lower...obliterates the warranty in the process. Seems like cutting off your nose to spite your face.

Cool man if that's how you feel. It was worth it to me though and many others regardless of what it seems like to you.

 

[Melvis]

I would of said watercooling but after watching many many reviews over the past few months, air cooling seems to be doing a better job over most AIO coolers.

 

[bug]

Cool man if that's how you feel. It was worth it to me though and many others regardless of what it seems like to you.

It's working great for you, ok. Enjoy. Nobody said you shouldn't do it.

But here we are, in a thread where someone asks is water is better than air cooling and the first advice they get is: delid! And then you weigh in arguing in favour of delidding. Don't you think that's "a little" going overboard?
It's like a teen comes to ask if front first parking is better and we start telling them how they really should learn to skid into the parking place because that yields much better results.