How can this be? Air better than water?

[UltraTaco]

After a 15 minute revelation, I am confused. Triple fan radiator was worse than noctua phase change cooler!
Even dual fan cooler was better than triple fan cooler!



6:37 is where it gets interesting, the results. I'm nt saying the rest of the video isn't interesting. Linus is very interesting man to watch, but the results are most interesting!

 

[freeagent]

When I left my h100 with beast fans behind for an air cooler, I thought for sure it was just a matter of time before I got another aio.. but nope.. I haven’t. As for triple vs dual rad, once they get heat soaked then there is no more advantage. Aio is good for tidy looks imo, but that’s about it..

 

[Alaric]

I think AIOs are still bling, for the most part, not necessarily an upgrade over the best air coolers. With TDP going down almost universally I think that trend will become more so rather than less. And with chips being so close to their limit out of the gate now, custom water may be facing a long slide back to the realm of the hardcore more exclusively.

I can't say it's a bad thing, either. The days of furnace like FX chips and the mobo requirements, along with hundreds of dollars in custom loops, have limited nostalgia value. LOL

 

[Blaylock]

Fully agree with the above statement. In fact I'm debating even installing the custom for my 3700x because I can't really get it to OC over the Boost clock. Temps aren't what's holding it back, it's the silicon.

 

[sobe]

I'm not seeing phase change in that video though?? Phase change units act like refrigeration units and cool far better than air.

 

[UltraTaco]

Phase change, as in liquid inside heat pipes changing into gas and then back.

 

[bignazpwns]

I don't know man. I got a 3900x that can clock. 4.6 on all cores just beats on my loop.

I used a h115i and it was pretty effective. And even heatsoaked in a hot room it was ok. Not great but ok. But you need good fans and the stock fans just don't cut it.


Personally I would pick an aio over air. I'm just so use to setting up airflow I don't wanna deal with a huge heatsink. My VRM with no fan never even gets warm. Even hamming it with an oc. Also the weight of a big heatsink. And with an AIO you don't need to worry about will I be able to clear my ram? The heatsinks today are great. And they do a great job. But I'll always go with an AIO if I don't stick with a custom loop.

 

[trents]

The results in this video contradict every cooler comparison chart I have ever seen which generally show something like the Corsair H100i to beat the big Noctua twin tower air coolers by 2-4c in stress testing.

Plus, Linus fails to mention what is to me the biggest advantage of an AIO water loop and that is it makes working inside the case to install/remove/service other components much easier, not just "cleaner" from an aesthetic perspective.

Not only this, but many or most itx cases will take a 240 mm water AIO but have nowhere near the room for a high end Noctua.

 

[freeagent]

Yes indeed, nice and tidy. Everything is easy to get to. Massive heatsink is good for cooling, but my 8 pin cpu is buried under aluminum. And its true, silicon seems to be at the edge.. its not that the tdp is out of control, because on paper, its not so bad at all.

 

[Bustos]

Stock cooler all the way!

 

[Blaylock]

Phase change, as in liquid inside heat pipes changing into gas and then back.

Nearly every modern air cooler utilizes heat pipes. While inside the heat pipe it is true there is a liquid-gas-liquid exchange it is not what we consider a phase change. A true phase change involves a compressor, condenser, and evaporator to create sub-zero cooling. The very best you can ever achieve with a heat-pipe air cooler is near ambient.

 

[mackerel]

I have a very simple way to look at this: cooling is mostly about getting the heat from CPU to the air. The cooler itself is just something that spreads it out. Fundamentally, I think it comes down mostly to effective cooler surface area. A cheaper 240mm rad is probably lower in effective area than a high end dual tower air cooler, so to me it is no surprise some of them wont perform as well. The heatpipes or water just move the heat around, and as long as they're sufficient don't really help much.

At a minimum I think you'd need a decent 280mm or 360mm class rad to start getting incremental benefit over high end air. Currently I have a tiny ITX build with a 120mm AIO on it, as that is the practical maximum it can take. 240mm ITX systems are big. I also have a cheap 240mm AIO but hate the pump hum on it more than air noise. As it isn't very thick it isn't that effective either. Wouldn't say it had a useful application beyond 150W or so at best, which is no problem for a high end Noctua.

 

[JrClocker]

Interesting...I have been using a thick 49mm thick 120 mm AIO (Corsair H80i v2) with 2 Corsair ML120 maglev fans in push/pull on my 5820K for a long time now. It seemed to cool "as good" or "better" than the thin 240 or 280 mm AIOs I tried from Corsair (same fan series). It kind of makes sense that a thick 120 mm AIO might cool "as good" as a thin 240 mm radiator as the surface area of heat dissipating veins is about the same...as long as you have enough fan to push through it.

Also, you have to remember your thermodynamics and look at the thermal resistance between the CPU die and the CPU lid that your cooler comes in contact with. If this thermal resistance is high, a high end custom loop or an air cooler will make little difference. (Hence, the whole de-lidding craze to reduce the thermal resistance between the CPU die and the CPU lid.)

Said differently, not every CPU series is the "same" when it comes to the thermal resistance between the CPU die generating the heat and the CPU lid which is meant to dissipate the heat. In other words, some CPU series will allow for better heat dissipating than others. (It's not a mystery. You can look this up on the datasheet for the given chip.)

The other thing that annoys me is that almost none of the CPU cooling vendors publish the thermal resistance of their coolers to the "air" around them. This is the true measure of "goodness" for any type of cooling system. All other types of electronic cooling devices and heat syncs publish these numbers. Asetec (spelling?) published these numbers for their AIOs early on...not sure if they do anymore. This number is important as it is in series with the thermal resistance of the CPU you are trying to cool...and allows you to COMPUTE the maximum temperature (vs just guessing or experimenting).

I believe that this number if left out so that the marketing people can get you to look at:

- oooo….it's bigger, it must be better!
- oooo...look at the shiny fans, it must be better
- oooo...this was has 2 fans, that one has 4 fans, it must be better

The whole objective of a cooling system is to get the heat energy (measured in watts) away from the part generating the heat into a part that can dissipate the heat. Thermal resistance is measured in °C/Watt.

I always get annoyed when I read a cooler review when the reviewer says "we are going to use the 'hot' CPU"...when the "hot" CPU is only dissipating 90 W or so. If you want to test the cooler, don't use a "hot" CPU, use a CPU that generates a lot of heat...like the 2011-v3 socket who's CPUs can easily generate more than 150 W of heat. "Hot" CPUs and CPUs that generate a lot of heat are not the same thing:

- A "hot" CPU could be due to high thermal resistance between the heat generating CPU die and the CPU lid...or it could be that the CPU is generating a lot of power that needs to be dissipated.

Intel's "extreme" series which use solder as the heat transfer mechanism to the CPU lid versus thermal paste is a perfect example. On these chips, the cooler will make a difference as the thermal resistance between the CPU die and CPU lid is much lower.

Intel is making a "cost cutting move" on the lower TDP chips vs the higher TDP chips (i.e. it's more expensive for the solder process vs the thermal paste process.)

 

[Blaylock]

Here we go again.

 

[ehume]

The results in this video contradict every cooler comparison chart I have ever seen which generally show something like the Corsair H100i to beat the big Noctua twin tower air coolers by 2-4c in stress testing.

Plus, Linus fails to mention what is to me the biggest advantage of an AIO water loop and that is it makes working inside the case to install/remove/service other components much easier, not just "cleaner" from an aesthetic perspective.

Not only this, but many or most itx cases will take a 240 mm water AIO but have nowhere near the room for a high end Noctua.

As mackeral pointed out, cooling is just moving heat from the CPU to the air. AIOs do it by moving heat further from the CPU.

The biggest problems I saw with the video have to do with case airflow. In his case, the rads should have been drawing air in though the top. The ambient air is cooler outside the case than inside. But what about inside? The pressure will build because the outflow is limited to that exhaust fan trying to overcome the restrictive "grill" at the back.

A better design would be to remove that rear grill. Then you wouldn't need an exhaust fan: the inside air would make its way out of the case, silently. At that point, you could mount an AIO, and point its air inward. But at that point, who needs AIO?

Before I retired as a heatsink reviewer I tested plenty of heatsinks and a few AIOs. The AIOs had more surface area so they were able to shed more heat, cooling better. But I always conducted my cooling outside a cases in free air, to try to test these coolers independently. From the video you can see what a huge difference a case can make: in this situation it nullified the advatage of 240mm and 360mm AIOs over admittedly high-end heatsinks (air coolers).

So, since 2010, I have always advocated cutting out the rear "grill" and eschewing an exhaust fan. After experimenting with several promising methods, I found that covering the cut edges with Gorilla tape kept them from cutting hands. For aluminum cases as well, finger guards can be attached with fan screws and nylon licnse place nuts (#10, available at auto supply stores).

 

[EarthDog]

Well...

If you intake cool air from the top, you blow warmed air over the CPU socket and VRMs... not optimal. I'd rather use slightly warmed case air to cool the CPU and have good internal airflow than chase after a couple of C just for the CPU.

I also recommend an exhaust fan out back regardless. While Im sure the intake CFM works just pushing things out the back, an exhaust fan creates more airFLOW which is the most important thing. Just letting air be 'pushed' out of most cases sounds like it would build warm spots inside the chassis and would not be optimal compared to adding an exhaust fan and the additional airflow it creates regardless what direction you run the top.

In general, top/sides = intake, top/rear = exhaust. The more intake CFM or exhaust CFM is a debate without end. I prefer slightly more exhaust than intake. The only drawback is any little dust that comes through the cracks (it isn't much in my house). But again, this isn't really a pressure thing as a case isn't a sealed thing.. .its all about the CFM.

 

[Johan45]

To add, one of the biggest drawbacks I have found with AIO or WC is the lack of airflow around the power section. With air cooling, there's a lot more air circulation in this area just because of the nature of the air cooler. The AIO, on the other hand, doesn't have any fans in the CPU/VRM vicinity and can allow air to stagnate causing the board/power section to heat up, in turn, adding more heat to the CPU. This was very apparent when running an AMD FX 8350 for example and many users added small fans in this area to compensate.

 

[trents]

Here we go again.

Indeed.

 

[freeagent]

I find it amusing that when I was into aio and advocating them everyone was nah d14 this and d15 that and I was like ok whatever.. I used my aio until I wasn’t satisfied with its performance, which was about 4 years.. I left aio and went big air and now that those people are caught up with aio are enjoying them and don’t want to hear about big air.. my advice is just run it until you aren’t satisfied with its performance. You paid for it might as well use it.

 

[Blaylock]

The best part about big air coolers are their longevity. I have heat tube air coolers that are about 12-13 years old that still occasionally get used. As far as pure heatsink w/ fans, forget about it. So long as they are making fans they will function forever.