Hardware & Technical AIO water coolers.

[Deleted member 110222]

Hi. So, my game is running at higher settings. This is great, but it means that my CPU is being pushed further, which means it runs a bit hotter.

I've been wondering, how much a difference does a 2-fan AIO system make to running temps over your typical aftermarket air cooler? I'm currently using a Hyper 212 for my cooling, equipped with one 120mm fan.

 

[Chris Simon]

Which CPU do you have? Is it overclockable? What does it mean "a bit hotter"? 60C? 70?
On a stock-clock CPU, running liquid cooler is a total overkill. It will be really cool, yes, but also pricier (like in power consumption) and noisier (two fans and a pump. Pumps in particular, especially in cheaper AIOs tend to produce a very annoying high-pitched sound)

If I were you, I'd look around for some nice coolers. Noctua makes some really cool.. ehm... coolers. Practically silent (one 120mm Noctua fan running at 300RPM) and can keep my highly overclocked Ryzen 5 under 60C without trouble.

 

[Caramel Clown]

You'd be surprised to find out that creme de la creme of air coolers match, or even beat pretty much most AIO coolers out there In an appropriate case I'll go with Noctua NH-D15 with dual fans over ANY AIO any day of the week. That's what I've used in the past, and use till today. It's also more quiet under load. No pump whine. Not to mention how much it helps to remove heat from all the other components that produce heat. Motherboard elements, and even storage all warm up during use.
My opinion - if you go liquid do custom loop. In my mind AIOs are really only worth it if your case is a real choker, and has terrible airflow.

 

[Jon flint]

You'd be surprised to find out that creme de la creme of air coolers match, or even beat pretty much most AIO coolers out there In an appropriate case I'll go with Noctua NH-D15 with dual fans over ANY AIO any day of the week. That's what I've used in the past, and use till today. It's also more quiet under load. No pump whine. Not to mention how much it helps to remove heat from all the other components that produce heat. Motherboard elements, and even storage all warm up during use.
My opinion - if you go liquid do custom loop. In my mind AIOs are really only worth it if your case is a real choker, and has terrible airflow.

You really rate the d15 that highly? I tried one before going the AIO route, with the cooler in place I could only have a single stick of ram and had to remove the vrm heatsinks as it blocked most of the mobo, and that's in a fairly large case.

 

[Chris Simon]

You really rate the d15 that highly? I tried one before going the AIO route, with the cooler in place I could only have a single stick of ram and had to remove the vrm heatsinks as it blocked most of the mobo, and that's in a fairly large case.

Wrong motherboard, wrong model of the cooler. They are also making coolers with longer heatpipes that sit higher.
Mine is sitting comfortably above the RAM stick closest the the CPU.

 

[Deleted member 110222]

Just had an alternative idea that I think I should try first just because it's so much cheaper.

Stick another 120mm on the other side of the 212.

 

[Jon flint]

Wrong motherboard, wrong model of the cooler. They are also making coolers with longer heatpipes that sit higher.
Mine is sitting comfortably above the RAM stick closest the the CPU.

Trouble is, if it sat any higher I wouldn't have been able to put the side panel on, the one I tried sat about 5mm from the glass side panel.

This is why I prefer not to air cool, the requirement to either use special low profile ram like corsair lpx and/or specific mobo models really limits options when building a system. By contrast, an AIO is quick cheap and just works.

 

[Chris Simon]

Trouble is, if it sat any higher I wouldn't have been able to put the side panel on, the one I tried sat about 5mm from the glass side panel.

This is why I prefer not to air cool, the requirement to either use special low profile ram like corsair lpx and/or specific mobo models really limits options when building a system. By contrast, an AIO is quick cheap and just works.

Well, yeah. In that case it's like C.C. said - it really depends a lot on your case and other components.

 

[Caramel Clown]

That's why I've mentioned - in an appropriate case You don't need an enormous case for D15, but it definitely takes a bit of planning before putting system together. Luckily, Noctua's webside has an extensive guide on what hardware is compatible with each of their coolers.

 

[Morbad]

Higher-end AIO coolers will generally match the very best air coolers at similar noise levels, and out perform them at higher noise levels.

However, top of the line AIOs are reaching into budget custom loop territory and have more points of failure than an air cooler.

I use them for some builds and on my test benches, because of convenience, but any really serious build I make uses an air cooler, or a custom loop with redundant pumps.

 

[Shadowdancer]

A water cooler with the radiator inside the case does not make sense. Even with the radiator outside the case, fluid cooling without a chiller only makes limited sense. Spend your money elsewhere.

 

[Ashnak]

Which CPU (or rather, which TDP) are we talking about here? There are smaller air coolers (well, ok, basically anything is smaller than a D15...) that also work very well. I prefer downblowers (for improved mainboard cooling). On my mobo and in my case (Asus H97 M Plus in a Anidees compact tower), a Cryorig C1 works very quiet and doesn't interfere with the RAM (but I prefer to use "naked" RAM, without those fancy aluminium stickers).

 

[Dural]

I've been using closed loop coolers with the rad inside the case for a long time, the last 5 or 6 builds. I have the air directed out of the case, I know several people here prefer positive pressure setups but I prefer negative. Venting the hot air straight out of the case is the reason why I started using closed loop coolers in the first place having done various experiments with air vents attached to fans and found that the case didn't fill up with hot air. You can obviously vent hot air from the case with fans but I prefer to have 3 or 4 quiet low speed input fans with filters and and a single rear exhaust fan with the cooler at the top doing most of the work drawing hot air out of the case. That all seems to work nicely and is only my take on it and there are of course any number of ways of cooling a PC effectively. I even milled out an aluminium block with water channels, fitted a 150w peltier and large heatsink with a fan and ran that in line with a closed loop system, it was fairly effective for what is was (dropped the CPU temps by 10 degrees or so) but completely unnecessary for what I was doing, still, experiments are fun which was the whole point of doing it.

The most vulnerable component on a watercooling system (leaks aside) is the pump, if that goes your CPU temp will rocket and the machine will hopefully shut down or take some other kind of protective action. The only brand I've used without issues is Corsair and am on cooler setup number 3 with them (they even have software controlled RGB lighting now ). Thermaltake, Arctic and Coolermaster have suffered component failures or other problems but that's just my personal experience.

There are some really good air coolers out there, you can get very good cooling performance provided you have adequate airflow through the case.

 

[The Chairmaker]

Interesting thing re liquid coolers is that due to the thermo dynamic properties of the liquids, they take longer to warm up, however when fully warmed up the radiators may not then be as good at dissipating heat, as with an air cooled system.

They also conversely take longer to cool down.

 

[Deleted member 110222]

Noted guys. I'll try a second fan on my heat sink instead.

 

[Chris Simon]

Noted guys. I'll try a second fan on my heat sink instead.

Or clean it, remove it and re-paste it, buy a better fan with higher static pressure (blades closer together), etc.
Unless you overclocked it, the CPU shouldn't just get suddenly hotter under normal circumstances.

 

[Morbad]

A water cooler with the radiator inside the case does not make sense.

It can make sense either as an intake or an exhaust.

In the former case, cool outside air passes through the radiator, and the air inside the case is a few degrees warmer, which is usually irrelevant if the components most limited by heat are cooled by the loop.

In the latter case, you avoid heating the case air further, but have slightly warmer ambient air going into the radiator...but this is no different than most air cooled setups, and again will usually only be a few degrees different in water/component temperature.

Even with the radiator outside the case, fluid cooling without a chiller only makes limited sense.

Beyond a certain point, a water block can remove vastly more heat from a processor than any heatpipe or vapor chamber that will fit in the same area, and surface area of a loop's radiator, even an internally mounted one, can be far higher than any practical air cooler's fin stack.

In practice, eight 6mm, or six 8mm, heatpipes will usually reach saturation at about 250-300w of heat, and the largest of dual tower heatsinks have similar fin area to a thin 240-280mm radiator. A good block, backed by a pump with sufficient flow rate, can move a thousand watts of heat off a part, and a good mid-tower case can often mount three 360mm radiators (or the equivalent) internally...enough to remove several hundred watts of heat from the coolant with a low deltaT.

I currently have a NH-D15S with two 2500rpm fans. The 360mm, internally mounted, radiator I used to have on it could move at least 100w more, with six 1000rpm fans, at the same temperature delta, allowing me 200-300MHz higher stable clock speeds on my HW-E CPU with lower load noise levels.

Interesting thing re liquid coolers is that due to the thermo dynamic properties of the liquids, they take longer to warm up, however when fully warmed up the radiators may not then be as good at dissipating heat, as with an air cooled system.

They also conversely take longer to cool down.

If this becomes relevant in actual use, something is wrong with the coolant loop; insufficient flow rate or an undersized radiator.

 

[Deleted member 110222]

Or clean it, remove it and re-paste it, buy a better fan with higher static pressure (blades closer together), etc.
Unless you overclocked it, the CPU shouldn't just get suddenly hotter under normal circumstances.

Will do.

 

[Shadowdancer]

It can make sense either as an intake or an exhaust.

In the former case, cool outside air passes through the radiator, and the air inside the case is a few degrees warmer, which is usually irrelevant if the components most limited by heat are cooled by the loop.

In the latter case, you avoid heating the case air further, but have slightly warmer ambient air going into the radiator...but this is no different than most air cooled setups, and again will usually only be a few degrees different in water/component temperature.

In the former case you're feeding all thermal loss of the cooled components back into into the case, effectively killing the components that need airflow like power supplies, chipset, or M.2 SSDs. In the latter case, you feed the loss of those components (for which you need intake fans anyway) into the CPU cooler making it less effective.

To make matters worse, you do not have the advantage of a phase change medium on the cooled components, which increases thermal resistance quite a bit. As was mentioned above, you get an advantage of thermal capacity, but as that gets used up, you lose efficiency.

The purpose of fluid cooling is to get heat away from the cooled components to dissipate it outside the enclosure where it doesn't matter and where you have a ready supply of cold air, and you want to do so with a fluid input that's as cold as you can get without producing condensation or freezing/gelling your medium. Putting it inside a computer case without a chiller is, to a large degree, masturbation: fun but not all that productive.

 

[Morbad]

In the former case you're feeding all thermal loss of the cooled components back into into the case, effectively killing the components that need airflow like power supplies, chipset, or M.2 SSDs.

The air coming out a properly sized radiator with sufficient air flow is maybe ~10C warmer than the air that went in.

Most power supplies on most modern cases intake air from outside the case entirely.

Most modern chipsets are little more than I/O hubs that produce half a dozen watts or so of heat at full load and have several dozen degrees of thermal headroom. Likewise, the VRMs on most quality boards for mainstream platforms (LGA-1151 or AM4) are complete overkill and will run maybe 40C over ambient during stress testing with heavily OCed processors...which is still ~40C lower than they are rated to handle if the inside of the case is 40C (which is unlikely).

Most M.2 SSDs are very easily cooled by sinking them to the motherboard with a thermal pad and even if that board is a few C higher than it was, the hottest running M.2 SSDs will never thermally throttle if given a thermal path to the motherboard. I typically see a 30-40C drop in M.2 VRM and controller temps during drive benchmarks just by stuffing a cheap thermal pad under them.

Ultimately, only the CPU and GPU are going to be anywhere near their temperature limits on most performance setups, and cooling should be focused on these components. I am always limited here (even when they are under water and all the radiators are dumping heat into the case) and always have enormous headroom everywhere else.

To make matters worse, you do not have the advantage of a phase change medium on the cooled components, which increases thermal resistance quite a bit.

Heatpipes are limited by the tiny volume of their working fluid. A good water loop moves multiple orders of magnitude more coolant, completely dwarfing the fact that boiling the working fluid in a heatpipe carries away more energy per unit of volume.

Final thermal resistance (the difference between room temp and component temp) is invariably going to be lower on a good water loop than on any air cooler after a certain thermal load is reached.

Putting it inside a computer case without a chiller is, to a large degree, masturbation: fun but not all that productive.

This is flatly incorrect as countless tests can demonstrate.

Any air cooled setup you can build, I can cool with internally mounted radiators substantially better, at the same or lesser noise level, provided I have room to mount sufficient radiator area.