Linux Adding Leakshield Driver Support For Reporting Liquid Cooling System Leaks

[skeevy420]

Air cooling like it nor that is brute forcing the problem.

That's funny because you started with:

Ducting with fans does work up to a point as long as you have the airflow space and you are not trying to melt the earth..

The reality is it depends on how you design your system, if you're trying to cram too much stuff in too small of a space, and what kinds of systems we're talking about; like how the most badass home gaming PC needs less cooling resources than a 4 CPU server board in a room with 40 of those servers. At some point we transition from utilizing air effectively to brute forcing air through a case.

Take my home gaming PC, when I bought the case it came with two fans, one in the front and one in the rear and that setup with a Wraith cooler on the CPU was just fine with my old RX 580. The 6700 XT I upgraded to blocked a significant amount of airflow from the stock front fan which lead to over heating and thermal shutdowns. That's when I upgraded and put in three more case fans - a 2nd front intake, a side intake, and a 2nd rear exhaust. The kicker is adding more fans made it so I could run my fans at less RPMs so my system is now quieter and cooler. I'd have added a bottom exhaust but I didn't buy a modular power supply and I have to tuck my excess, unused, cables somewhere...

I'm not sure if what I did would be considered brute forcing or not. I don't consider it brute forcing. I consider it utilizing air effectively. Yeah, I added more fans to move more air through the system, but at the same time I turned all my fan RPMs down to make it quieter since I'm fine with the system running upwards of 65-70C at load. 5 case fans at 30% capacity instead of 2 at 100% capacity. I could actually brute force more air through my system and knock the temps down another 10-15C. It's just loud and annoying so I don't. IMHO, that would be brute forcing air, not utilizing air effectively.

I'm using a Coolermaster N200 Micro-ATX. Any smaller of a case/form factor and I don't think low-noise effective air cooling would necessarily work as well or at all for me and my parts. I intentionally picked micro-ATX as the smallest I'd go for this exact reason since any smaller and you're forced into running your traditional fan setup at the extreme, you have to pick low-performance parts, or you have to spend a lot more on liquid cooling. That's what happens when you try to cram too much into a small space and violate the laws of thermal dynamics.

[oiaohm]

Take my home gaming PC, when I bought the case it came with two fans, one in the front and one in the rear and that setup with a Wraith cooler on the CPU was just fine with my old RX 580. The 6700 XT I upgraded to blocked a significant amount of airflow from the stock front fan which lead to over heating and thermal shutdowns.

The sad part is the 6700XT and RX680 heatsinks are over 3 times the size it need to be if most effective technology could be used..

You think you are not brute forcing this the problem is you are. Look at the airjet https://www.froresystems.com/ These has between 3 to 10 times better heat remove than our current fan based systems. Airjet increases the airspeed until you break the boundary layer that preventing heat from being abstracted.

Think about it Skeevy420 if you only need to move 1/3 to 1/10 the amount of air to get the same amount of cooling. The setup you described would be closer to 1/10 effectiveness vs theoretical best at heat removal.

Airjet gets inside 0.01 percent of the theoretical best.

"laws of thermal dynamics" the reality is current fan based air-based and water based cooling are not designed to make the most out of thermal dynamics.

https://en.wikipedia.org/wiki/Thermal_boundary_layer_thickness_and_shape

Thermal boundary this part of thermal dynamics that works against cooling with air and water. Increase flow velocity the lower the resistance to heat transfer. As you noted increasing airspeed with fans you run into growing noise problem. Airjet increase the velocity in compact unit to the ideal.

Heat pipes and other phase change parts are close to 98% effective at moving heat this is well better than the 1/3 to 1/10 of the fin stack connected to them. Immersion cooling phase change liquids are popular from effectiveness point of view.

Water cooling custom or AIO hits the Thermal boundary layer problem twice once on the radiator and once on the blocks connected to what ever you are attempting to cool.

This is the sad part we have been brute forcing PC cooling for a long time. skeevy420 you are not seeing that your cooling solution is 3 to 10 times larger than it in fact needs to be by theory and by what airjet products prove. Air cooling with the parts you have in theory could work with low noise in a ultra compact case but you would have to stop using fans and instead use something like the airjets that can generate high speed air that gets past the thermal boundary problem without generating massive amounts of noise.

Noise from a fan majority is turbulence not the movement of air in the intended direction/required direction for cooling.

airjet only 3 times better than the server air cooling the ones that run air at 4x the speed of what you PC fans can do and also generate like 90DB of noise. Yes gaming laptops generating 60-70 db of noise are in fact on the low noise side. This is why it 3-10 your PC is closer to 10 more effective to use something designed like the airjet..

Servers are going liquid cooling not because they cannot make fan base air cooling work with enough brute force its that once you get to that level of brute force it generating so much noise it comes impossible to put your workers in enough hearing protection to be safe.

The annoying part is we most likely have another 20 years of fan based cooling ahead of use to to the company behind airjet patents. Same problem we have had with AIO having pumps on the CPU block.

[mdedetrich]

I mean liquid cooling is superior to air cooling its just that if you have a standard/typical desktop CPU, its usually not necessary because you don't need to move that much thermal energy and hence a high end Noctua fan often does the job. However once you start getting top end consumer CPU sku/dual GPU territory and higher you end up seeing the benefits of the liquid cooling more and more. I mean we have Intel CPU's taking 300+ watts alone now, that generates a lot of heat.

[Teggs]

This is what they do in a server system, https://www.wiredzone.com/shop/produ...6-216-213-4809

I am aware of these solutions. Because an unknown Threadripper system was in play, and people do put Threadrippers in standard desktop towers (even one built for Greg Kroah-Hartman), I thought I'd pipe up. But it seems that zexelon is all over the possibilities.

[oiaohm]

I mean liquid cooling is superior to air cooling its just that if you have a standard/typical desktop CPU, its usually not necessary because you don't need to move that much thermal energy and hence a high end Noctua fan often does the job. However once you start getting top end consumer CPU sku/dual GPU territory and higher you end up seeing the benefits of the liquid cooling more and more.

This is a common mistake is that liquid cooling is superior to air cooling. Air cooling is not all the same things and liquid cooling is not a superior as many think.
https://www.icegiantcooling.com/ and airjet are two of the higher performing air cooling solutions.

Heatpipes are phase change cooling. What ice giant does is phase change cooling with massive improvement still using fans. High end Noctua heatsinks still depend on heat pipes with hot enough CPU you exceed the heat pipe transfer limit this is why items like the ice-giant exist.

Of course the lack of pump in the Ice-giant means you have less points of failure than AIO. The sealed copper of the icegiant you don't have the leak problem that AIO have.

.Linus tech tips put the icegiant heat to head with liquid cooling AIO the horrible point is that the icegiant out performed the AIOs including the largest ones on the market. So non custom solutions there is always some form of air cooler that kicks liquid cooling tail.

Yes there was attempted todo phase change cooling like the icegiant in a AIO form factor they all run into the same problem flexible pipes end up with too much gas leakage leading to lack of liquid inside the cooler(basically leaking in the other direction to a normal water based AIO ).

Custom liquid/water cooling is way simpler the design is very forgiving. Like you cannot go and do a custom loop based on icegiant tech. Yes there is immersion phase change solutions out there they perform better than your general liquid cooling they are also not suitable for home use.

Yes your highend consumer CPUs can be cooled by the highest end air cooler being the ice-giant better than your water based AIOs.

High-end servers end up using phase change immersion cooling. Liquid/water cooling as you mind range when noise with air cooling comes too much of a problem. Heat pipe and ice-giant like Air cooling in servers that need high uptime due to the high reliability.

The benefits of custom liquid cooling is that it is fairly simple to increase radiator size including put it outside case if required to get fresh cool air that does not make it superior to air cooling. Yes the benefit to increase radiator size with need does not exist with the AIOs or items like ice-giant.

[Vaporeon]

As already stated, Noctua NH-D15.

As for GPUs, the ATX add-on card specification has long since been inadequate (especially since they do not even use full length cards), but no GPU manufacturer wants to be the first one to jump.

[oiaohm]

As already stated, Noctua NH-D15.

Noctua NH-D15 it out performed by ice giant and the server rack-mounted equal. Yes there are server rackmount using the same kind of features as the icegiant using a phase change fluid using gravity return from radiator with solid metal pipes these are custom made for server rack-mount cases.

As for GPUs, the ATX add-on card specification has long since been inadequate (especially since they do not even use full length cards), but no GPU manufacturer wants to be the first one to jump.

No there is something wrong here.
https://en.wikipedia.org/wiki/PCI_Ex...ress_(standard).

https://www.voice-boards.com/dialogic-card-length/

PCI/PCIe Full-Length 111.15 × 312.00 × 20.32 is the size the card per slot is meant to be inside. 12.5 inch/ 312.00mm length is exceed by many modern day GPUs exceed the PCI Full-lengh max allowed size..
ISA card full length 13.5 inchs or 342.9mm. Yes they are still inside the ISA max size.

Something important full length ISA card is meant to have end support so it does not bend most modern GPU don't come with end support .

ATX by specification add-in card can be ISA or PCI/PCIe dimensions. PCI/PCIe no end support and should not need end support should be required if specifications obeyed with card constructed right and ISA for full length cards by specification should have end support to prevent bending.

Yes ISA standard card is meant to have a bracket on both ends.

Lots of GPU are too long for PCI/PCIe standard and missing bracket to be ISA standard so too short in length to ISA and too tall for both standards.

.It would be nice of GPU would be to either PCI/PCIe or ISA size standards not the current half way between they currently are..