Tiny Water Cooled System

Xev writes "Most people water cool a full PC, over at Hexus they have water cooled a MINI PC (SFF - small form factor). Creating the smallest water cooled system." Takes all kinds to make the world go round. I'm amused that the radiator is almost as large as the computer itself.

Comments

all this cooling

[jacquesm]

It would be *way* better to have a system that does
not use such enormous amounts of power in the
first place, the computations take up almost
no power at all, the rest is heat !

Re:all this cooling

[CTho9305]

I do not believe you are correct. In a modern processor, almost NO power is consumed when computations are not happening. CMOS logic is set up so that as long as its state does not change, there is no current flowing (well, transistors "leak", but that is relatively little current). However, when the clock ticks, all the gates start changing states. During the state change, power is consumed. This page has some excellent applets which show how and when the current flows.

During computation is the only time power is being used. Because of how CMOS works, power usage grows approximatekly linearly with clock speed - which makes sense - if power is consumed when stuff changes state, doubling the clock doubles the number of state changes per time, increasing power consumption.

filtration

[buzban]

i've never done anything this exotic, but whenever i add a larger fan to a PC, it sucks in a ton more dirt. anyone done anything like this with filtration? and if so, how much air flow did that block for you?

just curious...

Watercoolig -- yesterday and today

[(H)elix1]

Why watercool?

I started off with water cooling because I wanted to overclock... to get the CPU to run with a front side bus way out of spec you usually end up pumping more voltage to the processor. More voltage, more heat, less stable. Back when it was a buck or more a megahertz, you strapped a peltier plate on to really drop the CPU temp. The peltier plate alone usually kicked out more heat than the latest CPU's, so creative cooling - high speed fans, ducting, and eventually water cooling were required.

Fast forward to today. Mhz really does not matter. I can run 3-4 CPU releases behind and still have a screaming system. Stability is more important than an extra 200mhz, but the current generation of CPU's kick off the same kind of heat I had to deal with in an OC'd 566 (952mhz w/112 FSB water cooled + peltier, 833mhz w/98 FSB aircooled). Less heat still equals stability, however.... It only takes a couple 8K RPM fans in your office before pushing more air and buying louder speakers is not a solution.

A water cooling rig can be silent. I'm not sure what the point of a 8K fan on a tiny radiator, but my heat exchanger I got by with a couple low RPM ducted 120mm fans. Kits are becoming mainstream however - IE, you don't have to buy a couple cases of beer for someone with access to a machine shop to make your CPU heatsink.

Going the wrong direction

[shagoth]

It's about time the movement for hardware hackers shifted toward low-power, quiet computing. Having a giant radiator and a blower does niether.

There is no reason that home servers can't be PowerPC machines running with laptop harddrives other than the hardware hackers haven't yet found clever ways to come up with iMac motherboards. I recently changed out the drive in my home server for an IBM Travelstar 40 GNX and for all intents and purposes removed the last noise maker from my office.

A couple of months ago the server was reborn on a retired Apple Powerbook and the difference in the temperature from the traditional machine was profound. Since I live in the desert, I have to dump all the heat I make whether it comes off heat pipes, heat sinks or radiators so reducing waste heat is a good thing. The surplus hardware is out there for the scrounge (just like with Wintel stuff) and the power consumption and heat production is significantly less.

Similar answers are coming for commercial rack mount machines of which Apple's X-Serve is only the first. Remember that power costs money too, and not just fo r the machine but for the A/C to take the heat away.

Eyecandy

[Shinsei]

It looks very nice, but I don't see the practical uses of it. First of all, if I wanted to watercool my system, I would try a Peltier element to reduce the temperatures to as low as possible. I do of course have a low cpu speed myself, but 34 degrees celcius is my system temperature for an Athlon T-Bird 1.33Ghz system which is only aircooled - and the AMD CPU's are infamous for their heat generation. Also in this machine is a GeForce 4 TI4600 display adapter, which generates a whole lot of heat just by itself as well.

In my eyes, the system is only for eyecandy - as it is surely a beautiful sight for the eye - but it doesn't have much practical use. It is expensive, and I don't believe the system gains any much stability from it, as the Intel CPU's are already made for stability @ factory. If they were able to add a Peltier, perhaps display adapter cooling, and hd cooling - in that small case - then it would be a convenient solution. But as it is now - just eyecandy, imho of course. :)

Large cooler.

[Shanep]

I'm amused that the radiator is almost as large as the computer itself.

I used to be a network admin for a stock exchange disaster recovery site in the mid 90's.

We had BIG VAXen, as in *large*-double-door-fridge size.

A big chunk of the size of these boxes, was actually the cooling system itself. : )

hdd cooling?

[fw3]

Watercooling the cpu and eliminating the fan results in very little cooling for all the other parts of this cute little box. The HDD espeically will suffer reduced life expectancy.

Seems to me a false economy.

Only if need to be on the "Edge"

[Gerry Gleason]

The fastest processors of any generation will generate heat, becuase the faster you want to switch, the more power it will take. Theoretically, there is no minimum amount of energy to transmit or transform a bit or bits of information, so as your technology gets smaller and faster (from smaller geometry supported by each new generation of technology), the power required for a given speed goes down. So you can have the same speed at a lower power, or faster at the same or maybe even more power.

That said, there is a difference related to efficiency of the architecture. Some people think the RISC/CISC debate was ended when CISC (read x86 family processors) started using all the same internal architectural features as the top RISC processors, but they did it by pushing the technology, not getting rid of the complexity. It has been costing Intel more to produce the same performance as competing RISCs for a long time, but their market is so huge that they make up the difference in up-front engineering cost by having a bigger market. This also shows up in power consumption.

I find the Transmeta innovation interesting because they solve the instruction set complexity problem a different way. Translate on the fly to an efficient micro-architecture. Can't be quite as good as a simple ISA, but it does much better on the Power/Computation measure.

Bottom line is the architecture matters.

Does this actually increase reliability?

[panurge]

I'm actually curious to know whether having those humungous cooling plates on the top of CPUs and bridge devices might not actually increase unreliability due to increased pressures on the IC carrier, socket and mobo.

With all of these water coolers, the surface temp. can never go below ambient air, so getting a fast but quiet flow of air to well designed fins on the hotspots would seem a better option. Water cooling for vehicles works mainly because of the problems (with air cooling) of cooling places difficult of access, like around the valve guides on an IC engine, plus the desirability of maintaining a controlled temperature WELL ABOVE any sensible ambient. The objective here is to get the temp. down close to around 10-30C.

Using a Peltier device can get the die temp to below ambient but requires heat to operate which also has to be removed from the case- I have come across a case where people didn't understand this, cooled one device, an IR sensor,with the Peltier and then had the system fail because of overall thermal overload caused by the additional 120W needed to drive the Peltier stack. The answer would seem to me to be the one favored by Apple - well designed air cooling. As Intel and AMD cpus are the SUVs of the processor world, add heat pipes or fluorinert bags to transfer heat efficiently within the case to where the airstream can run most effectively.

I'm starting to think - why are we doing this?

[gelfling]

Pretty soon we're going to need reheat units attached to the coolers to break up the ice that collects on them then we'll need moisture collectors to manage the condensors and another power supply to run the whole thing which will generate more heat. and so on.

There has got to be a better way - make CPU's that run colder.

Re:What about condensation?

[SmallFurryCreature]

Condensation happens when the temperatures differ. So when warm air moist air touches a cold surface. The water in this setup is WARMER then the air around it. Why? Cause airflow itself, or at least the relativly low airflow created by small fans like this can not cool below ambient temperature. And since the water itself will be warmed up by the cpu it will typically be a few degrees higher then the air, so no condensation.

There are setups where they cool the water and then of course condensation is a real problem. Same if you use peltiers.

The only way you could with watercooling create temps lower then ambient is if you use evaperation. This is not the case here and I am not sure that anyone has ever tried it. Try it by wetting you hand and holding it in front of a fan, this will be a lot cooler then youre dry hand.

If you don't believe me on the condensation put something like a mirror in front of a fan and put it at max. No way condensation should form. (presuming reasonable normal conditions there are always exceptions).

Please note that you can achieve lower then ambient with just airflow without evaporation, I believe it is called vortex cooling or something. But this requires the kinda flow you get out of an airpump.