PDA

View Full Version : Apple steps into Liquid Cooling


MacBytes
Jun 9, 2004, 02:18 PM
Category: Apple Hardware
Link: Apple steps into Liquid Cooling (http://www.macbytes.com/link.php?sid=20040609151808)
Posted on MacBytes.com (http://www.macbytes.com)

Approved by Mudbug

wdlove
Jun 9, 2004, 03:19 PM
I'm very interested in the 2.5 G5 with the liquid cooling. Want to get thoughts from others as to the efficiency of this new device.

MacsRgr8
Jun 9, 2004, 04:13 PM
I'm very interested in the 2.5 G5 with the liquid cooling. Want to get thoughts from others as to the efficiency of this new device.

Me too.

Would this feature pave the road for higher clock frequencies, or is it just more quiet?
Is it more fragile?

edesignuk
Jun 9, 2004, 04:20 PM
Anyone know how they're doing this exactly, with all water cooling for PC's you have to bleed the system every now and then to clean it out, add anti-freeze (or some equivalent) etc. I can't see you doing this on a G5 :confused:

2GMario
Jun 9, 2004, 04:38 PM
i'm not a engineer, but this is one thought on how to atleast extend the life of the fluid

the fluid is just running threw some sort of block (aluminum, stainless steel, etc... some sort of metal) and some sort of piping (plastic or metal again)

in a car, your oil gets dirty from debry in the air that the air filter doesnt catch, and any metal shavings that occure cause of the metal on metal friction in the engine

you CAN, were you to take off your oil pan, put a magnet in the oil pan, and as the oil circulates threw the engine, then settles back into the oil pan, the magnet would catch the metal shavings / giving the oil longer life

maybe, in the resevrior or what ever thier using to store the fluid, they employeed some thing similar to make a attempt to purify the liquid as it passed threw the system ?

just a thought
-Mario

shamino
Jun 9, 2004, 04:56 PM
Anyone know how they're doing this exactly, with all water cooling for PC's you have to bleed the system every now and then to clean it out, add anti-freeze (or some equivalent) etc. I can't see you doing this on a G5 :confused:
Not necessarily. The early designs all made you do this, and many people prefer to use them today, but there are other designs. For instance, this model (http://www.frostytech.com/articleview.cfm?articleID=1576) heat sink has a small, sealed, passive liquid-cooling system in it. This is, basically, a pair of copper pipes with distilled water sealed inside at low pressure. My home PC, a Shuttle XPC (http://www.shuttle.com/hq/product/barebone/default.asp?B_id=25) has a similar device as a part of its cooling solution.

Apple's is probably more elaborate than this, because they're claiming software control over it, but it's still perfectly possible to have a sealed liquid-cooling system, even with active components.

It's worth noting that air conditioners and refrigerators are also sealed systems. Unless they spring a leak, you never bleed/refill them.

shamino
Jun 9, 2004, 05:02 PM
in a car, your oil gets dirty from debry in the air that the air filter doesnt catch, and any metal shavings that occure cause of the metal on metal friction in the engine
This is part of it. Another part, however, is that the oil runs alongside some very hot parts of the engine. The temperatures cause it to breakdown over time. Synthetic oils are less prone to thermal breakdown, which is one reason why cars that use them can go longer between oil changes.
you CAN, were you to take off your oil pan, put a magnet in the oil pan, and as the oil circulates threw the engine, then settles back into the oil pan, the magnet would catch the metal shavings / giving the oil longer life
I believe this is done in the resivoir for automatic transmission fluid. I don't know about for oil. Of course, it only works if your engine parts are magnetic. If they're aluminum, this won't do a thing.
maybe, in the resevrior or what ever thier using to store the fluid, they employeed some thing similar to make a attempt to purify the liquid as it passed threw the system ?
The inside of a computer doesn't get nearly as hot as the inside of an internal combustion engine. It's unlikely that a computer's coolant will ever suffer thermal breakdown. If the piping is non-corrosive (plastic, teflon-lined, glass-lined, or something else that won't chemically react to the coolant) and the coolant is clean (distilled water or other purified fluid), then it may never need to be changed.

NP3
Jun 11, 2004, 08:32 PM
"Better yet, the flow of the fluid is continuously controlled through the same OS X monitoring software that controls the fan speeds."


I hope we never get to the point where virus writers could whack this control system....

ugh, that would be ugly.

windowsblowsass
Jun 11, 2004, 08:58 PM
any one know how the system works is it a traditional system or is it something revelutionary like cooligy

Danrose1977
Jun 12, 2004, 04:03 PM
Based on Apple's user friendly approach I say it has to be a sealed system... my question is, is this passive or active, is the liquid pumped around or does the heat of the liquid cause it to rise and fall as necessary. If it is an active system, the OC possibilities are posibly good... perhaps 3Ghz?.. If passive it's probably at its limit...

shamino
Jun 13, 2004, 10:06 AM
Based on Apple's user friendly approach I say it has to be a sealed system... my question is, is this passive or active, is the liquid pumped around or does the heat of the liquid cause it to rise and fall as necessary. If it is an active system, the OC possibilities are posibly good... perhaps 3Ghz?.. If passive it's probably at its limit...
I would assume it's active. I can't imagine how software could control a passive system.

wdlove
Jun 13, 2004, 01:55 PM
I appreciate reading all the good input. Now I'm starting to feel more confident about the Liquid Cooling system that Apple has developed.

thatwendigo
Jun 13, 2004, 11:21 PM
Why hasn't anyone here even tried to quote Apple on the subject?

The Apple G5 PowerMac Design page (http://www.apple.com/powermac/design.html) says:
Take it up a notch without losing your cool. The top-of-the-line Power Mac G5 with dual 2.5GHz processors squeezes outrageous performance into tight quarters. To cool down those steaming circuits, Apple designed a sophisticated liquid cooling system that takes off the heat without bumping up the noise. Mac OS X dynamically adjusts the flow of the fluid and the speed of the fans based on temperature.

Dynamic adjustment of flow rate means that it's an active system with some kind of induction - pump or otherwise - that changes the cooling profile. The question now is what that would mean for overclockers, and how hard it would be to get the system to respond to greater temperature. Depending on where the controls reside, it could be as simple as letting the machine decide for itself (if there's a self-adjusting program that kicks the system in above certain temperatures), or as complex as hacking a ROM on the board.

We just don't know yet.

windowsblowsass
Jun 14, 2004, 12:33 AM
Why hasn't anyone here even tried to quote Apple on the subject?

The Apple G5 PowerMac Design page (http://www.apple.com/powermac/design.html) says:
Take it up a notch without losing your cool. The top-of-the-line Power Mac G5 with dual 2.5GHz processors squeezes outrageous performance into tight quarters. To cool down those steaming circuits, Apple designed a sophisticated liquid cooling system that takes off the heat without bumping up the noise. Mac OS X dynamically adjusts the flow of the fluid and the speed of the fans based on temperature.

Dynamic adjustment of flow rate means that it's an active system with some kind of induction - pump or otherwise - that changes the cooling profile. The question now is what that would mean for overclockers, and how hard it would be to get the system to respond to greater temperature. Depending on where the controls reside, it could be as simple as letting the machine decide for itself (if there's a self-adjusting program that kicks the system in above certain temperatures), or as complex as hacking a ROM on the board.

We just don't know yet.

shortest thatwendigo post ever remember where you were people

thatwendigo
Jun 14, 2004, 01:33 AM
shortest thatwendigo post ever remember where you were people

No, this is. :D

shamino
Jun 14, 2004, 10:20 AM
Dynamic adjustment of flow rate means that it's an active system with some kind of induction - pump or otherwise - that changes the cooling profile. The question now is what that would mean for overclockers, and how hard it would be to get the system to respond to greater temperature. Depending on where the controls reside, it could be as simple as letting the machine decide for itself (if there's a self-adjusting program that kicks the system in above certain temperatures), or as complex as hacking a ROM on the board.
This is all a guess, but if this follows the same pattern as other similar features, there will be a monitor program in MacOS that checks temperatures and other relevant sensors.

The variable-speed fans on the G5 (don't know about the laptops) are all software controlled. If the software doesn't run (e.g. a Linux installation that doesn't have any fan-control software loaded), then the fans all run at full-speed (making a lot of noise) to keep the box from overheating.

I would expect the fluid system to be similar. The flow rate will probably be controlled by software in MacOS that monitors temperature probes and posibly other sensors. If the software doesn't run, then the liquid cooling will run at some default rate - probably one that will maximize cooling but not be efficient in other terms (power consumption, noise, maybe something else.)

In terms of overclocking, I would hope you won't have to think about this. Any well-designed software for cooling should be based on temperature, not indirect factors like CPU processing load. After all, if it's a hot day at home and your ambient temperature is 85 degrees, those fans are going to have to run at a higher speed than if you're in a machine room at a controlled 65 degrees.

Whether the system can keep up with the heat generated by an overclocked G5 is another question.

But I would suspect that it can. Although I haven't been able to get actual numbers from IBM's web site, I would expect that the 90nm 970FX chip at 2.5GHz puts out less heat than the 130nm 970 chip in the older 2GHz models. (We do know, from an IBM brochure, that a 2GHz 970FX consumes less than half the power of a 1.8GHz 970.) If the 2GHz models could be cooled by fans alone, there shouldn't be any problem with the fans alone cooling the 2.5.

My gut feeling tells me that this liquid cooling is in order to allow slower fans (and thus, less noise) and as a proof-of-concept for future models, not because it's needed for this model.

Anticipat3
Jun 14, 2004, 03:10 PM
My gut feeling tells me that this liquid cooling is in order to allow slower fans (and thus, less noise) and as a proof-of-concept for future models, not because it's needed for this model.

My feelings exactly. I'm surprised this hasn't gotten any more ink or excitement -- another first for Apple! First Liquid Cooled mass-market PC. Do I like it? Well... The verdict is still out on that.

Enthusiasts have been liquid cooling their PCs for years, but even with some of the best systems anyone's ever rigged up, they can leak, and they can develop air bubbles, etc. My concern with Apple going this route is simple - reliability and repair problems. Fans are simple, and pretty reliable -- pumps are much less so, and when that pump does go out... there will be problems. Now think about getting it fixed -- forget doing it yourself, and think about the expensive proprietary apple pump, and grimace when you find out that it's a $500 repair. Now imagine the Powermac is a few years old, on its third owner... just try to FIND a pump, even.

However, as computers get faster and faster and cooling requirements go up and up, it's becoming obvious that an LCS is a great way to go. When working, they allow processors to run much faster than an air cooling system. There are thousands of reports of 2.4C Pentium 4 Chips running at 3.4+ Ghz, and while that's a better than average gain, you can bet that we'd still be at 2.0 and not 2.5 on those powermacs without the LCS. If apple can develop a reliable, effective LCS before anyone else, they're going to be able to ship faster, more reliable machines in years to come... not a bad plan.

It's disappointing for those of us who were waitiing for a "refined" rev B powermac - in that it's really more like Rev A hardware. Proof of concept, and way cool, but the problems that could result from a bad LCS make me worry.