PDA

View Full Version : Liquid Cooling and Cooligy


MacRumors
Feb 26, 2004, 07:35 PM
CNet provides some followup (http://news.com.com/2100-7337_3-5166111.html?tag=nefd_top) on nanotube and liquid-cooling technology.

Specifically they mention a startup named Cooligy. According to Cooligy's vice president of marketing, they expect their liquid cooling system to be used in a "workstation" this year.

Rumor-readers should recognize this company's name. In October 2003 (http://www.macrumors.com/pages/2003/10/20031007074845.shtml), Cooligy admitted to creating prototype designs using their technology for a number of companies -- including Apple. In November more specific rumors (http://www.macrumors.com/pages/2003/11/20031117130322.shtml) emerged of a liquid-cooled PowerBook targeted for 2004.

Meanwhile, recent whispers (http://www.macrumors.com/pages/2004/02/20040223092546.shtml) have again claimed that Apple is working on liquid cooled machines for release in the near future.

arn
Feb 26, 2004, 07:38 PM
FYI, This could all be true, true, unrelated.

arn

Grimace
Feb 26, 2004, 07:39 PM
Hmmm.....I can't wait for G5 Powerbook comments to start churning....

javabear90
Feb 26, 2004, 07:39 PM
would this mean that they could pump water through the nano tubes?? or just use them at conducters.

I guess this means new PowerMacs and PowerBooks tommorow!

JohnGillilan
Feb 26, 2004, 07:44 PM
This could only mean one thing . . .

apple tablet/pda next tuesday

nmcphers
Feb 26, 2004, 07:47 PM
Apple won't have any "workstation" that requires cooling. It will most likely be the 3.4 GHz Prescott...duh! That's hot!

jnasato
Feb 26, 2004, 07:47 PM
I don't know the reliability of liquid cooling in the PC world, but it must suck for a tube to disconnect or some other malfunction...

What are the benefits of liquid cooling? Lack of fan noise is understandable, but with current fan/cooling setups, the processor heat, etc., don't seem to be a problem. Does this just allow for more reliable overclocking of processors?

GetSome681
Feb 26, 2004, 07:53 PM
Originally posted by jnasato
I don't know the reliability of liquid cooling in the PC world, but it must suck for a tube to disconnect or some other malfunction...

What are the benefits of liquid cooling? Lack of fan noise is understandable, but with current fan/cooling setups, the processor heat, etc., don't seem to be a problem. Does this just allow for more reliable overclocking of processors?

Lack of fan noise is not true. You use the water to suck the heat out of the cpu, yet you still need to get that heat out of the water or else it'll just constantly heat up till it approaches the temperature of the cpu, and no more cooling takes place. For this you need some type of radiator, which would need ot be cooled, and is usually by fans. It's almost exactly the same process as in your car.

Personally, I don't know if I'd want a water cooled powerbook. If something goes wrong, it's trashed. Sure this is fine during your year warranty, but what if you accidently mishandle your powerbook, and you accidentally break the water cooling system. You system could be perfectly fine after the accident, but as soon as the cooling system breaks, the powerbook is done for. There's the possiblity for a LOT of unhappy users after their year warranty if this system was not properly implemented with this in mind.

dho
Feb 26, 2004, 07:53 PM
COOL, I hope this aides in a g5 pb coming out soon.

Not that I will be able to get one, but cool anyway

Puppies
Feb 26, 2004, 07:56 PM
Yeah, the reliability issue bothers me too. And with Apple's bad luck in this area recently...

Personally I wish they'd just make the notebooks a little thicker, if need be. The 970 isn't that hot of a chip compared to a Pentium 4-M.

analogkid
Feb 26, 2004, 07:59 PM
I woudnt be surprised to see a 970fx Powerbook in the late summer in time for school, complete with liquid cooling.
Whether they use Cooligy or just make their own design

Chomolungma
Feb 26, 2004, 08:06 PM
Originally posted by Puppies
Yeah, the reliability issue bothers me too. And with Apple's bad luck in this area recently...

Personally I wish they'd just make the notebooks a little thicker, if need be. The 970 isn't that hot of a chip compared to a Pentium 4-M.

Can you provide any links to show a reliability issue with Cooligy technology? If not, lets give them the benefit of the doubt?

Chomo

agreenster
Feb 26, 2004, 08:11 PM
This 'workstation' mention is interestingly vague.

greenstork
Feb 26, 2004, 08:23 PM
Originally posted by agreenster
This 'workstation' mention is interestingly vague.

Considering Apple has never really labeled any of their computers workstations and many other manufacturers have, I have serious doubts that the "workstation" mentioned in this rumor is an Apple.

jnasato
Feb 26, 2004, 08:28 PM
Originally posted by GetSome681
Lack of fan noise is not true. You use the water to suck the heat out of the cpu, yet you still need to get that heat out of the water or else it'll just constantly heat up till it approaches the temperature of the cpu, and no more cooling takes place. For this you need some type of radiator, which would need ot be cooled, and is usually by fans.
I see, thanks. I thought some refrigerator type technology was being used (instead of fans).

StarManta
Feb 26, 2004, 08:35 PM
Pure speculation... but a quad 2.5-GHz G5 would generate a lot of heat... seems like it would be a good system to put water cooling in.... ;-)

It may cost $5000 but hell, I'd be savin'.....

AidenShaw
Feb 26, 2004, 08:38 PM
Originally posted by greenstork
Considering Apple has never really labeled any of their computers workstations...


Put the word "workstation" into the search box at http://www.apple.com, you'll see hits on "eMac Workstations", "iMac Workstations", "portable workstations" (PowerBooks), "G3 workstations", "G4 workstations"....

The only time that Apple avoided the term "workstation" was when they wanted to falsely claim the first 64-bit desktop.... Somehow they thought people would believe that all of the 64-bit systems already sitting on desktops weren't "desktops", but "workstations".

And if you look at Cooligy's website, they push their technology for "servers", "workstations", and SFF PCs.

morkintosh
Feb 26, 2004, 09:27 PM
Originally posted by GetSome681
but as soon as the cooling system breaks, the powerbook is done for.

thatís not necessarily the case, just because it is liquid cooled doesn't mean that the liquid is water; in fact you need a lot of space to get enough water to actually cool a chip ... so powerbooks are out. The configuration in a consumer device is likely to be such that a leak would cause little to no damage.

Puppies
Feb 26, 2004, 09:52 PM
Except that the system wouldn't run anymore...

It just seems strange to me that Apple would be investigating exotic cooling methods, when their systems don't generate that much heat compared to x86 systems.

Telomar
Feb 26, 2004, 09:55 PM
Originally posted by jnasato
I don't know the reliability of liquid cooling in the PC world, but it must suck for a tube to disconnect or some other malfunction...

What are the benefits of liquid cooling? Lack of fan noise is understandable, but with current fan/cooling setups, the processor heat, etc., don't seem to be a problem. Does this just allow for more reliable overclocking of processors? Without going into a lesson on heat exchanger design I'll just stop at saying having a liquid cooler on the CPU does provide some benefits that an air cooled system doesn't that allows for better flexibility of design related to better heat transfer coefficients. The fans won't be removed so noise will still be present. The biggest advantage is just you can cool the CPU far more effectively though (more heat removed per second). Kind of important if you want to use hotter CPUs.

As for reliability liquid cooling systems have been used in laptops for quite some time to distribute the heat around to areas where it is easier to cool. Most people never even realise it's there.

Telomar
Feb 26, 2004, 10:00 PM
Originally posted by jnasato
I see, thanks. I thought some refrigerator type technology was being used (instead of fans). You'll find most fridges have a giant pipe network following the compressor that acts as an air cooler. Only reason the fridge doesn't need fans is because of size.

spankalee
Feb 26, 2004, 10:06 PM
I think the amount of liquid (water or not) in these systems is quite small. The idea is that the liquid has a very high specific heat and can be moved to an area where it's easier to cool. You still need a radiator, but since you can now put the radiator near the vents or outside the case cooling the radiator becomes much easier.

I hope Apple uses this in the G5 tower so that they can cut down on the fans and heatsinks and offer a second optical drive bay and more HD bays. I would love to see them cut off 3" inches from the height too so I can rack mount one sideways.

This could also work well for the iMac G5.

Photorun
Feb 26, 2004, 10:53 PM
Is Cooligy owned by Coolio?

nitropowered
Feb 26, 2004, 10:54 PM
Yes, the "liquid" does not have to be water. It can be something that does not conduct electricity. There is such thing as submerged cooling where they submerge the whole motherboard and such into a substance to wick away the heat. Don't quote me on that but I think I read about it. Maybe someone else can help clairify/verify this.

However i wouldn't mind having a pb just a bit thicker for the G5. I really want one this summer as I will be entering college and need a pb.

Stolid
Feb 26, 2004, 11:49 PM
Liquid cooling is great because of the fact that liquids (especially water after cost considerations) absorb and give off heat at a great rate. This means its better than air at cooling 'per square X' (in, mm, whatever). Its also great because if you can get water over a spread out area than it takes less CFM from fans to cool
If I cool a 1 square cm area with a heatsink fan I can get X watts of heat off it per second. If I use water I can spread that water's heat effectively into a large radiator, meaning that my fans are effectively cooling the same watts but over 100 square cm; the limiter is in what takes the heat from that 10 cm source -- and water absorbs faster (and therefore better) than air.
As for reliability - I've got a water cooled PC and its lasted 4 years without any maintenence. None.
Watercooling can be just as reliable as air cooling; 'what if you get a leak' -- the machine is hosed. 'what if your fan fails' -- the machine is hosed. More than likely for a 'sealed' solution like a powerbook the tubes will be copper (copper is another great heat transfer material, most good heatsinks today are copper, silver is better but boat loads more expensive). which means to get a leak you'll have to puncture copper tubes. If you did something to do that; I'd say that a heatpipe or hs/f solution would more than likley have major problems in that scenario too.

scratchh
Feb 27, 2004, 12:29 AM
nitropowered:
you can indeed cool by submerging the motherboard in a non-electrically-conductive medium. after reading a bit ab out it, and friend and i created a fish-tank computer by filling an aquarium with mineral oil and fitting the motherboard into it (except the power supply!). it is a totally passively cooled system though, based on the large volume of oil to transmitt the heat to and out of. not practical for serious overclocking.

the ideal coolant for liquid-based systems would deffinetly not be water. it has a very high specific heat, so it can absorb a lot of heat energy, but it doesnt have the best heat transfer properties. plus there are the obvious problems associated with leaks.

the issue in liquid cooled system break-downs is not the liquid on the components (since the coolant likely isnt water), just the fact that the computer would no longer be cooled at all (and thus not useable) if the liquid leaked out or the pump broke, etc... it would take vey careful and clever engineering to make an efficent and reliable liquid cooling system for a laptop.

my $0.02
cheers,
scratch

legion
Feb 27, 2004, 01:04 AM
Originally posted by scratchh
it would take vey careful and clever engineering to make an efficent and reliable liquid cooling system for a laptop.

my $0.02
cheers,
scratch

...and it's been done and is currently used in IBM Thinkpads (gotta love their research budget!) since the A20, which was in '99. They use liquid cooling in a sealed hinge. (and they do use water)

See here:
http://www.research.ibm.com/thinkresearch/pages/2001/20010808_cooling.shtml

Opteron
Feb 27, 2004, 02:41 AM
You can effectivley cool a comuter system (not inludin moving parts, HDD, CD, etc...) bu submerging the entire system in mineral oil. No fans, convection powers the system.

While this is very unlikley, to have computers submerged in oil, such a fluid could be usedina closed loop. Jus incase thepipes breake loose.

Opteron
Feb 27, 2004, 02:56 AM
It's also interesting to see how opinions change over time. This is a link to an old tread, before the XserveG5 was announced:

"http://forums.macrumors.com/showthread.php?s=&threadid=42560&perpage=25&pagenumber=2"

It's worth a read.

LoadRunner
Feb 27, 2004, 02:57 AM
Apple happens to hold a patent, for a small mobile cooling system. I forget most of the details. But one thing did stick out, a very interesting detail, witch was the coolant. If the coolant happened to escape it would not damage the computer processor, mother board, or any other components.

With a close circuit it's quite easy to detect a loss in pressure so you wouldn't have to worry about frying your computer.

beelesbob
Feb 27, 2004, 03:03 AM
Liquid cooled powerbooks are almost certainly a nono - There just isn't any point in doing it - in order for Liquid cooling to be any use you need a big radiator, and that with the pump and condenser are going to weigh a lot... It just isn't a suitable technology for laptops... As for desktops, I'm highly sceptical on that front, I think it's something that apple (a) don't need to do now the G5s run cooler and (b) wouldn't want to do unless really necessary, because saying that "your £3000 computer has liquid flowing inside it" is NOT going to inspire confidence in it's reliability. (Whether it's non-conductive fluid or not, your computer's still gonna fry when it springs a leak)

Bob

Telomar
Feb 27, 2004, 04:14 AM
Originally posted by beelesbob
Liquid cooled powerbooks are almost certainly a nono - There just isn't any point in doing it - in order for Liquid cooling to be any use you need a big radiator, and that with the pump and condenser are going to weigh a lot... It just isn't a suitable technology for laptops...Well apart from the fact that IBM has already proved it a suitable technology you're also wrong on the technicalities of it. Liquid cooling, as proposed by Cooligy, offers the ability to get a lot more heat out of the CPU then move it to an area where you can cool the system more effectively, it provides much greater flexibility in cooling system design than is currently possible and therefore ultimately better design.

If you've ever opened a powerbook you'd see they already have heat pipes all through them. There's no need for a condenser and in fact optimally you don't want to be boiling that water. The pump is smaller than the size of a regular fan and the infrastructure for removing the heat from the fluid is already there in the form of the heat sink it will just get moved to a better location.
Originally posted by beelesbob
As for desktops, I'm highly sceptical on that front, I think it's something that apple (a) don't need to do now the G5s run cooler and (b) wouldn't want to do unless really necessary, because saying that "your £3000 computer has liquid flowing inside it" is NOT going to inspire confidence in it's reliability. (Whether it's non-conductive fluid or not, your computer's still gonna fry when it springs a leak) I've seen servers with liquid cooling systems, in fact Virginia Tech's supercomputer is liquid cooled, as are most. You face just as much risk should a fan die as should your liquid cooling fail. Both are reliable barring extreme misfortune and there's no reason to expect it will spring a leak.

pbrennen
Feb 27, 2004, 05:47 AM
could someone explain this to me... why is there such a concern with heat for g5s? i was under the impression that a g5 pulls less watts at an equivalent clock than the g4.

cripdyke
Feb 27, 2004, 06:24 AM
I'm amazed at the number of people making "statements" about this system - liability to break after a year, what the coolant is, etc - without bothering to read the article in question.

This is an etched silicon wafer with a water-based solution (not entirely water, there are other substances in the liquid but the exact make up is not revealed) flowing through the channels of the wafer 1mm away from the surface of the cpu.

It's a wafer on a wafer. If you broke the cooling system, in all likelihood you'd have cracked the cpu as well.

it doesn't use fans, b/c as the article says, the solution is pushed through the cooling chip by electric induction (electric current can induce motion in an ionic fluid). You wouldn't want the solution to be non-conducting, then it could not be efficiently moved by induction.

Now, of course, what this is is simply a better heat sink. In most cases (pun intended) you have a fan which blows air over the heat sink...in some designs a fan is not required within a particular system.

As for whether or not you would want the liquid to boil...in this system it is designed not to, but there are systems that have successfully used passive cooling and system inertia to recondense the cooling fluid and move it back over the cpu to be boiled again. It's not a bad idea (as some have phrased it), just difficult to execute properly. To overcome some of the problems, usually a substance with a low boiling point is used. The same concept is used to extract eletrical power from geothermal sites where the temperatures are insufficient to boil water.

Next time, perhaps we might try reading the article before making statements about how a tech will function and what would be a good or bad way to execute it!!

cripdyke
Feb 27, 2004, 06:30 AM
PS. What I can't wait for is a boiling-coolant system that turns a generator - turning the waste heat back into electricity to prolong battery life. They've already created laptop computers with B-C systems, all they need is a small version of a steam-turbine. It wouldn't take much room inside the case. Wonder what it would do for battery life??

lyonsden
Feb 27, 2004, 06:43 AM
These guys (http://www.octools.com/) have done it (not once (http://www.octools.com/index.cgi?caller=articles/submersion/submersion.html) but twice (http://www.octools.com/index.cgi?caller=articles/submersion2/submersion2.html)) using Fluorinert (http://cms.3m.com/cms/US/en/2-178/iilzkFV/view.jhtml) from 3M (http://www.3m.com/US/index.jhtml). The even have a video (http://www.octools.com/index.cgi?caller=articles/submersion2/subvideo.html) (WMV sorry :( ) of the second experience.

Telomar
Feb 27, 2004, 07:05 AM
Originally posted by cripdyke
it doesn't use fans There's one siting on the radiator. It still uses a standard air flow heat exchanger to cool the water.

Originally posted by cripdyke
As for whether or not you would want the liquid to boil...in this system it is designed not to, but there are systems that have successfully used passive cooling and system inertia to recondense the cooling fluid and move it back over the cpu to be boiled again. It's not a bad idea (as some have phrased it), just difficult to execute properly. The benefit in boiling a liquid is you can remove more heat. The disadvantage is it can kill your heat flux if you get past nucleate boiling. Since in this system you're really interested in getting as much heat out of that small area, the CPU, as possible you want a high heat flux. Ideally at peak operation it'd be nucleate boiling and other times convection but in reality I don't know of too many correlations that accurately predict it and that's a lot of control for the cooling system. There's better and easier things you could do than boil the liquid using Cooligy's design.

AidenShaw
Feb 27, 2004, 09:04 AM
Originally posted by legion
...and it's been done and is currently used in IBM Thinkpads (gotta love their research budget!) since the A20, which was in '99. They use liquid cooling in a sealed hinge. (and they do use water)

See here:
http://www.research.ibm.com/thinkresearch/pages/2001/20010808_cooling.shtml


Interesting, it sounds like they are using two heatpipes in series - one to carry the heat from the CPU to the bottom of the hinge, and a second in the top half to carry the heat away from the hinge into the top radiator.

Cooling and Cooligy were discussed in length recently at http://www.macrumors.com/pages/2004/02/20040223092546.shtml

A more conventional, and also effective, example of the use of liquid-filled heat pipes is in the Dell Latitude D600, a Pentium M (Centrino) system.

Here's a picture of the heat exchanger and fan:
http://docs.us.dell.com/docs/systems/latd600/sm/thermal4.jpg

The plate #1 sits on top of the CPU, beside the fan. Two liquid-filled heatpipes conduct the heat from the plate to the rectangular radiator seen above the plate.

The fan sucks air from the bottom and side of the case, and blows it out louvers in the back of the case.

The radiator drops into position between the fan and the back of the case, so that the fan is blowing the air across the fins of the radiator.

It seems most laptops have been using heatpipes for some time, so there's really nothing special about the Dell except that the manual has a nice picture that shows how the heatsink, heatpipes and radiator are arranged.

Also note that the "plumbing" of a heatpipe is a self-contained single tube. No connections to leak, just one solid assembly.

A good description of how a heatpipe transfers heat can be found at http://www.thermacore.com/hpt_how.htm

http://www.thermacore.com/images/heat-pipe-201.gif

Hemingray
Feb 27, 2004, 10:55 AM
I came into this thread not knowing a thing about liquid cooling, and there have been some really great posts to help me understand. Thanks!

This technology sounds great. As long as the liquid doesn't leak (obviously) and it's of sturdy construction, and dissipates more heat than fans, then I'm all for it.

Just one thing though: how will the liquid perform in the long-term? Like 5 years down the road, is there any chance that the liquid could evaporate or its state change? I'm assuming though, since this would be in a vacuum, that as long as the vacuum remains sealed that no evaporation would occur, but I'm certainly no expert on the subject. Anyone care to comment?

mrsebastian
Feb 27, 2004, 11:14 AM
the natural thought is of g5 pb of course, but this technology would be extremely usefull in so many applications. every six months at work, we take the time to open up our macs clean out the fans, dusting, and making sure everything is running good. with liquid cooling wouldn't it be possible to make an enclosed system without having to pump "dirty" air through it, thereby keeping the components in a more sterile environment?

morkintosh
Feb 27, 2004, 12:07 PM
Originally posted by nitropowered
Yes, the "liquid" does not have to be water.

-- SNIP --

as I will be entering college and need a pb.

umm... no it doesn't; the point of the liquid is to absorb heat and transfer it away from the processor. Any substance can do this all you need is:
1) a cool substance (as it absorbs more heat)
2) someway to cool the now hot substance

this process loops with a cool substance being warmed and transferring heat thus being cooled. Think of your heat sink now ... it is cool (has a high absorption rate) it gets hot (lowers the rate of absorption), the held heats is "blown away" by a fan making it cool (raising the absorption rate); so on an so fourth.

Water is effective because it easily transfers heat, but any other liquid could do this to varying degrees of success. Liquid nitrogen, for example, dissipates heat very well ... the problem there is the power requirements to keep such an inert gas in liquid form negate the gains from the cooling in the first place.


next fall take a thermo dynamics class, it will prove useful to you

a17inchFuture
Feb 27, 2004, 12:28 PM
I have asked this question several times on several different forums, and no one has answered me yet. Maybe you guys will help me out, please?

The question is, were pb g5's to be announced at the wwdc on june 28th, would they be available for purchase right then, or would one have to wait a period of time before online buying? I know I wouldnt receive it for a while, for sure, but could I buy it then?

Thanks very much guys.

sedarby
Feb 27, 2004, 12:56 PM
Originally posted by a17inchFuture
I have asked this question several times on several different forums, and no one has answered me yet. Maybe you guys will help me out, please?

The question is, were pb g5's to be announced at the wwdc on june 28th, would they be available for purchase right then, or would one have to wait a period of time before online buying? I know I wouldnt receive it for a while, for sure, but could I buy it then?

Thanks very much guys.

The reason no one has answered you is because no one here knows when the PB G5 will be announced or when it will be available. If we knew that then we wouldn't be here pining away for the date of release and speculating our tails off! :) :D

a17inchFuture
Feb 27, 2004, 01:12 PM
Read closely, that wasnt my question.

I wanted IF it WAS announced on teh 28th, is it typical that it would be available for purchase online (not necessarily to receive, as I know that it usually takes some time until they are released a la the mini) immediately, or would it take some time? I know the mini was buy-able almost immediately, so does anyone else know?

army_guy
Feb 27, 2004, 01:24 PM
Originally posted by StarManta
Pure speculation... but a quad 2.5-GHz G5 would generate a lot of heat... seems like it would be a good system to put water cooling in.... ;-)

It may cost $5000 but hell, I'd be savin'.....

A QUAD G5 is out of Apples leage, its not thier market and I dont think IBM would let it happen as the FULL power series takes care of that section not the cuttdown G5. $5000? try $10-15k the cost is not linear from a DP to a QUAD, the price difference is massive.

army_guy
Feb 27, 2004, 01:29 PM
Liquid cooling is very reliable assuming you follow some very simple rules, the main problem is leaks at the fittings whether they just break or comeoff you have to use good fittings and seals/threads etc.. The other problem is the pump, eheim pumps rules here, having dual pump is better if your really concerned. Also the combination of TEC and liquid cooling is very effective (compared to VAPOR cooling, the only difference is you can cool multiple heat sources no a single heat source with VAPOR aka VAPACHILL/MACH2) but has very steep learning curve and risk with noobs.

Hemingray
Feb 27, 2004, 01:29 PM
Originally posted by army_guy
A QUAD G5 is out of Apples leage, its not thier market and I dont think IBM would let it happen as the FULL power series takes care of that section not the cuttdown G5. $5000? try $10-15k the cost is not linear from a DP to a QUAD, the price difference is massive.

How do you figure? If the dual processor 1.8 costs $2,499, the dual 2.0 costs $2,999, you honestly expect the price to jump from $3K to $10-15K for a quad? No way. You could buy over THREE dual 2.0's for that price! They wouldn't sell hardly any. IF Apple did quads, I would be surprised if they priced it over $5,999.

army_guy
Feb 27, 2004, 01:32 PM
Much bigger motherboard, higher power components and much more expeience in PCB design (longer traces, very dense multiple layer PCB's etc.) Lets just say its no small feat.

Hemingray
Feb 27, 2004, 01:42 PM
I realize it's not as simple as slapping on a couple extra processors, but from a marketing standpoint they'd be killing themselves with a pricetag that high. Okay, maybe $6,999... ;)

army_guy
Feb 27, 2004, 01:42 PM
Originally posted by legion
...and it's been done and is currently used in IBM Thinkpads (gotta love their research budget!) since the A20, which was in '99. They use liquid cooling in a sealed hinge. (and they do use water)

See here:
http://www.research.ibm.com/thinkresearch/pages/2001/20010808_cooling.shtml


That isnt liquid cooling, the term liquid cooling implies the use of a pump, radiator and heat block. That IBM thinkpad thing is basicaly a sort of heatpipe.

army_guy
Feb 27, 2004, 01:48 PM
Originally posted by Hemingray
I realize it's not as simple as slapping on a couple extra processors, but from a marketing standpoint they'd be killing themselves with a pricetag that high. Okay, maybe $6,999... ;)

Its requires much more experience, more people, more R&D, more resources etc.. and also much more time to market, the overall cost comes from both the component costs and what I mentioned above. OK its only 2 extra CPUS and a few mosfets and capacitors, so what, now can you get it working?

Put it this way, IBM can do it for Apple but would it be worth it?
How many machines could they sell with $10k pricetages?

IMO it would be easier to use the power4 instead of the G5 for anything more than 2 cpus.

legion
Feb 27, 2004, 01:51 PM
Originally posted by a17inchFuture
Read closely, that wasnt my question.

I wanted IF it WAS announced on teh 28th, is it typical that it would be available for purchase online (not necessarily to receive, as I know that it usually takes some time until they are released a la the mini) immediately, or would it take some time? I know the mini was buy-able almost immediately, so does anyone else know?

Probably would not be available for sale for a month after announcement and then not delievered for another 2 months (so 3 months after announcement)... that is if Apple stays true to form.

(all speculation, of course)

legion
Feb 27, 2004, 01:53 PM
Originally posted by Hemingray
I realize it's not as simple as slapping on a couple extra processors, but from a marketing standpoint they'd be killing themselves with a pricetag that high. Okay, maybe $6,999... ;)

which is why quad processor systems are not for the consumer market(!)

army_guy
Feb 27, 2004, 01:56 PM
Originally posted by legion
which is why quad processor systems are not for the consumer market(!)

exactly!

Unless that consumer can spend $10k+ on a machine with out blinking an eye then its a no no for Apple.

legion
Feb 27, 2004, 01:56 PM
Originally posted by army_guy
That isnt liquid cooling, the term liquid cooling implies the use of a pump, radiator and heat block. That IBM thinkpad thing is basicaly a sort of heatpipe.

No, liquid cooling doesn't "imply" anything that isn't apparent by the terms "liquid" and "cooling"

In this case, the radiator is the exposed hinge and the pump is one created by the forces of thermodynamics.

Hemingray
Feb 27, 2004, 01:59 PM
Originally posted by army_guy
Its requires much more experience, more people, more R&D, more resources etc.. and also much more time to market, the overall cost comes from both the component costs and what I mentioned above. OK its only 2 extra CPUS and a few mosfets and capacitors, so what, now can you get it working?

It almost sounds like you think Apple is incapable of accomplishing this. I'm sure they've already invested in said people/R&D/resources for quads. The rumors have been surfacing for years. The point is, a $10-15K price tag is simply too high. Apple's more than paid for any R&D on quads they've done or are ever going to do. You're talking about a company with billions of dollars and zero debt.

Hemingray
Feb 27, 2004, 02:02 PM
Originally posted by legion
which is why quad processor systems are not for the consumer market(!)

...yet...

And dual processors are? Five years ago you probably would have said the same thing about duals. Where do you draw the line? A consumer could be anyone from a high school gamer to a person who edits digital video for a living. Then you start getting into "prosumer" and all that.

army_guy
Feb 27, 2004, 02:03 PM
Also I should add you people are assuming a QUAD CPU machine with an AGP slot? These are non existant from INTEL, IBM or SUN and from anyone else for that matter (the mother board just gets too big!!) . The only QUAD cpu MB with an AGP slot will be from TYAN for Opteron (I think there is also another OEM offering a full size MB ie QUAD Opteron/AGP PRO 110 /16 Memory Slots/DUAL SCSI), the application being 3D/EDA/MCAD tools.

army_guy
Feb 27, 2004, 02:05 PM
Originally posted by legion
No, liquid cooling doesn't "imply" anything that isn't apparent by the terms "liquid" and "cooling"

In this case, the radiator is the exposed hinge and the pump is one created by the forces of thermodynamics.

I suppose but this application would be unable to cool a DUAL CPU + AGP card +chipset + others etc...

army_guy
Feb 27, 2004, 02:09 PM
The QUAD G5 would target a non-existant market in Apple terms and also due to the cost of the machine. Iam baseing my costs on QUAD xeon/Itanium/Opteron(err with the price cuts, otherwise $14000 in CPUS alone for the 800 series 2.2GHz)

army_guy
Feb 27, 2004, 02:13 PM
Another problem is whether the QUAD G5 would be point to point?
The 800 series Opterons cost is high only on that term alone so that $10k is realistic. The only way to produce a so called cheap QUAD CPU machine would be to downgrade the expected specification i.e. QUAD 1.5GHz G5, 8-12 memory slots + shared BUS, few as possilbe HT links, sata drives (not SCSI and no RAID).

AidenShaw
Feb 27, 2004, 02:27 PM
Originally posted by army_guy
The QUAD G5 would target a non-existant market in Apple terms and also due to the cost of the machine.

People seem to assume that a Quad would be four times faster than a single - even though a Dual is almost never even close to twice as fast as a single on most applications.

This is why quads and higher show up in servers - a server isn't doing one application - it's often doing dozens to hundreds of things at once.

Another issue influencing the price is that a quad needs a much better memory system than a dual. Apple touts how the G5 uses a pair of 64-bit DIMMs to get "128-bit memory". Well - Intel quads usually have 4 to 10 DIMMs clocked in parallel for 256-bit to 512-bit wide memory!

So, a quad would be much higher priced, and much less likely to scale...

army_guy
Feb 27, 2004, 02:38 PM
Originally posted by AidenShaw

Another issue influencing the price is that a quad needs a much better memory system than a dual. Apple touts how the G5 uses a pair of 64-bit DIMMs to get "128-bit memory". Well - Intel quads usually have 4 to 10 DIMMs clocked in parallel for 256-bit to 512-bit wide memory!

So, a quad would be much higher priced, and much less likely to scale...

Intel way of getting that much memory in a 32-bit machine is outdated, slow and cumbersome. Its not 512-bit wide, the objective is to flick through multiple 4GB blocks (4GB blocks for 32-bit CPU) memory performance is pathetic using this kind of technique (it works but its not how I how call an engineers elegant solution). Another reason INTEL wants people to adopt the so called ITANIUM.

AidenShaw
Feb 27, 2004, 03:40 PM
Originally posted by army_guy
Intel way of getting that much memory in a 32-bit machine is outdated, slow and cumbersome. Its not 512-bit wide, the objective is to flick through multiple 4GB blocks (4GB blocks for 32-bit CPU) memory performance is pathetic using this kind of technique (it works but its not how I how call an engineers elegant solution). Another reason INTEL wants people to adopt the so called ITANIUM.


What *are* you talking about?

It has nothing to do with getting *more* memory in the system, it's for getting adequate bandwidth to/from the quad or octo memory controller.

So, you are saying that if Apple uses 2 DIMMs clocked in parallel to get more bandwidth it's "fast and good", but if IBM uses 4 DIMMs to get twice the bandwidth of Apple it's "outdated and slow".

There are no memory banks, it's a flat 36-bit (64 GiB) memory space - it is not segmented.

hulugu
Feb 27, 2004, 04:04 PM
Originally posted by GetSome681
Personally, I don't know if I'd want a water cooled powerbook. If something goes wrong, it's trashed. Sure this is fine during your year warranty, but what if you accidently mishandle your powerbook, and you accidentally break the water cooling system. You system could be perfectly fine after the accident, but as soon as the cooling system breaks, the powerbook is done for. There's the possiblity for a LOT of unhappy users after their year warranty if this system was not properly implemented with this in mind.

The TiBook already had a liquid cooling system, it's a sealed metal tube that runs from the processor and video card to the back of the machine. If you get a chance, open one up and you can easily see the tubes. Breaching the system would require an impact where leaking tubes would be the least of your problems.
I wouldn't mind an extended version of this system in a G5 Powerbook.

hulugu
Feb 27, 2004, 04:13 PM
Originally posted by army_guy
That isnt liquid cooling, the term liquid cooling implies the use of a pump, radiator and heat block. That IBM thinkpad thing is basicaly a sort of heatpipe.

So what would be the difference between the two? Doesn't the heatpipe tech in the TiBook and the Thinkpad use liquid to move heat away from the processor, etc. to the back of the laptop?

AidenShaw
Feb 27, 2004, 04:54 PM
Originally posted by hulugu
Doesn't the heatpipe tech in the TiBook and the Thinkpad use liquid to move heat away from the processor, etc. to the back of the laptop?

The heatpipes are actually phase-change devices.

The liquid in the sealed tube boils due to the heat of the CPU. The act of boiling (evaporating) absorbs much more heat for the temperature difference than if the liquid stayed in liquid form.

The hot vapor moves to the far end of the heatpipe, where it condenses back into liquid - giving up the heat that it absorbed when it boiled.

The liquid is then drawn back to the hot end by the capillary action of a wick (or by gravity is some stationary applications), repeating the cycle.

So, the heat pipe cools with a liquid, but vapor is also involved.

Some people like to reserve "liquid cooling" to mean a system where the coolant remains liquid at all times and is circulated by a pump or convection. The coolant in the Cooligy technology is always liquid, and uses a pump.

bryanc
Feb 27, 2004, 05:33 PM
Picture this:

a current hemispherical iMac, but with a transparent case and internal illumination, and it's G5 processor (and other heat generating guts) are cooled by a pair of immicible liquids that convect around inside the transparent dome like a lava-lamp!

Now I'd buy that just to watch it crunch SETI units!

Cheers

dudewheresmymac
Feb 27, 2004, 05:58 PM
you kno what would be cool....

if there was slushy machine that pumped the frozen goodness over teh processor....

i kno that would never happen but hey we can all dream

stingerman
Feb 27, 2004, 06:15 PM
Originally posted by army_guy
The QUAD G5 would target a non-existant market in Apple terms and also due to the cost of the machine. Iam baseing my costs on QUAD xeon/Itanium/Opteron(err with the price cuts, otherwise $14000 in CPUS alone for the 800 series 2.2GHz)

r u trolling? Of course there is a market for Apple. It's the same market SGI goes after as well as other high-end companies. Apple's customers can actually afford 4-way servers and workstations. When you are editing or compositing a LOTR type movie, do you think they care more about a $6K workstation or saving cutting their workflow in half?

Apple will go after the low-hanging fruit first and they have the customer base already for those segments that would buy such a PC or Server, so I really don't think it is off the wall. In fact, the 970FX has 3 coherent processor interconnects, allowing it to form an 8-way unified cache SMP computer. So the technology is already there. Apple has the engineering to develop advanced SMP systems. And the 970FX lends itself very well to such connections. OS X is highly threaded and would benefit intrinsically and Apple (as well as Adobe) have already designed their renderers to scale thread wise both up and out.

So for Apps like FCP, SHAKE, and maybe we'll even see Apple release TREMOR (now that they have the hardware for it) a quad PM/Xserve would be great and Apple has a built in market for it.

dadman
Feb 27, 2004, 08:48 PM
Originally posted by GetSome681
Lack of fan noise is not true. You use the water to suck the heat out of the cpu, yet you still need to get that heat out of the water ...

Not necessary because the water could carry the heat away to a radiator with a much larger surface area than the heat sink clinged on the CPU(s).

With a big enough heat sink, say, the metal chasis (or a modified version of it to improve the surface area), a fan is not necessay.

But then, CPU heat is always not the problem with Macs (or more precisly, Motorola/PowerPC chips), we still have the harddisks and power supply to take care of.

EscCtrlPigUp
Feb 28, 2004, 02:17 AM
Originally posted by nitropowered
Yes, the "liquid" does not have to be water. It can be something that does not conduct electricity. There is such thing as submerged cooling where they submerge the whole motherboard and such into a substance to wick away the heat. Don't quote me on that but I think I read about it. Maybe someone else can help clairify/verify this.
I read an article about 2 years ago: a group of people in New Zealand submerged a PC motherboard in a non-conductive liquid from 3M and using (I think) nitro filled cooling coils to keep it way cool. They managed to over-clock the Pentium 3 by nearly 800MHz before the system became unstable. Get this, the temperature was something like 30 degrees. Really interesting. Problem was the cost. They submerged the entire motherboard and all the cards in a Styrofoam ice bucket then filled it up. The 3M liquid cost them nearly $2,000 for about 2 gallons and I don't remember about the nitro. Another problem was the pump kept freezing (big surprise?) and the liquid slowly heated up. Cooligy has the right idea with their pump system and I can't wait to see Apple use it. The G5 or Xserve would be a great use for the technology. The problem for notebooks is the heat exchanger. In the current systems the giant heat-sink is spread out over half the machine and the fan turns on when needed. With liquid cooling the fan would need to be on all the time since it would heat up in seconds with such a small loop. Just a thought.

army_guy
Feb 28, 2004, 07:52 AM
Originally posted by AidenShaw
What *are* you talking about?

It has nothing to do with getting *more* memory in the system, it's for getting adequate bandwidth to/from the quad or octo memory controller.

So, you are saying that if Apple uses 2 DIMMs clocked in parallel to get more bandwidth it's "fast and good", but if IBM uses 4 DIMMs to get twice the bandwidth of Apple it's "outdated and slow".

There are no memory banks, it's a flat 36-bit (64 GiB) memory space - it is not segmented.


I disagree the xeon has a 400/533MHz FSB its bottlenecked anyway plus you have to share that between 4 CPUS hence only 133MHz each, not very good is it. All that bandwidth your on about isnt utilised. This doesnt apply to 64-bit machines as there address space isnt limited to 4GB.

army_guy
Feb 28, 2004, 07:58 AM
Water is conductive?

Be more specific?

You dont use tap water in a liquid cooled computer, at the very least you use distilled water but preferably de-ionised water hence non-conductive. However the water WILL become conductive over time due to it being in contact the copper/alluminium hence the coolant must be changed every 4-6 months, I do this via "hot swap the coolant" very easy to do assuming your using appropriate fittings.

AidenShaw
Feb 28, 2004, 10:15 AM
Originally posted by army_guy
I disagree the xeon has a 400/533MHz FSB its bottlenecked anyway plus you have to share that between 4 CPUS hence only 133MHz each, not very good is it.

All the more reason to have wider memory paths to the north bridge....

And, how does the architecture of Apple's quad PPC970 compare to the Xeon MP systems?

And what about DMA I/O transfers? Those use bandwidth across the north bridge, but never go across the FSB to the CPU. In a server, I/O bandwidth is often as important as CPU speed.


This doesnt apply to 64-bit machines as there address space isnt limited to 4GB.

Boy, you are confused here....

What does address space have to do with bus speeds, bus width and bandwidth?

Nothing at all.... You're just trying to create another 64-bit myth.

Oh, BTW, the PPC970 isn't even a 64-bit chip in this context.

It's a 42-bit chip, compared to the Pentium 4/Xeon at 36-bits. (The later Moto G4 chips are also 36-bit chips.)

The Pentium4/Xeon support 64 GiB of RAM, they are not limited to 4 GiB.

legion
Feb 28, 2004, 01:51 PM
Thanks Aiden... I was about to post the same reply.

army_guy,
Just because water is deionized does not make it non-conductive. Water by its very nature as a polar molecule and will carry electricity no matter what you do to it. I really have no idea where you're going with this but I'm assuming you've just been told these untruths by others. This seems to apply to your beliefs about Xeon MP systems and Intel memory addressing too.

shyataroo
Feb 28, 2004, 02:33 PM
Pure water does NOT conduct electricity Contrary to popular opinion... becuase there is nothing in pure water that electrons can travel through hydrogen and oxygen are both gasses so if Apple were to have a pure water cooled powerbook the water leakage would be no problem at all besides just a annoyance and a way of cleaning out any hidden dust ...


P.S. Holy Carp my sig sucks lol

army_guy
Feb 29, 2004, 02:10 PM
De-ionised water is not conductive enough (no ions) to pose a threat in a liquid cooled PC, ive tried and tested it on running hardware. If its 1-2+ years in a system then its conductive thats why you have to change it every few months. Also there are problems whens wixing alluminium and copper this can wreck havok with the water unless youre using some additive/anti-corrosion/anti-freeze mixture. Ask any liquid cooling person or specialist. I can refer you to Swiftech, DangerDen, Asustech, DTeck, Legris and Aquacomputer for info regarding anything you want to know.

army_guy
Feb 29, 2004, 02:19 PM
The Pentium4/Xeon support 64 GiB of RAM, they are not limited to 4 GiB.
I remember Carmack mentioning this (and critisising INTEL on the technique they use) and why 64-bit machines would bring more realistic games.

army_guy
Feb 29, 2004, 02:26 PM
The Pentium4/Xeon support 64 GiB of RAM, they are not limited to 4 GiB.

The xeon limit is 4GB the only way to use more is to use memory hacks/tricks or whatever they call it "Large Address Extension" which you would then need a very specific version of windows (W2k Data Center) or a specific linux distribution. What ever the trick is, its cumbersome and your still limited to 4GB threads/per CPU. The application also HAS to support this aswell, if it doesnt allthough the OS can address more than 4GB the application will not.

Opteron
Feb 29, 2004, 03:58 PM
Pure water does NOT conduct electricity Contrary to popular opinion... becuase there is nothing in pure water that electrons can travel through hydrogen and oxygen are both gasses so if Apple were to have a pure water cooled powerbook the water leakage would be no problem at all besides just a annoyance and a way of cleaning out any hidden dust ...


But due to the fact that water is such a good solvent, as soon as it touches and dust, or whatever it will become a couductant.

wdlove
Feb 29, 2004, 06:19 PM
I wonder if they might be thinking about adding Cooligy to the Power Mac also? It certainly would end the noise problem.

davetrow1997
Feb 29, 2004, 07:49 PM
FYI, This could all be true, true, unrelated.

arn

Arn, you must be a physician...
True, true, unrelated, indeed...
If you aren't, what other profession talks like that, I'm interested to know?

AidenShaw
Feb 29, 2004, 08:48 PM
The xeon limit is 4GB the only way to use more is to use memory hacks/tricks or whatever they call it "Large Address Extension" which you would then need a very specific version of windows (W2k Data Center) or a specific linux distribution. What ever the trick is, its cumbersome and your still limited to 4GB threads/per CPU.


Hmmm. Exactly like OS X, right? Only 4 GiB per process, but you can have more per system.

OS X can't give a process more than 4 GiB, even though the G5 can have up to 16 GiB.

32-bit Windows is exactly the same - the system can have up to 64 GiB, but each process is limited to 4 GiB.

You get so excited, army_boy, but OS X and 32-bit Windows have exactly the same capabilities on a system with more than 4 GiB. And who cares if only some of the server versions of Windows have the PAE support? Sometimes I've run datacenter server on my laptop if I needed the features in it. (Isn't there both a workstation and server version of OS X? Again, doesn't this make Windows and OS X more alike than different?)

Of course, if you are running on a 64-bit IA64 or AMD64 CPU, you can run 64-bit Windows and use as much memory per process as you have. But you don't have the 64-bit option with a G5 and OS X - only 32-bits regardless of the RAM in the system.

army_guy
Mar 1, 2004, 07:17 AM
But due to the fact that water is such a good solvent, as soon as it touches and dust, or whatever it will become a couductant.

Well dont contaminate it dude ;)

wdlove
Mar 1, 2004, 11:08 AM
Arn, you must be a physician...
True, true, unrelated, indeed...
If you aren't, what other profession talks like that, I'm interested to know?

Yes, arn is a physician. He just passed his boards in November. His speciality is a Nephrologist, treatment of the Kidney.

legion
Mar 1, 2004, 01:12 PM
Pure water does NOT conduct electricity Contrary to popular opinion... becuase there is nothing in pure water that electrons can travel through hydrogen and oxygen are both gasses so if Apple were to have a pure water cooled powerbook the water leakage would be no problem at all besides just a annoyance and a way of cleaning out any hidden dust ...



Wrong, wrong, wrong... A) Gasses can conduct electricity-- even noble gasses! B) Water is conductive. The Bohrs model is incorrect, if that's all you've been exposed to. Electron shell jumping occurs and water by it's very nature is polar. When you have water, you actually have OH- and H+ with a weak hydrogen bond between the two. When electricity passes through the water, the hydrogen is "completed" and (actually it's a multiple stage process and the OH- molecule breaks, H+attaches to H-, and due to the structure of O, you end up with O2 and H2; see the Heisenberg Uncertanity Principle for electron movement) However, the electron transfer is exactly what conduction is about, hence water conducts electricity. This "pure" water nonsense is just sily because both physicists and chemists/chemical engineers construct all equations using an assumption of "pure" water-- if it wasn't "pure" any volitaile molecules would have to be diagrammed.

army_guy
Mar 1, 2004, 01:52 PM
Ive been using liquid cooling for 4-5 years and not once have I destoryed H/W as a result of using de-ionised water, its not conductive period. Unless you fail to change the coolant for several months or years then it becomes conductive enough to destroy electrical components.

davetrow1997
Mar 2, 2004, 02:50 AM
Yes, arn is a physician. He just passed his boards in November. His speciality is a Nephrologist, treatment of the Kidney.

Thanks for the info... I am also a physician, that's why I picked up on the choice of words... I am in my internal medicine residency. Actually, there was a recent article in the NEJM (1 Jan 04) regarding thymoma and pure erythroid aplasia, titled True, True, and Related.

Funny...

davetrow1997
Mar 2, 2004, 02:54 AM
Thanks for the info... I am also a physician, that's why I picked up on the choice of words... I am in my internal medicine residency. Actually, there was a recent article in the NEJM (1 Jan 04) regarding thymoma and pure erythroid aplasia, titled True, True, and Related.

Funny...

Here's an attachment of the Computed Tomography scan demonstrating the anterior mediastinal mass. ;o)

legion
Mar 2, 2004, 04:54 AM
Ive been using liquid cooling for 4-5 years and not once have I destoryed H/W as a result of using de-ionised water, its not conductive period. Unless you fail to change the coolant for several months or years then it becomes conductive enough to destroy electrical components.

I guess I'll try one last time...

DI water in an ultrapure state is still conductive. It measures a conductivity of 0.1 microsiemen. Deionization is the same as reverse osmosis and is just a means to purify water. However, and this is a big however, the second DI water, even in an ultrapure state, meets the air (in essence outside of a vacuum) carbon dioxide dissolves into the water. The dissolved CO2 increases the already conductive H2O dramatically. The idea of having water (any water) touching exposed (non-insulated) electronics and not doing damage is ridiculous.

If you don't buy this, then I guess the only other means of proof is by experiment. How about filling a glass tank (glass as an insulator and to prevent grounding principles) with DI water and having someone stand in it while I put positive and negative leads into it... any volunteers (and I really hope no one jumps at this kind of opportunity)
:rolleyes:

windowsblowsass
Mar 2, 2004, 09:03 AM
holy sh*! this guy is using liquid nitrogen istead of fans to cool hix submerged motherboard
http://www.octools.com/index.cgi?caller=articles/submersion/submersion2.html

Telomar
Mar 2, 2004, 09:39 AM
I guess I'll try one last time...

DI water in an ultrapure state is still conductive. It measures a conductivity of 0.1 microsiemen. Deionization is the same as reverse osmosis and is just a means to purify water. However, and this is a big however, the second DI water, even in an ultrapure state, meets the air (in essence outside of a vacuum) carbon dioxide dissolves into the water. The dissolved CO2 increases the already conductive H2O dramatically. The idea of having water (any water) touching exposed (non-insulated) electronics and not doing damage is ridiculous.

If you don't buy this, then I guess the only other means of proof is by experiment. How about filling a glass tank (glass as an insulator and to prevent grounding principles) with DI water and having someone stand in it while I put positive and negative leads into it... any volunteers (and I really hope no one jumps at this kind of opportunity)
:rolleyes:Actually all polar molecules are self ionising and are therefore conductive. The extent of ionisation depends on the polarity of the molecule. Water is a weak electrolyte. It isn't the best conductor but it does conduct.

Furthermore, deionised water actually leaches out metal cations from alloys and is a bugger to contain. Even in things like PVC you get leaching issues. You pretty much have to treat the water in most cases and make the solution slightly basic.

legion
Mar 2, 2004, 11:00 AM
Actually all polar molecules are self ionising and are therefore conductive. The extent of ionisation depends on the polarity of the molecule. Water is a weak electrolyte. It isn't the best conductor but it does conduct.


I already mentioned this in an earlier post in reply to army_guy's claims :) , however since it didn't seem to impress upon him, I was going with a quantitative post on the conductivity measurements.

Telomar
Mar 2, 2004, 02:40 PM
I already mentioned this in an earlier post in reply to army_guy's claims :) , however since it didn't seem to impress upon him, I was going with a quantitative post on the conductivity measurements.Oops. It was 3 am and I was a bit too tired to read some of the earlier stuff :o

eric_n_dfw
Mar 2, 2004, 08:12 PM
http://www.techtv.com/screensavers/shownotes/story/0,24330,3626775,00.html
(no video on web site yet aparently)
Kenneth Goodson, Stanford professor and Cooligy co-founder, demonstrates a completely silent PC water-cooling pump.

Not only completely silent, but no moving parts either. As discussed earlier in this thread, they're using ionization to move coolant through the system.

They also show something he called a "Microchanel Collector", which is barely bigger than the silicon chip it is supposed to cool, taking the place of today's huge water blocks and/or heat sinks.

Pretty cool stuff - take a look at www.cooligy.com