The future of consumer cooling

[derraa]

I was thinking, as processor improve, they generate more heat. They are not getting bigger, so what do you think the future of cooling holds. Water cooling is too space consuming and not very simple and economical. Do you think that freon will be the future? I was thinking, that you could build an eficient phase change in a small area. A one peice system could be made. a compressor to compress the freon, it could then run into a waterblock with fins on the top to cool the compressed freon. It could then run into a bottom chamber.(heat insulatioon between them) where it would decompress. Do you guys think this is doable? It would mound ussing four bolt system. I could make the blocks, but where could I get a small compressor? Thanks
Derraa

[WebMasta33]

freon = bad. Bad bad polutant gas.

Plus, compressors are bulky, you still the need the expansion/contraction coils to release the heat.

You want a environmentally friendly solution?

Click -> http://www.sciencenews.org/20020105/fob2.asp

Just recently been toned down for practical home use. Expect to see magnetic cooling in refrigirators in about 5 years or so.

[resago]

smaller = less power consumption = cooler

also, when chip makers start using semi-organic transistors, power consumption and heat will be reduced at least 60%.

[Brians256]

You want a good short-term replacement? Use plain propane! It has all the efficiency of normal CFC refrigerants. It is as efficient as R12 and is far better than R134a (the environmentally "safe" option). It is CHEAP, because we don't have any complex synthesis steps to make it. It's practically a waste gas from oil drilling.

Why don't we use it? Because it's dangerous in large quantities. It's one thing to have a small bit of propane in your refrigerator. In fact, it's far less dangerous than the gas stove in your kitchen because there is simply not that much gas in a refrigerator. But, to make thousands of refrigerators requires large storage tanks that big companies don't want to make safe (because it costs money up front). So, we don't do it.

Incidentally, society would make up the investment money because of the the cheap nature of propane and the increased energy efficiency, but no one wants to pay an extry $10 per refrigerator to save hundreds of dollars over the product lifetime. Just look at the American consumer's decision on front-loading washers. A little up front cost and you make it back (and then some) in reduced utility bills.

Did I mention that you don't have to use special recovery techniques to capture "used" propane? You don't have to use special lubrication oil for the compressor?

Instead, we spend bunches of dollars and energy on synthesizing artificial R134a gas that won't burn. Well, at least the chemical plants are making money off of the decision. Oh, and the companies that own the electrical plants.

[Joe]

I think the future is not with water, but with Amonia or similar coolants. For a roomtemp phase change setup. Sorta like Heatpipes but much more versitle.

[Brad]

the topic was consumer cooling, not geek ways to cool

I think we'll see bigger heatsinks, still mostly Al. With 80mm or 92mm fans at the back of the case sucking air out, through a duct from the cpu

[WebMasta33]

Originally posted by Brians256


Just look at the American consumer's decision on front-loading washers. A little up front cost and you make it back (and then some) in reduced utility bills.

Hey now... I'm American and I have a front loading washing machine A little up front cost doesn't cover it. If you want a large basin front loader, it's about 1/3 to 2/3 more for a front loader, but the water savings in a year alone is more than worth it.

[derraa]

do you think a new compound with higher thermal conductivity properties than copper will be released. I know CuSil is a hoax, but what about that stuff that OCWC is working on?

[Brians256]

Ammonia systems? Maybe. With an adsorption system like the original refrigeration systems, you can drive the fluid around by the heat of the CPU but it takes a lot of heat and the system would probably be expensive. Frankly, I don't see normal consumers paying for it.

I see heatsinks being the mainstay for quite some time, with chip designers limiting performance so that the chips will be adequately cooled by normal heatsinks. Speed addicts such as ourselves will probably keep a good niche market going for better and more efficient cooling systems like water or the heat-pipe systems that Joe mentions.

Semiconductors manufacturers are getting better at thinning the backside of wafers to reduce heat resistance, and they are also going to start using different materials such as isotopically pure silicon. That's my best guess, such as it is.

[Butcher]

WTF is this with front loading washers, there are other types?

[Lonely Raven]

Back to the topic a bit...

I work for Compaq Computers right now...and I was talking to
some of our high end Alpha server guys, and they were telling
me that Compaq has some Water Cooled laptops based on
some sort of cooling used in the Alpha Servers....

Now, these guys are not geeks like us, so to them, some sort
of mini phase change system might be looked on as water
cooling....

I've not seen this system, I don't know if it's been released
or if it's just in Beta stage...but the dude I got the info from is
pretty high level Field Engineer. So I suspect something is
happening...I just wish I could find out more!!

I'll post more as I hear it...but right now I'm playing Elf at
Keebler Corp for Compaq. It supposed to be a 3 year contract
so who knows when I'll see real engineers again...(sigh)

[Lonely Raven]

Crap, I thought I'd do some research and see how much of this
has been leaked out, and I found this:

http://www.infoworld.com/articles/pi...417piwater.xml

Note the date....

I guess water cooling laptops is nothing new...

[derraa]

I would have thought that a watercooler is too space consuming for a laptop. Thats interesting though. I think a consumer product, however, must be small, factory sealed, and no possible matinence. Joes phase change idea sounds good. And a front loading washer is is a cloths washer with a sideways bin. Th water then can be far less because as the bin rotates, the water will cover it unlike with a top loader.

[WebMasta33]

As I remember... I thought the laptop cooling system for compaq wasn't water cooling, but rather it was just a few heat pipes that brought the heat from the proc to a heat sink in the back.

[Joe]

yeh, but whats in heat pipes? water or water vapor.

[Detho9000]

Hi everyone

I think you're all thinking of cooling processors that are like the
ones that we have today but only faster. I don't necessarily
agree with this because I think there's going to be a drastic
change in the architecture and style of which processors are
produced as well as the way they process data.

With biologic and even quantum processing quickly becoming a
reality it's only a matter of time before that these technologies
replace what is currently being used. I know it seems like an
overkill to stick a quantum processor that could potentially
process all the SETI work units in a matter of minutes in a
Compaq Presario, but if it's cheap to produce, and the public
wants it, it will be done.

Heh, imagine seeing a Sunday add for CompUSA or Best Buy:
Buy the new Compaq Presario 2600 with the new Intel Quantium Processor!!!
Package includes:
20GB QDR Memory!
60TB Western Digital solid state hard drive!
Microsoft Windows BB
And much more!
Only 599.00$*
*after 100.00 mail in rebate

What in the hell does all your babbling have to do with the future of cooling Detho???

My point is that the trend over the last 70 years is to get smaller,
faster and cooler. I think computer processing will get down to
the point where it may not need any extra cooling at all...just air
running over the chip will be enough.
I'm not saying that the next processor that Intel or Amd produces
is going to be a quantum or biologic, but I think within the next
few years you'll see chip manufacturers starting to veer away
from the relatively inefficient and expensive x86 processor to
more efficient processing methods. So elaborate cooling methods
like water or refrigeration may not be even be a consideration when overclocking and souping future processors.
*sigh* Where's the fun in that?!?

my .02 cents

-D9K

[Joe]

"With biologic and even quantum processing quickly becoming a
reality it's only a matter of time before that these technologies
replace what is currently being used."


I agree that they will be out... but not for another 10 - 20 years.
You do realize that both the technologies you meantioned have been tested to prove that in THEORY they can work. NONE have been successfully tested or even designed in a capacity near what existing PC's are. I seriously think that if you hold out for that, and remain naive to the heat problems of today and in the next few years, you will be in for a rude awakening. The only thing thats of interest that will be coming around in the next few years is SOI, different silicon substrates, and new interconnect designs. All will be focused on smaller/cooler and Faster. But when they make them faster they make them warmer. So for all intents and purposes, the die shrink they are doing is compensated with more heat produced in a smaller space = just a more dense hot peice of silicon.

The CPU's that are being sold today will be used WELL into the next 5 years. Look at Pentium 166's and Pentium 2 266's that are STILL in use. I mean P-166's are going on 6 - 7 years old, and are still in use around in places. I dont see why you think that if a new CPU design comes out that everyone will throw out the "legacy" ones for a new one.

Quantum computing could verywell be one of the most important advances to man when it happens... A technology that profound wont sneak up on us in the next few years... Even IBM who is at the forefront of that R&D said it will be 20 years till they see consumers owning Quantum devices. Biological devices have a LOAD of thier own issues... what abotu the most recent tests where the best they could get was 95% accuracy doign computations with a test sample. Ummm 95% isnt good enough. That technolgoy is nifty but I would put hard money that if it ever makes it to a mainstream product ( which I dont think it will) it will be 10 years down the road.

Technology like Quantum computing is a pipe dream just like Antimater propulsion and power generation. They are both technologies that WILL work, it has been proven, just getting it to work on the scale, and functionality we want will take decades to arrive.

So till then, buy a waterblock, and cool that lil peice of silicon!

[Detho9000]

First of all, quantum computing HAS been succesfully tested and
all the tests can be reproduced.

Secondly, I didn't say that things like biologic and quantum
computing would be in the average consumer's computers
anytime soon. If a quantum processor was available to the public
in even the next 10-15 years I would be really surprised.

What I said was that the current architecture of the x86
processor will go to the wayside for a more efficent processor
within the next couple years and that current elaborate cooling
methods may not be nessecary to OC a more efficent processor.

-D9K

[Joe]

Yeh a Quantum has been tested, I said that, but the scale of the test was VERY small in all cases.

the next evolution of the CPU's is already set in stone... its 64bit RISC or RISC/CISC cores. There will still be X86, there will still be Silicon, there will still be heat issues. But what you will see is cores that are very much like what the CRAY machines use.

These will be multiple, parallel, lower power cores in one CPU case. This is also how the REAL SMT implementation will be from AMD and Intel, basicly 2 - 4 cores on one chip

The technologies you meantioned (Quantum, Bio) are giant leaps that will take place starting in 10 years. Till then it will just be a series of small steps or shuffles forward and backward. I mean if you wanted a cpu that was very efficient and very powerful, why didnt the Alpha succeed? Simple... you must retain backward compatability. Which means what every is brought out will have an X86 subset, with the new RISC sets.

Intels EPIC set SUCKS. It needs FAR too much optimization to do anything slightly fast.

While you are looking at the future as speeding up I am looking at it to slow to a crawl. I think we will hit 4Ghz, and level off big time. At those speeds WHO needs to upgrade... there is NOTHING today like email, and web surfing that taxes a 1Ghz CPU.. at 4Ghz anything short of realtime photorealistic 3D rendering, wouldnt use it.

I think the graphics cards is where you are going to see new developments happening, cause thats an area that could really use it.

It was so much easier back in the day when you know the P2 and P3 were going to come out in the next few years. But I have to tell ya, the next fee years now looks DAMN booring.. I mean really when I see people clocking P4's to 3Ghz it does NOTHING for me...

speeds over 2Ghz are rather pointless now as the CPU's are so much faster than the memory systems ( especialy Rambust with its high latency) that there is almost no noticeable performance gain over a 1Ghz CPU.

What Do I see for the next 5 years? - 64bit x86 RISC/CISC cores. Cause Intel will go to that, and AMD is already planning on that. For Enterprise stuff, 64bit RISC will remain the same as its been for the last 10 - 15 years. Why screw with something that works?

[iceheart]

I remember 6-7 or so years ago everyone said that "CPU's over 200mhz will be useless, they will need liquid nitrogen cooling and the speed won't be any advantage" and stuff like that. I'd like to think that we are more aware of what the future will bring, but somehow I'm not so sure.

[Brians256]

Heat pipes can have any substance in them that (at the internal pressure) easily goes between the liquid and gas states. However, the only one that I've seen mentioned by name is R134a. That was some Japanese heat-pipe heatsink combo.

I bet the reason that they don't work as well as we would like is that they cannot be tuned on-the-fly like a real refrigeration system can. Take this with a grain of skepticism, because I really don't know for sure! However, I think what is happening is that the heatpipe is tuned for a given temperature and heatload, but the CPU really changes that by a large amount.

Oh, and yes. Quantum computing has been tested and it really works! But, so do many technologies. Right now, quantum computing is too expensive to produce yet. Just because a lab can produce one-off copies doesn't make it practical. You want to spend $12 million on a computer that can only make one computation before it self-destructs and has to be rebuilt? Also, the complexity of the current prototypes is basically nil. They cannot do general purpose stuff yet, only simple tests. That's what Joe is getting at (I'm guessing).

[jaydee]

Originally posted by derraa
I was thinking, as processor improve, they generate more heat.

Like said before this is not true, and is just the opposite. CPU's are getting cooler. Look at thre T-Bird and the XP. The newer XP core runs cooler already with the new core. The new P4 2.2gig core runs cooler than the old P4 core. It is just a matter of time when we can use a HS only again to cool the CPU effectivly. Hell, Cyrix/VIA's C3 has already done it but on a poor performance level.

[Joe]

the amount of heat reduction is MINIMAL. look at the .25 to .18um core shrink... that was supposed to be the end-all be-all for cooling... and what do you know, a few months after they .18um shrink we are back at the same heatlevels. Give it a few months we will be right back where we were before.

Heat densities keep rising with every core shrink.

[jaydee]

Originally posted by Joe
the amount of heat reduction is MINIMAL. look at the .25 to .18um core shrink... that was supposed to be the end-all be-all for cooling... and what do you know, a few months after they .18um shrink we are back at the same heatlevels. Give it a few months we will be right back where we were before.

Heat densities keep rising with every core shrink.

The wattage is decreasing so heat will follow, but yes you are correct to a point. The higher the mhz per a core the hotter it will get as it will use more wattage. The XP1600+ at 1400mhz was about 15% cooler than my T-Bird1400 at 1400mhz, but now that the XP is up to 2000+ at 1670mhz the temps are right back up there where the T-Bird1400 was.

Even though the densities are decreasing the wattages used are also decreasing so it really cancels each other out.

BUT it will come to a point that the wattage will be so low it will not produce that much heat even at the peak of a certain core. I don't think this will happen utill they get well below .13 microns though. By then they will probably have some revolutionary core out anyway.

[futRtrubL]

Currect me if I'm wrong but I believe that current PC processors don't actually run x86 anymore, incoming x86 commands are translated into another architecture (RISC I believe). They descided to stick with the installed x86 userbase instead of going for better product (this has happened before). Anyway, even if the architecture was still x86 limited heat gains would be seen on going to a more eficient one, commands could only be streamlined so much.
I don't see wattages declining anytime soon, they will rise until heatsink design allows them to go no higher, then they will level off or increase die size for better transfer or allow cooling of both sides. Maybe heatpipes will be incorporated with hsfs. But beyong that, I don't see maintainance needing systems like water going mainstream.

Edward