The Perfect Block..

[bds71]

@Kayin: i love seeing those types of articles - and i have actually read that particular article before today (though i read it again just to see what has developed since 2006). unfortunately, it's sad to see that we had so many potential candidates and none came to fruition in the last 6 years. (for whatever reasons). if i had to guess, i would say (technology-wise) that our best otpions (and closest to production ready) would be vapor chamber and/or microchannel. material-wise, we are a long way (5-10 yrs, if not more) from **real** graphene/carbon nano tube products that begin reaching their theoretical potentials. the stuff we see now is doubtfully actual nano-tubes (its become the new marketing buzz word so everyone and their dog has "nano-tubes" - yea right). at our current levels of technology, water probably IS as good as it gets - that's why i think vapor chamber and microchannel tech will be the popular solutions in the near future, because they utilize the properties of steam/water. everything else is TIM related (getting the heat from the heat spreader to the cooling device), and i don't think indigo xtreme will be beat (by much anyway) any time soon. especially with gimmicks. one thing that *i* think could beat out IX (way down the road) is graphene. i think *that* is as good as it gets (or will ever get). it is the theoretical "perfect solution." unfortunately, it would not be possible to make a whole block from the wonder material due to its anisotropic features, but as a TIM it would have no peers. note: experiments have already shown devices used with graphene heat spreaders have "self-cooling" properties

sorry - i love "tech" talk like this. most people read the news blogs....i read tech reviews and nano tech blogs (and water cooling forums)

[plasmatique]

Received this message of Aquacomputer

All new kryos beginning middle of this month will be completly vacuum sealed.

Stephan Wille
Geschäftsführer/General Manager
Aqua Computer GmbH & Co. KG
Gelliehäuser Str. 1
37130 Gleichen

[asura]

General Electric (no they don't just make lightbulbs) are the biggest player in the development of artificial diamond production. CVD chemical vapour deposition makes for the fastest and easiest growth, then usually put through a HPHT (high pressure, high temperature) process to anneal the synthetic stone - certainly the ones made for jewellery usually are. No idea what the largest size of stone producible would be, but I'm sure that with a properly designed substrate you could produce a coldplate with either micro fins or pins; or chop channels into a flat slab (practically no need for a substrate or design thereof)

What's more you could have it in any colour you'd like; ain't that a joy.

Even this is quite expensive, moissanite is a much cheaper common(ish) diamond stimulant, which is thermally conductive, how conductive I don't know, can't find any solid values it's often quoted as being "similar to diamond" if it is high enough then it would be a cheep alternative to diamond, it' still hard (9.25) but is easily cut by diamond tools.

[NaeKuh]

Just curious, if money was no option, what would be the perfect block?
I think I read somewhere diamonds have better thermals then copper or other metals used, or that may of been inregards to something else with computers.

We thought it was graphite... then tony from OCZ showed us how bad that idea was..
Turns out that graphite was hydrophobic.. and well... thats bad if u want water to transfer heat.

I would assume the perfect block would be something with an insane amount of thermal transfer abilities, so yes a diamond block would do it, however it would kill you.

Most likely im going to look into the next generation of carbon nanotubes as the perfect candidate, however, i have a feeling we wont be using water on those.

I was talking to skinnee a while back ago, and i said.. if we do happen to use carbon nanotubes, we will have pony tails sticking out of the real 120mm slot.
As the tubes will be so efficient, it would pull any and all heat outside, and the long strands can disperse that heat though convection and pull greater then whatever performance we see on water now.

However cam gave me a funny look and said NO... i dont want a pony tail on my PC.

[MrBean]

@ Mods, please delete, duplicate post!

Ta...

[MrBean]

Isothermal Systems Research (ISR) have developed a CPU cooling technology called Spraycool, which replaces traditional heatsinks with a rather novel form of liquid cooling that utilizes jets of atomized fluid sprayed onto a cold plate in a fully enclosed environment

@ Kayin: You are probaly alluding to testing this, or a similar technology.

I have replied in an earlier thread about your reference to "think submarine propellers' cooling that you were testing, did you read my comments then?

You are probably looking at atomizing the water, but also lowering the pressure that would cause the water to "boil" at a much lower temperature....

Hot, cold?

[badbaker`]

So really, even if we had the means and money for a perfect block there isn't a whole lot of gain over the best thats out now?

[frank anderson]

Well this thread peaked my interest a bit so I did some research, gotta love the internet and google..

I found this..

Thermal Conductivity of Metals..

Cost of production and profit aside, which most companies and manufacturers base their work models on, given say you have access to this and money is not a object.. Keeping in mind that Temperature and Thermal Conductivity are two separate things and both needs to be taken into consideration.

from a Metal perspective, pure copper wins hands down as it has both the flexibility and best balance of heat transfer and ability to hold heat, so ignore my previous comment regarding gold being the best..

The next step would be diamond or graphite, but I can imagine what a nightmare it would be to handle this.. Laser cut facility? If money was not a object and you are after absolute performance, the cost of having a diamond block made would probably be in the range of $1000 USD give or take, yea I have a relative that owns a precision tool bit manufacturing plant and his specialty is... precision diamond cutting tools... His cost would be roughly $1000 give or take, so even if it does hit retail, assume a rough 100% mark up and we are talking about almost $2000 just for a plate... not to mention the work involved of having it cut and shaped.. For the same price, I can think of other methods to have your rig cooled to sub zero temperatures and look just as hot..

If you are after bling, sure having a diamond plate will sure grab attention, if people can see it... Can they? now drop $2K on a phase change unit and your computer will look like a Harley Kawasaki hybrid on steroids.

[Church]

Hmm, why do i think that cost of diamond block would be way way more then even $10K? I doubt that big enough single piece diamond crystals, even if artificially made, are cheap enough.
.. $4,000 per carat. That's about 30% less than the price of a natural diamond ..
1 carat was ~ 0.2grams ~ 5mm big. While @lab they had grown up to 25 carats, "For economic reasons, the growth of most synthetic diamonds is terminated when they reach a mass of 1 carat (200 mg) to 1.5 carats"

So, to sum it up, you can make 24*7 multi stage phase cooling (extreme cooling with very subzero temps) for less then even 5% of diamond waterblock (which never will cool below ambient temps) costs. I rest my case.

[asura]

Oh completely impractical I agree (though I disagree with the costs quoted in the first link by about 20% for the natural, and more than that for the synthetic - it's still entirely pie-in-the-sky)

The reasons for stopping at 1-1.5 carats are twofold they want to guarantee getting 1ct cut stones; going from say 98pt to 1ct (That's a difference of .004g) has a cost increase of 10-30% per ct due to the status attached to a 1ct stone. And anything too much over 1ct will be a known stone - e.g. every 5ct stone will have at least one grading report, and can probably be traced back to its point of discovery; conflict diamonds excepted.

Here endeth the lesson.

Copper, silver, or (more likely) an alloy of the two is the best bet for current technology.

[gmat]

So really, even if we had the means and money for a perfect block there isn't a whole lot of gain over the best thats out now?

Basically, as waterblocks are concerned, exactly. There is probably still room for evolutions (micro-fin manufacturing techniques for example) but you wont have a revolution. Copper is pretty much the best material available (and it's far from cheap by the way), and the only way to leap ahead in performance is to go phase change.

[Church]

Imho at LC there should be other areas to enhance instead, because as others already noted, block innovation performance-wise is close to standstill, with little room to grow.
In my eyes there are two bigger ones:
*) universality and getting prices down (single specific gpu or single specific motherboard model fullcover blocks that get thrown out during upgrades are like stick in my eyes)
*) experimenting with rad sizes/forms/mounting ways to cover wider count of miscellaneous case models / growing frontal area by going wider (eg. new 180 & 200mm fan rads by magicool) and so on.

[iboomalot]

in my mind the following would kick ass

entire metal area would be silver with diamond micro dust infused with the metal. The block would be very tall to allow for 2" tall cooling fins.

This would give alot of weight , alot of surface area , and the metal/diamond mix/alloy would transfer heat like no other.

God knows what it would cost but I bet it would wick heat like no other

[Church]

very high fins = no concentrated flow where it counts, and problems with getting heat that far from coolable object. and geesh. you guys still hadn't dropped diamond idea. what's next from ressurected wunderwaffe from zombie threads, mercury as coolant? :/

[Kayin]

Actually, there is a diamond/copper matrix out there that has thermal conductivity as much as three times copper's base conductivity. You're still talking WAYYYY more expensive than copper. Though, I seem to remember liquid gallium loops for Opterons that were ex-US military...

http://www.techpowerup.com/?3105

http://hardforum.com/showthread.php?t=1132942

http://it.slashdot.org/story/08/07/2...-water-cooling

[asura]

Copper is pretty much the best material available (and it's far from cheap by the way)...

Have to disagree with you on your second point here, copper is very cheep.

It's about £5.85/Kg by bullion, way less than half the cost of say nickel or tin - yes it's about three or four times that of aluminium, but wait...

My primary working material is silver, that's about £800/Kg for sheet/wire; £600/kg bullion - a little over 100 times more expensive than copper.

Secondary material 9ct/18ct gold, that's 15,000/Kg for 9ct or 29,000/Kg for 18ct both sheet/wire cost; 24ct gold = £30,000/Kg bullion or over 5,000 times the cost of copper.

I don't work in platinum yet, and hopefully won't for a number of years, and not just because of the cost - £35,000/Kg for sheet/wire or nearly 6,000 times the value of a block of copper massing the same.

Copper is cheep!