Printable View
Sun Microsystems has discovered a way to connect a computer's chips face-to-face rather than linking them through the web of fine wiring currently laid down on circuit boards, a Sun official said.
By eliminating the wiring, the discovery could speed up data movement by as much as 100 times while reducing power usage, he added.
Full article HERE?.
Is this article some sort of hoax?
I realise it is posted in good faith, but if you follow the link to Sun's web page, then look at Sun's announcements or at their link to press comments about Sun's recent announcements, there is absolutely no mention if this at all.
In any case, it sounds like bullsh*t in the first place.
nah, its not a hoax theres some more info here they're trying to use capacitive coupling instead of wires.
Hmm, an effect with a range of only a few microns, that means that the interface would have to be awfully close to the surface of the chip, and you would also have to have extremely accurate placement of one chip with respect to the partner chip. Sounds as if it would be easier to physically combine the chips into one device, but perhaps that's the intention. I can see such a device costing megabucks though.
I would personally like to know how they solve the heat problem...
but it's a good idea
Submerged in a liquid nitrogen bath. :)Quote:
Originally posted by james_bond_rulez@26 September 2003 - 11:21
I would personally like to know how they solve the heat problem...
hehe we wont have any more big ass mobo's....but our cases are gonna be filled up with fans lol...
DWk
I think the days of air cooling will soon be drawing to a close. But this new method doesn't sound like it produces any more heat than equivalent conventional designs, I think the problem will more likely be space on the chip and as lynx said alignment issues in mass production.
If not air cooling, then what do you see as the next step?Quote:
Originally posted by ilw@26 September 2003 - 06:31
I think the days of air cooling will soon be drawing to a close. But this new method doesn't sound like it produces any more heat than equivalent conventional designs, I think the problem will more likely be space on the chip and as lynx said alignment issues in mass production.
water cooling is always coming closer to most users.
How about a heatsink carved from dry ice (solid C02).
Livy, I don't think that water cooling is really all that practical for the average home user. Outside of the fact that you have to monitor and replenish the H2O level, a radiant heat disperser is not terribly efficient beyond ambient room temp. Unless the whole system is dramatically improved from what I've seen available ( and it could be done certainly) then I don't think so...Quote:
water cooling is always coming closer to most users.
lynx, didn't someone post pics of just this- done by some Finnish overclockers?Quote:
How about a heatsink carved from dry ice (solid C02).
I have a feeling that you would have to replace dry ice at a much faster rate than water in a water-cooled system. In any case, a properly designed liguid cooled system (not necessarily water) should not need 'topping up', since it should not leak.
In any case, it is unlikely that you would want to cool the processor below ambient temperature because of the possibility of condensation, which is often a problem when using peltier-effect devices.
Most liquid-cooled systems fail simply because they do not have sufficient mass in the heatsink, which results in rapid temperature changes when the pump/water-level fails. If the heat sink is large enough, temperature rises can be detected and the system shut off long before permanent damage occurs. In any case, a failsafe system would have a both coolant-level and flow warnings built in.
I think this is largely irrelevant for the board users, the implication is that these chip-to-chip connections are going to be for super-computers with massively parallel processor arrays, not home pc's.
even water in room temp has excellent heat capacity (something like it takes 1000+ joles of energy to raise the temp of 1 liter of water 1 degree celcius)
so even if you system is somewhere like 40-50 degrees, water can bring it down to near room temp if there is enough water.
the problem is how to get the water to recycle the heat, get the heat out.
Unless someone comes up with something bettter fast, then water cooling is IMO definitely the next step, as lynx pointed out you don't have to cool below ambient temp, the point is to conduct any heat above ambient away quickly, and even though water is not that great a conductor it is significantly better than air. The higher density of water means that moving water can dissipate heat very quickly, and then the water can be pumped to a much larger heatsink than would be possible on a processor. I think the problems gonna be how do we cool our rooms
james_bond_rulez if i remember well I think the specific heat capacity of water is 4200 J/kg/C , water's very high shc is one of the factors in its favour in cooling.