Re: The market of ASICs (One GigaKey / Second?)
jbass-/[email protected] Wed, 11 Aug 2004 15:34:05 -0600
| Newsgroups | gmane.comp.distributed-net.hardware |
|---|---|
| Message-ID | <[email protected]> |
Elektron <elektron_rc5-FFYn/[email protected]> writes: > It would probably melt (since that's about the cooling capacity of 10 > big air conditioners, but I'm not sure if you mean 17.5 KW). Depending > on the power consumption for one chip, the biggest problem could be > getting the heat off the chips before they melt. Specific heat of air is very low, and coupling to it as a coolant requires a LOT of energy from the velocity squared component of the problem. 17.5KW/hr is about 15 million calories per hour, or about 4 kilocalories/min. If we allow a 4C rise in coolant from input to output of the processor stack, then we need 1 kilogram/min of water, which is about 132lbs/min, or rougly 16 gallons/min of water as a coolant at room temps. If chilled water is available, it takes considerably less. On the other hand, 65KW/hr which is closer to the machines power requirement if the ZBT/SDRAM and processors are also used, becomes nearly an unmanageable amount of air movement for a 1 cubic foot processor arrary. Chilled water handles the problem without any problem. I'm thinking about burying a thermal sink to heat/cool the house anyway, this just gets incorporated into it. A freon water/water heat pump is pretty practical these days, and doesn't have all the losses of pumping huge volumes of high volocity air. John _______________________________________________ Hardware mailing list Hardware-Ra3b/QYEcJ3d140v2zMXi0fjHoOT/h/[email protected] http://lists.distributed.net/mailman/listinfo/hardware