Quote:
Originally posted by Bitman
I suspect we might see the the diamond and sapphire stuff, just because it can handle the huge temperatures that the 50 GHz chips will produce.
|
Silicon's advantage is that silicon dioxide is glass - one of the best insulators available. Current demands are for new high-k dielectric superior to glass. IBM had one from a Dupont material. But in production, it pealed off the wafer. IBM abandoned to process with terrible consequences to its customers. Intel is said to have a secret material; but has yet to even say what material is uses.
We were designing an early 1980s computer using silicon on saphire. Technology is that old. But diamond and saphire based semiconductors still have limitations such a current leakages not found in silicon - because of glass.
Largest reason for heat is also same reason for speed limitations - leakage - which is why new high-k materials are required. Still nothing has come close to replacing the king - silicon.
So what was that new transistor? Have yet to read anything. Just that the latest (51st) International Solid State Circuits Conference appears to have suggested Moore's law has hit a brick wall. Not much room left to make transistors smaller. Atomic sizes put a limit there. Maybe in massively interconnected ICs in one package or layering transistors or breakthrough material. Talk is now of one trillion transistors in one package.
Just that nothing promising appeared yet that will continue Moore's law. So much for the new Intel optics transistor.
Moore's Law demands a 10 gigahertz computer very soon. This is mind boogling to one who once worked in the kilohertz and single digit megahertz range - and used Mecl technology for 10 Megahertz - exotic - digital ICs. At 10 Ghz, tiny traces now become tuned antennas. Computer chips of 150 watts heat dissipation are an unfortunately possibility. At what point can this no longer be advanced? An so the question of sublight processors. Or has Moore's linear law now started to become a decaying exponential?