Memory was just one example.
When they can lay these out on a chip to make VLSI circuits, they could build massively parallel reconfigurable devices. Imagine if the folks at Starbridge Systems, who are already building reconfigurable ASICs, made their next Pensa Processor with this stuff. Obviously I think the Pensa stuff is interesting already with just regular silicon.
Hot Tamales
The main problem regarding mechanical nanocomputers as I understand it is that the heat generation problem isnt just going to go away. Even if moving bonds dont have friction, moving parts will still generate thermal radiation. Frequencies of operation will have to be selected with consideration of the normal modes of the components in motion. Add to that the regular heating that results from resistive current flow and you get a major heating problem. The reconfigurable circuit system would help it recover from real-time circuit failures, but I still wonder how long such a package would last. They might end up having an in-use half-life of a few days, such that you have to keep replacing them or use them only at 1% capacity to stretch out their effective useful lives (and keep the data intact long enough to port!).
Slashdot Mirror
Memory was just one example.
When they can lay these out on a chip to make VLSI circuits, they could build massively parallel reconfigurable devices. Imagine if the folks at Starbridge Systems, who are already building reconfigurable ASICs, made their next Pensa Processor with this stuff. Obviously I think the Pensa stuff is interesting already with just regular silicon.
Hot Tamales
The main problem regarding mechanical nanocomputers as I understand it is that the heat generation problem isnt just going to go away. Even if moving bonds dont have friction, moving parts will still generate thermal radiation. Frequencies of operation will have to be selected with consideration of the normal modes of the components in motion. Add to that the regular heating that results from resistive current flow and you get a major heating problem. The reconfigurable circuit system would help it recover from real-time circuit failures, but I still wonder how long such a package would last. They might end up having an in-use half-life of a few days, such that you have to keep replacing them or use them only at 1% capacity to stretch out their effective useful lives (and keep the data intact long enough to port!).
JF