Re:Room-Temperature Ambient-Pressure Superconductor LK-99 pr...
by sigmoid10 on Tuesday August 01, 2023 @04:17AM
(#36953654)
Silicon is not valuable because it is a good conductor, in fact it is actually more like an insulator in its natural state. However, mixing the right materials into it, it becomes a very good semiconductor. So it can let current flow easily or block it effectively, making it a good basis for transistors. The problem here are the normal (usually copper) wires between transistors that cause heat to build up whenever current flows through them. That limits how densely you can pack wires in a chip. With a superconductor, we could not just have much smaller and much faster processors, we could also have many designs that don't require cooling anymore. Imagine a power monster chip like the RTX 4090 running state of the art LLMs locally on your phone. That's the kind of stuff that's at stake here and that's also why everyone and their uncle wants to be author on the original paper.
[ Parent
| on HN ]
Re:Room-Temperature Ambient-Pressure Superconductor LK-99 pr...
by evouga on Tuesday August 01, 2023 @04:42AM
(#36953804)
How much of the cooling cost of current chips is due to wires, vs. the transistors themselves?
Superconductors will not help with the latter.
Re:Room-Temperature Ambient-Pressure Superconductor LK-99 pr...
by __alexs on Tuesday August 01, 2023 @04:50AM
(#36953854)
> The problem here are the normal (usually copper) wires between transistors that cause heat to build up whenever current flows through them.
IIRC with the high frequencies of modern processors switching losses tend to be a larger factor than resistive losses. If you can remove the resistive losses that leaves you with a greater heat budget from switching losses which might help drive up frequency even more.
Re:Room-Temperature Ambient-Pressure Superconductor LK-99 pr...
by sigmoid10 on Tuesday August 01, 2023 @04:57AM
(#36953901)
Switch losses are also resistance losses, they just get amplified by things like the skin effect.