Hacker News new | past | comments | ask | show | jobs | submit login

If this is the transistor moment what technology would an IC equivalent require?



First we need a new paradigm of computing that is consistent with how light behaves.


Some questions from someone who is simply interested in the topic and seeks to expand their understanding.

1. For electrical circuits we are interested in properties like resistance, capacity, inductance etc. for passive elements and in the characteristic curves for active ones.

What are their equivalents in photonic circuits? Which properties matter here?

2. In digital computing I think we abstracted the electrical analog properties and the differential equation level of the problem away pretty much. Can't we achieve that with photonic ones as well? Or do you mean a paradigm shift in higher level things like memory bandwidth and latency vs caching etc. ?

3. When I look at the state of photonic circuits I think miniaturization and fabrication are the most obvious problems. Let's assume for a moment that these can be solved, what else makes photonic circuits worse than electric ones?


Unfortunately I am no expert. Most of these questions go beyond my knowledge.

But you should definitely watch this video [0].

[0]: https://youtu.be/EwueqdgIvq4


Imprecise neural networks?


3-4 football fields




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: