Hacker News Reader

Comment by marcus_holmes | Hacker News Reader

Optical computing. I remember talking to a bloke working on it back in 1990, and he said they had a working optical transistor so all the hard bits were done. Soon we'd have computers that used multiple frequencies of light to process operations concurrently on the same hardware!

Still waiting...

View on HN

In a sense we did get it, the research into optical computing didn't lead to photonic cpu's, any advantages of optical computing are mute when each building block can't scale below a few square micrometers due to limits on the physics.

However, the research did find it's way into telecom, high speed multiplexing switches, with massive bandwidth in a compact formfactor would not have been economically viable without photonics.

He's right! There are transistors which have light at different frequencies on the same hardware. This paper[1] talks about how they've achieved 1.125 Tb/s throughput with various techniques.

So he's right about optical computing, but just not in the way you anticipated :)

Just extremely fixed functions with mind-blowing capacity.

[1]: https://www.nature.com/articles/srep21278

These computers do exist on tabletops in labs; the problem is that it's comparatively difficult/expensive to mass-produce and miniaturize these components compared to silicon (where you can etch billions of transistors at once on a modern CPU). There are really cool things you can do, especially in terms of continuous/non-discretized-time computations, but I wouldn't hold your breath for this to become mainstream.

Are you talking about analog computing? We tried that both with electronic and mechanical computers and it obviously didn’t work out since nobody uses them anymore. What benefits do optical computers have in that regard?

I once heard somebody talk about analog computers who is old enough to have used them.

The impression I got was that they did work excessively well on a very narrow niche of problems (simulations, mostly). But "programming" them was a huge pain, and for the general purpose arena, they were not useful.

They would be comparing them with the digital computers of the same time.

There's a reason people used analog computers back them, and there's a reason nobody uses them now. Most of the reason is about how noise interferes with small components, and it's a physical reality that won't go away.

It's already being used commercially: http://www.lucedaphotonics.com/en

Reply on news.ycombinator.com