Hacker News Reader

Google TPUs use optically routable fibers like circuit switching to be able to change the whole topology of the setup depending on training load/matrix dimensions.

View on HN

I remember reading about their setup and it blew my mind. They have MEMS mirrors that can rotate to redirect one of thousands of fiber lines into any other fiber line by using other optics. That way you can take in like 1000 lines and optically connect it to any of the other 1000 lines.

I wonder how did they pull that off. When I imagine end of a fiber optic, the light doesn't come out of it nicely in one direction.

TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings

https://arxiv.org/abs/2304.01433

Thanks for the source.

Basically they use lens array (one lens per input fiber) to focus each specific beam on its assigned mirror and mirror directs the light to the chosen output that also each has a lens in front of the fiber going out.

Reply on news.ycombinator.com