10 Comments
User's avatar
Abhinav Upadhyay's avatar

This is something I wrote recently. A comparison between the costs of context-switching between CPUs and GPUs. TLDR: GPUs have zero-overhead scheduling, i.e., they switch from executing one thread to another without any overhead. CPUs on the other hand have to do quite a bit of dance to switch from executing one process to another (on a given core).

https://twitter.com/abhi9u/status/1711053912600707543

Alejandro Piad Morffis's avatar

Thanks, Abhinav! Would you take us up on the challenge of writing a short Note with an intuitive overview of the article suitable for all audiences?

Michael Woudenberg's avatar

In the current zeitgeist of AI more often than not people are lacking a basic contextualization of the layers of AI, the complexity of AGI, and where we currently are. This essay provides a solid foundation to work from.

https://www.polymathicbeing.com/p/the-layers-of-ai

Michael Woudenberg's avatar

Do you want that distilled into a note?

Alejandro Piad Morffis's avatar

As you prefer! Ideally something short that can be shared with anyone and points to the larger piece ;)

José J. Rodríguez's avatar

I publish on Tuesday. Does that count? 😆

Alejandro Piad Morffis's avatar

Yeah but only if you say it!