@yiningkarlli@mastodon.gamedev.place
2026-05-10 06:09 UTC
@zeux@mastodon.gamedev.place My thinking was: At least on theory side, programming languages is mostly math (lambda calculus, context free grammars, etc). ML is mostly math too (gradient descent, statistics, etc). Sure you have to deal with files, but dealing with the files isn't typically thought of as like, a core part of the problem to be solved.
Graphics is like: so we have a bunch of .obj files and a bunch of texture files. In graphics we consider data movement as a core problem since it really impacts perf.
Replies (1)
-
@yiningkarlli@mastodon.gamedev.place 2026-05-10 06:22
@zeux@mastodon.gamedev.place So as an example: if you are designing a programming language, your starting point is thinking about context free grammars, syntax trees, expressions, etc, which are all math. If you are designing a neural network, your concern is mostly about network topology, loss functions, activation functions, etc, which are all math. If you are designing a renderer, you better be thinking texture caching and stuff early on, and you don't actually need to worry about shader math until much later.