I have a pseudo-physical ODE system consisting of essentially dozens of masses each connected to subset of the rest with springs (with friction), and I'd want to track evolution of the system. It would appear that preprocessing step of NDSolve grows quickly with the size of the system. Would there be any interesting Methods or such to work around this?

It would seem that things get particularly complicated in three dimensions, two is still quite fine.

LOL, no. The slowdown actually originated from Simplify!

@b3m2a1 This is great! I am really interested in the details of paclet development.

drive.google.com/file/d/1cI_acZNRN1Piq8e9hkvjzL0b7jhJMFnr/view

(too big gif for imgur)

Discretized surface wants to be flat and consist of equilateral triangles, but you can't get everything!