The trickiest part of speeding up a program is not doing it,...

The trickiest part of speeding up a program is not doing it, but deciding whether it’s worth doing at all. There are few clear principles, only rules of thumb.

www.joshbeckman.org/notes/465084415