Optimizing for size versus performance often eliminates optimizations like unrolling loops.

At the assembly level, it's sometimes true that a single instruction or smaller sequence of instructions takes more cpu cycles.

Analyze the different instruction sequences here:

https://www.nxp.com/docs/en/supporting-information/MC680X0OP...

Most of the options listed have a disadvantage which is why they aren't enabled by default. The LTO and codegen unit options increase compile time, removing std is worse for the development experience, using xargo (or cargo's build-std feature) requires a C compiler to be present and increases compile time (as std is being compiled).

Well, not being able to use the standard library certainly makes it more difficult to write safe and performant programs.