Comment by AlexSW

I'm struggling to confirm that my intuition of why the compiler optimisation is faster than the naive switch expansion.

Is it about the layout of where the assembly instructions end up, and spacing around them? Or the CPU pipelining working better? Or...?