Sadly that was one of the things I did benchmark, but didn't write down the results. I read a lot that naive switch is faster because the compiler knows how to optimise them better, but for my architecture and benchmarks the computed gotos were faster