[fpoling]

I wonder are there any compilers that recognizes switch-in-loop pattern of interprets and optimizes it to be branch-predictor friendly using multiple indirect jumps?

[ack_complete]

MSVC can do this for x86 if the cases are densely packed and the default is blocked: https://gcc.godbolt.org/z/GPo9fMx8G

However, this optimization is less effective on modern CPUs with more powerful indirect predictors using global history.

[dhosek]

I believe that Rust does this which implies that it might be common to the clang back end.