Yeah, Go's (Thompson's NFA) algorithm's edge comes when trying to match pathological strings, which I also explained in this comment here: https://lobste.rs/c/lz68me

So, the performance would be similar to other implementations when trying to match small or non-pathological regexes.