I was told (in a CPU architecture lecture) this is a problem in branch prediction too. With two very similar branches, the time taken to decide starts to outwei

I was told (in a CPU architecture lecture) this is a problem in branch prediction too. With two very similar branches, the time taken to decide starts to outweigh the benefits so something has to step in and say "they're obviously basically the same, just choose the first one" (or some other heuristic maybe to about bias, I don't recall).