In bytecode interpreter VMs, one often encounters the "computed goto" [1] pattern in use to dispatch opcodes. This is one that tends to be a little faster than

In bytecode interpreter VMs, one often encounters the "computed goto" [1] pattern in use to dispatch opcodes. This is one that tends to be a little faster than a switch statement, enough to matter in the dispatch inner loop. Of course, if you are going to JIT compile the bytecode, that'll usually be a lot faster. But at that point you're changing one form of low-level wizardry for another. [1] https://eli.thegreenplace.net/2012/07/12/computed-goto-for-efficient-dispatch-tables https://eli.thegreenplace.net/2012/07/12/computed-goto-for-e...