Open franz1981 opened 7 months ago
Some of the proposed improvements are coming from https://github.com/cescoffier/fortune-benchmark/tree/main/fortune-reactive-mariadb/src/main/java/me/escoffier/fortune/reactive/mariadb so IMO they are not actually so valid (eg the cost of decoding in digits) , but given that I was having fun reading a nice article on it I said "why not trying implementing it?"
It contains 3 improvements (unfinished):
MinJumpTableSize
(see https://stackoverflow.com/questions/44988841/force-tableswitch-instead-of-lookupswitch for more info): we should find a way to reduce the number of branches here or find a different way to perform it, because the branch predictor AND the JIT doesn't enjoy it