Notes: ### Memoization - indiscriminate memoization => ~2x slower on heterogenous workloads - it may fare better on more regular loads, like program execution for K
