Why exact LRU is not used
Least recently used performs close to optimal, and real systems approximate it rather than implementing it exactly.
Why?
Least recently used performs close to optimal, and real systems approximate it rather than implementing it exactly.
Why?