ABSTRACT
Data should be cheap to store and fast to retrieve or process. Unfortunately there are fundamental reasons why we cannot design a computer memory that is at the same time cheap, compact and fast. No signal can propagate faster than light, and wires reaching every single memory cell would need too much room. As a result, given a storage technology and a desired access latency, there is only a finite amount of data reachable within this time limit. The simplest, and most widely used compromise to escape this problem is the memory hierarchy.
