Proof of Optimality based on Greedy Algorithm for Offline Cache Replacement Algorithm

Abstract

The optimal offline cache replacement algorithm is a MIN algorithm that chooses which data item to remove when a new data item is brought from lower level of cache or main memory. The optimal offline algorithm evicts a cache item whose next request is the furthest away. For a particular program, the performance behavior of the cache memory is determined by memory and block sizes and by the nature of the replacement algorithm. Replacement algorithms, however, deserve analysis because they are based on a variety of assumptions and design considerations. This paper presents a simpler proof based on the greedy algorithm. The paper demonstrates that every ideal solution can be repeatedly transformed into the solution provided by the greedy algorithm without increasing the miss of the optimal solution, hence demonstrating that the greedy solution is providing optimality.This paper also presents a new replacement algorithm named Greedy Weight-based Cache Replacement Algorithm (GWCRA), on the basis of the Greedy algorithm and it also incorporates the weighted access-based parameter, recency and frequency. The GWCRA achieves an average speedup of 57.29% when compared to LRU and SRRIP which is 55.58% and 55.65%, respectively.