In this paper, we present a new block replacement policy in which we proposed a new efficient algorithm for
combining two important policies Least Recently Used (LRU) and Least Frequently Used (LFU). The implementation of the
proposed policy is simple. It requires limited calculations to determine the victim block. We proposed our models to implement
LRU and LFU policies. The new policy gives each block in cache two weighing values corresponding to LRU and LFU
policies. Then a simple algorithm is used to get the overall value for each block. A comprehensive comparison is made
between our Policy and LRU, First In First Out (FIFO), V-WAY, and Combined LRU and LFU (CRF) policies. Experimental
results show that the LR+5LF replacement policy significantly reduces the number of cache misses. We modified simple scalar
simulator version 3 under Linux Ubuntu 9.04 and we used speccpu2000 benchmark to simulate this policy. The results of
simulations showed, that giving higher weighing to LFU policy gives this policy best performance characteristics over other
policies. Substantial improvement on miss rate was achieved on instruction level 1 cache and at level 2 cache memory.
Authors
Adwan AbdelFattah
Aiman Abu Samra
Pages From
16
Pages To
21
Journal Name
International Arab Journal of Information Technology
Volume
9
Issue
1
Keywords
Cache memory, replacement policy, LRU, LFU, miss rate.
Abstract