LRU2-MRU collaborative cache replacement algorithm on multi-core system

The access conflict from different threads or processes for parallel applications, can lead the system performance to degrade for multi-core system with shared cache. Cache replacement algorithm for L2 shared cache can be used to solve the problem efficiently and reasonably. Although LRU cache replacement algorithm can better reflect the locality of program and is widely used, it is not optimal for reducing the shared cache miss ratio and MPKI (misses per thousand instructions), and can not predict whether the data is be used frequently. In this paper, based on the consideration of time prediction, the disadvantage of LRU and the conflict between parallel application and shared cache, we propose the LRU2-MRU collaborative cache replacement algorithm to solve these problems. We use 10 benchmark programs to show that the LRU2-MRU collaborative cache algorithm may reduce the miss ratio of L2 shared cache by 4.61%, and the MPKI is average 4.54% lower than LRU.