基于经验搜索的多级存储层次优化

Memory Hierarchy Optimization Based on Empirical Search

存储墙是影响单机性能优化的重要因素,其缓解依赖于对程序进行存储优化。论文提出基于经验搜索的多级存储层次优化方法,将优化多级存储层次问题转化为对优化参数的经验搜索问题,并基于遗传算法选择全局最优解。实验表明,该技术可以自适应不同应用程序,大大降低存储访问时间,降低存储因素对程序性能的影响,从而有效地缓解存储墙问题。

Memory wall is an important factor that effects program performance optimization.And its alleviation relies on memory optimization of the program.We propose the approach of Memory Hierarchy Optimization based on Empirical Search.h turns the problem of optimizing across multiple levels of the memory hierarchy to an empirical search to optimization parameter problem,and selects the best overall solution.Experiments show that this approach can greatly decrease time on memory access and memory access factor to performance,therefore,effectively alleviate the problem of memory wall,moreover,it can automatically adapt different programs.