摘要
合理的cache设计是缩小处理器和存储器速度差距的主要解决方法,也是影响系统性能的关键因素之一。cache替换策略是影响cache性能的主要因素,目前最常用的替换算法是LRU算法,为了降低模块复杂度和实现的难度,从LRU算法简化出一种PLRU(Pseudo LRU)替换算法。通过采用开源的SimpleScalar仿真工具,对LRU、RANDOM、FIFO、PLRU等各种常见的cache替换算法进行了性能比较和分析,并对PLRU进行实现。实验结果表明,使用PLRU替换算法cache的缺失率与LRU算法基本相同,但是有着更小的面积和更短的关键路径。
Suitable design of cache is the main solution to decrease the processor - memory speed gap and becomes the critical performance factor of systems. The replacement policy is the main factor of cache performance. To simplify the complexity of implementation,we use PLRU replacement policy which come from the most commonly used replacement policy LRU. We also use the SimpleScalar to simulate several replacement policies of cache to compare their performance. And we implement the PLRU at last. Our results show that the PLRU can approximate LRU with much lower complexity for a wide variety of cache sizes and degrees of associativities.
出处
《微处理机》
2010年第1期16-19,共4页
Microprocessors
