贪婪双尺寸频率算法的优化与改进

针对贪婪双尺寸频率算法不能反映频率未来走势的问题，在贪婪双尺寸频率算法的基础上，提出了一种新的缓存替换算法。该算法通过对文件和系统的访问次数进行周期性的计数，得到了文件的平均周期访问频率、最近周期访问频率和周期相对频率，并通过周期相对频率来体现文件访问频率的未来走势，弥补了传统的贪婪双尺寸频率算法的不足。新的缓存替换算法具有良好的适应性，提供了周期次数N和频率影响程度k这两个参数。使用者可以通过调整这两个参数让算法适应实际的访问情况。在符合齐普夫定律的数据下进行实验，结果表明该算法比传统的贪婪双尺寸频率算法具有更高的缓存命中率。

Optimization and Improvement of Greedy Dual Size Frequency Algorithm

To solve the problem that the Greedy Dual Size Frequency （GDSF） algorithm can not reflect the future trend of frequency， we proposed a new cache replacement algorithm. The algorithm obtained the average cycle access frequency， the most recent cycle access frequency and the relative frequency of the file by periodically counting the number of accesses to the files and the systems. This method reflected the future trend of the file access frequency through the periodic relative frequency to overcome the shortcomings of the traditional GDSF algorithm. The proposed new cache replacement algorithm had good adaptability to help the user adjust parameters of the number of cycles N and the degree of frequency influence k for adapting the actual access situation. Experiments based on the data of Zipf’s law show that the algorithm achieves a higher cache hit ratio than the traditional GDSF algorithm.