The Research Of LRFU And Its Adaptive Algorithm

Cache as an important technology to improve the performance of the computer can avoid frequent access to the database and disk files, it is an effective means to make up for the difference between the processor and storage systems. However, with the rapid development of the processor, it needs higher performance of Cache. Computer is widely taken to expand capacity cache to improve cache performance, because of hardware processing technology restricted by technology and power consumption, it has become a hot spot to improve the cache performance by design good caching algorithms.After the traditional classical algorithm and data access pattern research,inappropriate selection of strategy will lead to cache thrashing, data block has just been replaced and soon will be visited again, it needs to be transferred to the cache and get some other block out of the cache, processor wastes a lot of time in dispatching data block when frequently operate in the cache and the main memory replacement, also seriously affect the system overall performance. A good strategy can avoid the "bump" phenomenon, which can effectively improve the performance of the cache and the overall performance of the system.In view of the deficiency of LRFU(Least Recently Frequently Used) algorithm and its adaptive algorithm, some related improvement strategies are put forward, and the improved strategy is divided into the following parts:1) Based on the local quantitative analysis model, analysis of the data local strength and give the relevant adjustment of the strategy, improve the adaptability of the algorithm for different access patterns.2) The 2 way set associative architecture: the queue was divided into two queues for hot and cold, improve the algorithm to capture the ability of low frequency and high frequency, the abolition of restrictions on hot and cold queue capacity threshold, the two queues using competition mechanism, and the introduction of ARC(Adaptive Replacement Cache) prediction mechanism and improve the cache hit rate, enhance the flexibility and adaptability of the algorithm.3) Dynamic change replacement of key factors: according to the deficiencies of the lrfu algorithm and adaptive algorithm with linear access patterns, combined with local quantitative analysis model and arc algorithm in dynamic adjustment of ideas and judge the current access pattern, dynamic change replacement of key factors, to improve the hit rate of cache.The experimental results show that: the improved method has better adaptability in the linear, probabilistic and local access modes, and improves the overall hit rate of the cache.