A Survey on Page Prefetching Techniques for Modern Memory Systems

The memory resources of operating systems have become a critical bottleneck for system performance. Alleviating frequent page faults is an important direction for breaking through this bottleneck, and page prefetching has emerged as an effective strategy. By loading the required pages into memory in advance, prefetching can significantly reduce the occurrence of page faults. An ideal prefetcher needs to balance accuracy and timeliness. However, inaccurate or untimely prefetching may lead to memory pollution or increased system I/O overhead, thereby affecting overall performance. Therefore, developing efficient and accurate prefetching strategies is crucial for improving the performance and user experience of memory-intensive applications. This paper will introduce prefetching-related work from three aspects. First, it begins with an introduction to related work from three different perspectives on optimizing I/O prefetching. Second, it presents several case studies of prefetching optimization from alternative viewpoints. Finally, in response to the evolution and iteration of storage systems, it introduces innovative work tailored to these novel scenarios.