Applying Deep Learning to the Cache Replacement Problem
Applying Deep Learning to the Cache Replacement Problem
复制标题
DOI：
10.1145/3352460.3358319
复制
发表时间：
2019-10
期刊：
影响因子：
--
通讯作者：
Zhan Shi;Xiangru Huang;Akanksha Jain;Calvin Lin
中科院分区：
文献类型：
--
作者：
Zhan Shi;Xiangru Huang;Akanksha Jain;Calvin Lin
尽管深度学习在许多领域取得了成功，但它不太适合用于硬件预测器，因为这些模型太大且速度太慢，但本文展示了如何使用深度学习来帮助设计新的高速缓存替换策略。我们首先证明，对于缓存替换，强大的LSTM学习模型可以在离线设置下提供比当前硬件预测器更高的精度。然后，我们执行分析来解释这个LSTM模型，得出一个关键的见解，使我们能够设计一个简单的在线模型，使离线模型的精确度与数量级的成本相匹配。结果是Glider缓存替换策略，我们在一组来自SPEC 2006、SPEC 2017和GAP(图形处理)基准测试套件的33个内存密集型程序上进行了评估。在单核设置中，Glider的表现优于第二届缓存替换锦标赛的顶级终结者，将LRU的未命中率降低了8.9%，相比之下，Hawkeye的未命中率降低了7.1%，MPPPB降低了6.5%，Ship++降低了7.5%。在四核系统上，Glider将IPC比LRU提高了14.7%，相比之下，鹰眼提高了13.6%，MPPPB提高了13.2%，SHIP++提高了11.4%。
Despite its success in many areas, deep learning is a poor fit for use in hardware predictors because these models are impractically large and slow, but this paper shows how we can use deep learning to help design a new cache replacement policy. We first show that for cache replacement, a powerful LSTM learning model can in an offline setting provide better accuracy than current hardware predictors. We then perform analysis to interpret this LSTM model, deriving a key insight that allows us to design a simple online model that matches the offline model's accuracy with orders of magnitude lower cost. The result is the Glider cache replacement policy, which we evaluate on a set of 33 memory-intensive programs from the SPEC 2006, SPEC 2017, and GAP (graph-processing) benchmark suites. In a single-core setting, Glider outperforms top finishers from the 2nd Cache Replacement Championship, reducing the miss rate over LRU by 8.9%, compared to reductions of 7.1% for Hawkeye, 6.5% for MPPPB, and 7.5% for SHiP++. On a four-core system, Glider improves IPC over LRU by 14.7%, compared with improvements of 13.6% (Hawkeye), 13.2% (MPPPB), and 11.4% (SHiP++).