Learning Memory Access Patterns

TL;DR

This paper explores using deep learning, specifically recurrent neural networks, to improve memory prefetching by learning access patterns, demonstrating superior performance over traditional methods on benchmark datasets.

Contribution

It introduces a novel approach of applying RNNs to memory prefetching, relating it to NLP models, and shows promising results compared to existing strategies.

Findings

Neural networks outperform traditional prefetchers in precision and recall.

Recurrent neural networks can effectively learn complex memory access patterns.

This work pioneers neural network-based prefetching in computer architecture.

Abstract

The explosion in workload complexity and the recent slow-down in Moore's law scaling call for new approaches towards efficient computing. Researchers are now beginning to use recent advances in machine learning in software optimizations, augmenting or replacing traditional heuristics and data structures. However, the space of machine learning for computer hardware architecture is only lightly explored. In this paper, we demonstrate the potential of deep learning to address the von Neumann bottleneck of memory performance. We focus on the critical problem of learning memory access patterns, with the goal of constructing accurate and efficient memory prefetchers. We relate contemporary prefetching strategies to n-gram models in natural language processing, and show how recurrent neural networks can serve as a drop-in replacement. On a suite of challenging benchmark datasets, we find that neural networks consistently demonstrate superior performance in terms of precision and recall. This work represents the first step towards practical neural-network based prefetching, and opens a wide range of exciting directions for machine learning in computer architecture research.