Learning Semantics, Not Addresses: Runtime Neural Prefetching for Far Memory

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Huang, Yutong, Guo, Zhiyuan, Zhang, Yiying
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915533290471424
author Huang, Yutong
Guo, Zhiyuan
Zhang, Yiying
author_facet Huang, Yutong
Guo, Zhiyuan
Zhang, Yiying
contents Memory prefetching has long boosted CPU caches and is increasingly vital for far-memory systems, where large portions of memory are offloaded to cheaper, remote tiers. While effective prefetching requires accurate prediction of future accesses, prior ML approaches have been limited to simulation or small-scale hardware. We introduce FarSight, the first Linux-based far-memory system to leverage deep learning by decoupling application semantics from runtime memory layout. This separation enables offline-trained models to predict access patterns over a compact ordinal vocabulary, which are resolved at runtime through lightweight mappings. Across four data-intensive workloads, FarSight delivers up to 3.6x higher performance than the state-of-the-art.
format Preprint
id arxiv_https___arxiv_org_abs_2506_00384
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning Semantics, Not Addresses: Runtime Neural Prefetching for Far Memory
Huang, Yutong
Guo, Zhiyuan
Zhang, Yiying
Machine Learning
Distributed, Parallel, and Cluster Computing
Operating Systems
Memory prefetching has long boosted CPU caches and is increasingly vital for far-memory systems, where large portions of memory are offloaded to cheaper, remote tiers. While effective prefetching requires accurate prediction of future accesses, prior ML approaches have been limited to simulation or small-scale hardware. We introduce FarSight, the first Linux-based far-memory system to leverage deep learning by decoupling application semantics from runtime memory layout. This separation enables offline-trained models to predict access patterns over a compact ordinal vocabulary, which are resolved at runtime through lightweight mappings. Across four data-intensive workloads, FarSight delivers up to 3.6x higher performance than the state-of-the-art.
title Learning Semantics, Not Addresses: Runtime Neural Prefetching for Far Memory
topic Machine Learning
Distributed, Parallel, and Cluster Computing
Operating Systems
url https://arxiv.org/abs/2506.00384