Work

HybridPersist: A Compiler Support for User-Friendly and Efficient PM Programming

Year: 2025

Type: article

Abstract: Persistent memory (PM), with its data persistence, has found widespread applications. However, programmers have to manually annotate PM operations in programming to achieve crash consistency, which is labor-intensive and error-prone. In this paper, to alleviate the burden of programming PM applications, we develop HybridPersist, a compiler support for user-friendly and efficient PM programming. On the one hand, HybridPersist automatically achieves crash consistency, minimally intruding on programmers with negligible annotations. On the other hand, it enhances both performance and correctness of PM programs through a series of dedicated analysis passes. The evaluations on well-known benchmarks validate that HybridPersist offers superior programming productivity and runtime performance compared to the state-of-the-art.

Source: Proceedings of the ACM on Programming Languages

DOI: 10.1145/3763089

Authors: Yiyu Zhang, Yongzhi Wang, Yanfeng Gao, Xuandong Li, Zhiqiang Zuo

Institution: Nanjing University

Language: English

Topic: Logic, programming, and type systems

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: diamond

Funders: National Natural Science Foundation of China, Government of Jiangsu Province, Natural Science Foundation of Jiangsu Province

Awards: G2126020117, G8268568699