Efficient Data Race Detection of Async-Finish Programs Using Vector Clocks

TL;DR

FastRacer is a new data race detection technique that efficiently uses vector clocks to reduce overheads in async-finish programs with locks, improving performance over previous methods.

Contribution

It introduces FastRacer, a novel approach that optimizes vector clock management for dynamic data race detection in structured parallel programs.

Findings

Significantly reduces runtime overhead compared to FastTrack.

Uses less memory by optimizing vector clock sizes.

Competitive with state-of-the-art race detectors for async-finish programs.

Abstract

Existing data race detectors for task-based programs incur significant run time and space overheads. The overheads arise because of frequent lookups in fine-grained tree data structures to check whether two accesses can happen in parallel. This work shows how to efficiently apply vector clocks for dynamic data race detection of async-finish programs with locks. Our proposed technique, FastRacer, builds on the FastTrack algorithm with per-task and per-variable optimizations to reduce the size of vector clocks. FastRacer exploits the structured parallelism of async-finish programs to use a coarse-grained encoding of the dynamic task inheritance relations to limit the metadata in the presence of many concurrent readers. Our evaluation shows that FastRacer substantially improves time and space overheads over FastTrack, and is competitive with the state-of-the-art data race detectors for async-finish programs with locks.