Skip to main navigation Skip to search Skip to main content

ROME: Maximizing GPU Efficiency for All-Pairs Shortest Path via Taming Fine-Grained Irregularities

  • Weile Luo
  • , Yuhan Chen
  • , Xiangrui Yu
  • , Qiang Wang
  • , Ruibo Fan
  • , Hongyuan Liu
  • , Xiaowen Chu
  • The Hong Kong University of Science and Technology (Guangzhou)
  • Harbin Institute of Technology
  • Hong Kong University of Science and Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

All-Pairs Shortest Path (APSP), a fundamental problem in graph analytics, can be solved efficiently by reducing the computational workload through vertex reordering. However, it fails on GPUs due to fine-grained granularity, shape, and dependency irregularities, which cause severe hardware underutilization. We introduce ROME, a system that tames these irregularities by spatially restructuring computation into regularized workloads and temporally overlapping them with an asynchronous pipeline. ROME achieves 14.7-244.5× speedup over the state-of-the-art multicore CPU solution and 11.2-338.0× speedup over the state-of-the-art GPU solution. Notably, our results achieve mostly above 20% and up to 34.7% of peak min-plus OPs across all tested graphs.

Original languageEnglish
Title of host publicationProceedings of the 31st ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming, PPoPP 2026
EditorsTony Hosking, Madan Musuvathi, Kenjiro Taura
Pages204-217
Number of pages14
ISBN (Electronic)9798400723100
DOIs
StatePublished - 28 Jan 2026
Event31st Annual ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, PPoPP 2026 - Sydney, Australia
Duration: 31 Jan 20264 Feb 2026

Publication series

NameProceedings of the 31st ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming, PPoPP 2026

Conference

Conference31st Annual ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, PPoPP 2026
Country/TerritoryAustralia
CitySydney
Period31/01/264/02/26

Keywords

  • All-Pairs Shortest Path
  • GPU Computing
  • Irregular Workloads

Fingerprint

Dive into the research topics of 'ROME: Maximizing GPU Efficiency for All-Pairs Shortest Path via Taming Fine-Grained Irregularities'. Together they form a unique fingerprint.

Cite this