SMBQ : An Efficient Stealing Multi‐Bucket Queue Scheduling Framework for Multicore Systems

Ensuring rapid responses to high‐priority tasks, balancing workloads across threads, and efficiently utilizing computing resources are essential to improving the performance of resource‐constrained multicore platforms. However, existing priority scheduling methods face two major challenges. (1) Global ordering strategies can reduce redundant computation, but frequent cross‐thread synchronization creates contention and significantly increases scheduling latency. (2) Local ordering strategies reduce communication overhead by using thread‐local queues, but their lack of global priority coordination can lead to priority drift and delayed execution of high‐priority tasks. To address these challenges, we propose the Stealing Multi‐Bucket Queue (SMBQ), an efficient priority scheduling framework for multicore systems. The framework combines a multiqueue strategy with priority‐aware probabilistic work stealing to mitigate priority drift while improving scheduling efficiency, using task batching to reduce locking operations and improve cache locality. Experimental results show that the proposed scheduler significantly outperforms state‐of‐the‐art priority schedulers such as OBIM and PMOD, providing an efficient and stable solution for priority task scheduling in multicore systems.

Authors

Institutions

Publication Details

Journal
Concurrency and Computation Practice and Experience
Published
2026-11-18
DOI
https://doi.org/10.1002/cpe.71021
Primary Topic
Parallel Computing and Optimization Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

SMBQ : An Efficient Stealing Multi‐Bucket Queue Scheduling Framework for Multicore Systems

Haofei Meng, Zhenhua Zhang, Wenwu Yu, Chuhao Chen et al.
Concurrency and Computation Practice and Experience
Parallel Computing and Optimization Techniques
article

SMBQ : An Efficient Stealing Multi‐Bucket Queue Scheduling Framework for Multicore Systems

Haofei Meng, Zhenhua Zhang, Wenwu Yu, Chuhao Chen, Penglin Hou
article en

Abstract

Ensuring rapid responses to high‐priority tasks, balancing workloads across threads, and efficiently utilizing computing resources are essential to improving the performance of resource‐constrained multicore platforms. However, existing priority scheduling methods face two major challenges. (1) Global ordering strategies can reduce redundant computation, but frequent cross‐thread synchronization creates contention and significantly increases scheduling latency. (2) Local ordering strategies reduce communication overhead by using thread‐local queues, but their lack of global priority coordination can lead to priority drift and delayed execution of high‐priority tasks. To address these challenges, we propose the Stealing Multi‐Bucket Queue (SMBQ), an efficient priority scheduling framework for multicore systems. The framework combines a multiqueue strategy with priority‐aware probabilistic work stealing to mitigate priority drift while improving scheduling efficiency, using task batching to reduce locking operations and improve cache locality. Experimental results show that the proposed scheduler significantly outperforms state‐of‐the‐art priority schedulers such as OBIM and PMOD, providing an efficient and stable solution for priority task scheduling in multicore systems.

Concurrency and Computation Practice and ExperienceVol. 38(23)
Harbin Engineering University (CN), Southeast University (CN)
Parallel Computing and Optimization Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon