ISACPP: Interference-Aware Scheduling Approach for Deep Learning Training Workloads Based on Co-Location Performance Prediction
IEEE Transactions on Parallel and Distributed Systems · 2025-08 · 36 (8) · 1591-1607
DOI: 10.1109/tpds.2025.3577796 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Parallel and Distributed Systems, 2025-08. Verfasst von Zijie Liu und 5 weiteren Autor:innen. Die Arbeit wurde bisher 8-mal zitiert (Crossref) und nennt 71 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Nanjing University, Startup Foundation for Introducing Talent of Nanjing University.
Abstract (Originalsprache)
Traditional exclusive cloud resource allocation for deep learning training (DLT) workloads is unsuitable for advanced GPU infrastructure, leading to resource under-utilization. Fortunately, DLT workload co-location provides a promising way to improve resource utilization. However, existing workload co-location methods fail to accurately quantify interference among DLT workloads, resulting in performance degradation. To address this problem, this article proposes an interference-aware scheduling approach for DLT workloads based on co-location performance prediction, dubbed ‘ISACPP’. ISACPP first builds an edge-fusion gated graph attention network (E-GGAT) that incorporates DL model structures, underlying GPU types, and hyper-parameter settings to predict co-location performance. Since the co-location state changes as each workload is completed, ISACPP proposes a multi-stage co-location interference quantification model derived from the predicted co-location performance to identify the GPU device with the minimum overall interference. Experimental results demonstrate that ISACPP can accurately estimate the co-location performance of DLT workloads with a maximum prediction error of 8.72%, 1.9%, and 4.4% for execution time, GPU memory consumption, and GPU utilization, respectively. Meanwhile, ISACPP can significantly shorten workload makespan by up to 34.9% compared to state-of-the-art interference-aware scheduling methods.
Förderung
- National Natural Science Foundation of China
- Natural Science Foundation of Nanjing University
- Startup Foundation for Introducing Talent of Nanjing University
- Natural Science Foundation of Jiangsu Higher Education Institution of China
- China Telecom Corporation Limited Jiangsu Branch Intelligent Cloud-Network Operation Center Lingxiao
- Postgraduate Research
- Practice Innovation Program of Jiangsu Province
Zitieren
APA
Liu, Z., Cheng, Y., Chen, C., Hu, J., Fu, R., & Zhang, D. (2025). ISACPP: Interference-Aware Scheduling Approach for Deep Learning Training Workloads Based on Co-Location Performance Prediction. IEEE Transactions on Parallel and Distributed Systems, 36(8), 1591-1607. https://doi.org/10.1109/tpds.2025.3577796
Vancouver
Liu Z, Cheng Y, Chen C, Hu J, Fu R, Zhang D. ISACPP: Interference-Aware Scheduling Approach for Deep Learning Training Workloads Based on Co-Location Performance Prediction. IEEE Transactions on Parallel and Distributed Systems. 2025;36(8):1591-1607. doi:10.1109/tpds.2025.3577796
BibTeX
@article{Liu2025,
author = {Liu, Zijie and Cheng, Yi and Chen, Can and Hu, Jun and Fu, Rongguo and Zhang, Dengyin},
title = {ISACPP: Interference-Aware Scheduling Approach for Deep Learning Training Workloads Based on Co-Location Performance Prediction},
journal = {IEEE Transactions on Parallel and Distributed Systems},
year = {2025},
volume = {36},
number = {8},
pages = {1591--1607},
doi = {10.1109/tpds.2025.3577796}
}
EUVIMED