舰船电子工程2025,Vol.45Issue(2):119-123,5.DOI:10.3969/j.issn.1672-9730.2025.02.025
GPU云资源预约式智能调度技术研究
Research on Intelligent Scheduling of GPU Cloud Resources with Reservation-based System
摘要
Abstract
A GPU cloud resource reservation based intelligent scheduling technology(DACO)is proposed to address the im-balance between supply and demand of GPU cloud resources and the limitations of real-time scheduling strategies.This technology consists of two parts,which are a prediction duration module and a scheduling module.The proposed technology is tested by Analog verification,and the results show that the proposed technology can effectively improve resource utilization and task response time un-der the condition of ensuring the completion rate before the deadline.关键词
GPU调度/蚁群算法/机器学习Key words
GPU scheduling/ant colony algorithm/machine learning分类
信息技术与安全科学引用本文复制引用
张云鹏,臧亚楠,田秋雨..GPU云资源预约式智能调度技术研究[J].舰船电子工程,2025,45(2):119-123,5.基金项目
江苏省重大科研设施预研筹建项目(编号:BM2021800)资助. (编号:BM2021800)