Publications

2025

1. HPSC

   DRL-MOSHRS: A Deep Reinforcement Learning Approach for Multi-Objective Scheduling in Heterogeneous HPC Systems

   Yifei Yang, Linchang Xiao , Tianyufei Zhou , and 3 more authors

   In 2025 IEEE 11th International Conference on High Performance and Smart Computing (HPSC) , 2025

   Bib

   @inproceedings{11038772,
     author = {Yang, Yifei and Xiao, Linchang and Zhou, Tianyufei and Yang, Chengrun and Liu, Xuezheng and Hu, Miao},
     booktitle = {2025 IEEE 11th International Conference on High Performance and Smart Computing (HPSC)},
     title = {DRL-MOSHRS: A Deep Reinforcement Learning Approach for Multi-Objective Scheduling in Heterogeneous HPC Systems},
     year = {2025},
     pages = {25--30},
     doi = {10.1109/HPSC66065.2025.00009},
     keywords = {Costs, Machine learning algorithms, Scheduling algorithms, High performance computing, Buildings, Clustering algorithms, Deep reinforcement learning, Scheduling, Heterogeneous networks, Optimization, deep reinforcement learning, high-performance computing, heterogeneous computing, resource scheduling, multi-objective optimization, elastic scaling},
   }

2. HPSC

   TaskFlare: A Heterogeneity-aware Unified Scheduling Framework for Diverse Domain-Specific Applications in Supercomputing Environments

   Tianyufei Zhou , Yifei Yang, Chengrun Yang , and 3 more authors

   In 2025 IEEE 11th International Conference on High Performance and Smart Computing (HPSC) , 2025

   Best Paper Bib

   Received the Best Paper Award at HPSC 2025.

   @inproceedings{11038795,
     author = {Zhou, Tianyufei and Yang, Yifei and Yang, Chengrun and Xiao, Linchang and Liu, Xuezheng and Hu, Miao},
     booktitle = {2025 IEEE 11th International Conference on High Performance and Smart Computing (HPSC)},
     title = {TaskFlare: A Heterogeneity-aware Unified Scheduling Framework for Diverse Domain-Specific Applications in Supercomputing Environments},
     year = {2025},
     pages = {66--71},
     doi = {10.1109/HPSC66065.2025.00016},
     keywords = {Technological innovation, Processor scheduling, Scientific computing, Parallel programming, Scalability, Dynamic scheduling, Resource management, Artificial intelligence, Usability, Optimization, Task Scheduling, Resource Management, Heterogeneous Computing, Scheduling Frameworks, Performance Optimization},
   }

3. PDCAT

   NAAM: Enhancing Automatic Task Mapping Efficiency on NUMA Machines

   Yifei Yang, Tianyufei Zhou , Linchang Xiao , and 4 more authors

   In Parallel and Distributed Computing, Applications and Technologies , 2025

   Abs Bib

   Modern supercomputing nodes are typically highly heterogeneous and exhibit Non-Uniform Memory Access (NUMA) characteristics. Automatically mapping tasks on such NUMA architectures while maintaining high operational efficiency poses significant challenges. In this paper, we propose a novel automatic mapping algorithm called NAAM (NUMA-Aware Auto Mapping), which accounts for both intra-node and inter-node heterogeneity. NAAM dynamically adjusts mapping strategies for various tasks across heterogeneous NUMA systems by performing a dynamic search to identify optimal mappings for typical heterogeneous hardware configurations. Finally, we conduct extensive experiments to validate the effectiveness of our method, with results showing that NAAM outperforms baseline approaches by up to 8\% in terms of speedup over the default mapping.

   @inproceedings{10.1007/978-981-96-4207-6_40,
     author = {Yang, Yifei and Zhou, Tianyufei and Xiao, Linchang and Yang, Chengrun and Liu, Xuezheng and Hu, Miao and Wu, Di},
     editor = {Li, Yupeng and Zhang, Yong and Xu, Jianliang},
     title = {NAAM: Enhancing Automatic Task Mapping Efficiency on NUMA Machines},
     booktitle = {Parallel and Distributed Computing, Applications and Technologies},
     year = {2025},
     publisher = {Springer Nature Singapore},
     address = {Singapore},
     pages = {441--453},
     isbn = {978-981-96-4207-6},
     doi = {10.1007/978-981-96-4207-6_40},
   }