吴明瑜

副教授

mingyuwu AT sjtu DOT edu DOT cn

个人简介

吴明瑜,上海交通大学副教授、博士生导师。主要研究兴趣为操作系统和语言运行时系统。成果发表在ASPLOS、USENIX ATC、EuroSys、IEEE TC等国际高水平会议和期刊,获EuroSys 2025最佳论文奖、APSys 2018最佳论文runner-up奖、华为火花奖等,部分成果已经被OpenJDK接收并合入主线。(英文主页

个人经历

  • 2020.11 - 至今,上海交通大学软件学院
  • 2015.09 - 2020.08,上海交通大学软件学院,博士生,导师:陈海波
  • 2011.09 - 2015.06,上海交通大学软件学院,本科生

教学工作

  • 软件学院 编译原理与技术(SE3355)2021/2022/2023/2024年秋季学期
  • 软件学院 云操作系统设计与实践 (SE3356) 2022/2023/2024年春季学期
  • 软件学院 系统设计与开发大型课程设计(系统软件)2022/2023/2024年秋季学期
  • 软件学院 计算机系统设计和实现(研) (EI6708) 2023/2024年春季学期

研究项目

  • Java的NVM原生支持: 非易失性内存(NVM)模糊了内存与持久性存储之间的界限,而Java缺乏对于NVM的高效支持,无法充分利用其大容量、低时延、持久化的特征。本项目为Java提供了NVM原生支持,研究结果包括:
    • Java语言及运行时扩展(Espresso)
    • 延迟持久化模型探索(GCPersist)
    • NVM带宽友好的垃圾回收算法(NVM-friendly-GC)
    • 轻量级对象置放策略(GCMove)
  • 垃圾回收优化: 垃圾回收是语言运行时的重要组成部分,其行为与应用性能息息相关。我们针对交互式服务、大数据分析、数据库等不同类型应用的内存行为提出了针对性的垃圾回收优化算法,其中部分成果已被OpenJDK接收并进入主线:
    • 面向交互式服务的高吞吐低时延垃圾回收器(Platinum/Jade)
    • 高可扩展全堆垃圾回收算法(ScissorGC)
    • 面向SGX的高效垃圾回收(SAGC)
  • 面向服务器无感知计算/函数即服务的语言虚拟机支持: 服务器无感知计算(Serverless Computing)是近年来发展迅速的一种云计算模式,人们也常称它为函数即服务(FaaS),该领域通常使用以Java为代表的高级语言进行开发,但语言虚拟机对FaaS的支持仍存在很多不足。为此我们对语言虚拟机进行了一系列针对FaaS场景的优化,包括:
    • 半FaaS执行模式以及自动FaaS卸载框架(BeeHive)
    • 基于语言虚拟机的轻量级执行抽象(Flock)
    • 语言虚拟机感知的FaaS内存管理(Desiccant)

发表论文

  • [EuroSys] Jade: A High-throughput Concurrent Copying Garbage Collector. Mingyu Wu, Liang Mao, Yude Lin, Yifeng Jin, Zhe Li, Hongtao Lv, Jiawei Tang, Xiaowei Lu, Hao Tang, Denghui Dong, Haibo Chen, Binyu Zang. Nineteenth European Conference on Computer Systems, Athens, Greece, April 22-25, 2024.

  • [EuroSys] Characterization and Reclamation of Frozen Garbage in Managed FaaS Workloads. Ziming Zhao, Mingyu Wu, Haibo Chen, Binyu Zang. Nineteenth European Conference on Computer Systems, Athens, Greece, April 22-25, 2024.

  • [EuroSys] Serialization/Deserialization-free State Transfer in Serverless Workflows with RDMA-based Remote Memory Map. Fangming Lu, Xingda Wei, Zhuobin Huang, Rong Chen, Mingyu Wu, Haibo Chen. Nineteenth European Conference on Computer Systems, Athens, Greece, April 22-25, 2024 (Best paper award).

  • [IEEE TC] Toward An SGX-friendly Java Runtime. Mingyu Wu, Zhe Li, Haibo Chen, Binyu Zang, Shaojun Wang, Lei Yu, Sanhong Li, Haitao Song. IEEE Transactions on Computers, pp. 44-57, January 2024.

  • [IEEE TC] Flock: Towards Multitasking Virtual Machines for Function-as-a-Service. Ziming Zhao, Mingyu Wu, Xujie Cao, Haibo Chen, Binyu Zang. IEEE Transactions on Computers, pp. 3153-3166, November 2023.

  • [ASPLOS] BeeHive: Sub-second Elasticity for Web Services with Semi-FaaS Execution. Ziming Zhao, Mingyu Wu, Jiawei Tang, Binyu Zang, Zhaoguo Wang, and Haibo Chen. The 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Vancouver, Canada, March, 2023.

  • [USENIX ATC] Zero-Change Object Transmission for Distributed Big Data Analytics. Mingyu Wu, Shuaiwei Wang, Haibo Chen, Binyu Zang. USENIX Annual Technical Conference, Carlsbad, CA, USA, July 2022.

  • [VEE] Transparent and Lightweight Object Placement for Managed Workloads atop Hybrid Memories. Zhe Li, Mingyu Wu. The 18th ACM SIGPLAN/SIGOPS International Conference on Virtual Execution Environments, Lausanne, Switzerland, March 2022.

  • [EuroSys] Bridging the Performance Gap for Copy-based Garbage Collectors atop Non-Volatile Memory. Yanfei Yang, Mingyu Wu, Haibo Chen, Binyu Zang. Sixteenth European Conference on Computer Systems, Online Event, United Kingdom, April 26-28, 2021.

  • [USENIX ATC] Platinum: A CPU-Efficient Concurrent Garbage Collector for Tail-Reduction of Interactive Services. Mingyu Wu, Ziming Zhao, Yanfei Yang, Haoyu Li, Haibo Chen, Binyu Zang, Haibing Guan, Sanhong Li, Chuansheng Lu, Tongbao Zhang. USENIX Annual Technical Conference, Boston, MA, USA, July 2020.

  • [VEE] GCPersist: An Efficient GC-assisted Lazy Persistency Framework for Resilient Java Applications on NVM. Mingyu Wu, Haibo Chen, Hao Zhu, Binyu Zang, Haibing Guan, The 16th ACM SIGPLAN/SIGOPS International Conference on Virtual Execution Environments, Lausanne, Switzerland, March 2020.

  • [VEE] ScissorGC: Scalable and Efficient Compaction for Java Full Garbage Collection. Haoyu Li, Mingyu Wu, Binyu Zang, Haibo Chen, The 15th ACM SIGPLAN/SIGOPS International Conference on Virtual Execution Environments, Providence, Rhode Island, April 2019.

  • [APSys] Analysis and Optimization of Java Full Garbage Collection. Haoyu Li, Mingyu Wu and Haibo Chen. The 9th ACM SIGOPS Asia-Pacific Workshop on Systems, Jeju Island, South Korea, August 2018 (Best paper runner-up).

  • [SOSE] DwarfGC: A Space-Efficient and Crash-Consistent Garbage Collector in NVM for Cloud Computing. Heting Li and Mingyu Wu, IEEE Symposium on Service-Oriented System Engineering, Bamberg, Germany, March 2018.

  • [ASPLOS] Espresso: Brewing Java For More Non-Volatility with Non-volatile Memory. Mingyu Wu, Ziming Zhao, Haoyu Li, Heting Li, Haibo Chen, Binyu Zang and Haibing Guan. The 23rd ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Williamsburg, VA, USA, March 2018.

  • [PPoPP] POSTER: Recovering Performance for Vector-based Machine Learning on Managed Runtime. Mingyu Wu, Haibing Guan, Binyu Zang, Haibo Chen. Proceedings of the 22nd ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming. ACM, 2017.