USTC-SYS Reading Group

The Systems Reading Group at University of Science and Technology of China provides an opportunity for students, researchers and faculty to discuss and keep abreast of current research in systems research, and to learn by example “how to do high-quality systems research”.

The meeting is held at 7:00 pm every Wednesday online (Tencent Meeting), and should not be longer than 1 hour.

Operating System:EuroSys 2021USENIX ATC '21OSDI '21

Storage System: FAST '21

Networking System: SIGCOMM 2021NSDI '21

Database System: SIGMOD 2021VLDB 2021

Architectures/Language: ASPLOS 2021

Cloud Computing: SoCC '21

Machine Learning and Systems: MLSYS 2021

Si Wu and Cheng Li are organizers of USTC-SYS Reading Group, who advise students to select a paper and present it clearly.

Weiqiang Yu works as the assistant for the fall21 semester to help the organization work.

The primary focus of the discussion will be understanding what the paper says. With that in mind, the discussion leads should focus on clearly identifying

  • what is the problem being solved?
  • what are the challenges? what makes the problem difficult?
  • what are the key insights/techniques/lessons used to address those challenges? (i.e., how does it work)

We will also discuss other points, but we will try to keep these points as secondary concerns as much as possible

  • does it work? (i.e., evaluate the evaluation)
  • why does the problem matter? i.e., why should we (not) read the paper?
  • other points that arise

Reference:How to read a paper

Consider structuring the presentation around the key discussion points. When guiding the discussion, the goal should be to guide the rest of the group through the discussion points above (roughly in order). We strongly suggest that the number of slides should not exceed 40.

Discussion leads should propose a paper and justify why we should discuss the paper.

The current suggestion are papers covering system topics and published at top system conferences such as OSDI, SOSP, ASPLOS, SIGMOD, VLDB, ATC, EuroSys, NSDI, FAST etc in the recent 3 years. Please send your choice to wqiang7@mail.ustc.edu.cn to get approval.

Spring 2022

Assistant: Zhiwei Bai

Lecturers Conference Paper Title Paper Link Slides Video Rank Sheet Summary Report
Mon.Day.Year Name ABBR+Year(Bold) Title(Bold and Italic) slides video
Aug.25.2021 Si Wu 2021 Fall Kick-off slides video
Apr.6.2022 Chuannan Zhang SOSP'21 Boki: Stateful Serverless Computing with Shared Logs https://www.cs.utexas.edu/~zjia/boki-sosp21.pdf slides video
Apr.13.2022 Ziyue Qiu, Chuqing Gao NSDI '21 Segcache: a memory-efficient and scalable in-memory key-value cache for small objects https://www.usenix.org/conference/nsdi21/presentation/yang-juncheng slides video
Apr.20.2022 Yu-Ang Cao FAST'22 Removing Double-Logging with Passive Data Persistence in LSM-tree based Relational Databases https://www.usenix.org/system/files/fast22-huang.pdf slides video
Apr.27.2022 Youhui Bai, Quan Zhou Topic Sharing Break gpu memory wall in training large models slides video
May.4.2022 Quan Zhou SC '21 ZeRO-Infinity: Breaking the GPU Memory Wall for Extreme Scale Deep Learning https://arxiv.org/pdf/2104.07857.pdf slides video
May.11.2022 Zhiqi Lin, Haiquan Wang OSDI‘22 Alpa: Automating Inter- and Intra-Operator Parallelismfor Distributed Deep Learning https://arxiv.org/pdf/2201.12023 slides video
May. 25. 2022 Qianli Wang ATC'21 One-sided RDMA-Conscious Extendible Hashing for Disaggregated Memory https://www.usenix.org/conference/atc21/presentation/zuo slides
Jun. 1. 2022 Guanbin Xu OSDI'22 A New Approach to Deep-Learning Model Sparsity via Tensor-with-Sparsity-Attribute https://www.microsoft.com/en-us/research/uploads/prod/2021/08/SparGen.pdf slides video
Jun. 8. 2022 Yiduo Wang Topic Sharing Scaling Metadata of Distributed File System slides video
Jun. 15. 2022 Yufei Wu FAST'22 InfiniFS: An Efficient Metadata Service for Large-Scale Distributed Filesystems https://www.usenix.org/conference/fast22/presentation/lv slides video
Jun. 22. 2022 Xinyang Shao DFS slides video

Scientific Writing

presenter: Cheng Li

lecture video
1 lec1
2 lec2
3 lec3
4 lec4
5 lec5
  • public/rg/readinggroup.txt
  • 最后更改: 2年前
  • 王一多