SC20 Is Everywhere We Are

SC20 Virtual Platform

Guided Interest Groups (GIGs) are community learning experiences where lead student volunteers shepherd interested students through the SC technical program.

  • GIGs are open to all students attending the conference; sign-ups will open first for students participating in the Students@SC Cohorts.
  • Students can register for a GIG which is lead by a student volunteer.
  • Participating students are expected to attend sessions planned by the GIG leaders as well as pre- and post-session discussions facilitated by the session lead.
  • Once students are registered and confirmed in the GIG, the lead for the area will ensure they receive invitations to all meetings.

Please take the time to review the different GIGs and submit your selection.

You must register for a group by November 8.

Register for a GIG

GIG 1

HPC at a Glance: Nodel Level Performance Engineering and Distributed Algorithms

vijay

Lead

Vijay Thakkar

Lead Bio

Vijay is a second year masters in computer science student at Georgia Tech working on accelerated computing and parallel algorithms under Dr. Rich Vuduc. He first fell in love with HPC during his junior year when he started the student cluster team during his time at Boston University. He continues to mentor a cluster competition team at Georgia Tech, and is the graduate mentor for Team Phoenix at SC20. Currently, Vijay is working in collaboration with ORNL on a novel algorithm for large scale knowledge graph mining on the Summit supercomputer. Apart from his personal research and SCC involvement, Vijay enjoys volunteering at conferences, serving as a student volunteer at SC19, HotChips 32 and as a lead student volunteer at SC20.

Description

This GIG will be a general overview of the core topics in HPC, intended to give participants an at a glance view of many different facets of the core HPC algorithms and software engineering practices. Performance engineering for HPC codes can broadly be attributed to two main types of optimizations: 1) Distributed algorithms that dictate the communication patterns, studied with mathematical rigor and often lead to asymptotic speedups in applications. 2) Code optimizations done at the scale of a single machine (node level), often paired with a performance model specific to a hardware platform such as GPUs. This goal is to give participants a window into these two words, how they intersect, and what the state of the art is. Additionally, discussions will center around how design of current and future supercomputing systems is conducted in the context of broader trends in semiconductor and HPC algorithms space.

Schedule (Eastern Time)

Nov 16 | Kickoff Meeting

  • 2:30 pm: Accelerated Computing, Distributed Algorithms and Softaware/Hardware Codesign – Setting the Stage

Nov 17 | Invited Talk: Data-Centric Architecture of a Weather and Climate Accelerator

  • 10:15 am: Pre-Session Meeting
  • 10:45 am: Invited Talk
  • 11:30 am: Post-Session Meeting

Nov 17 | Paper: GPU-Accelerated Applications

Nov 18 | Paper: Exascale and Beyond

Nov 18 | Paper: GPU Algorithms and Optimizations (Optional)

Nov 19 | Society Awards: Gordon Bell Prize Finalist Session 2

GIG 2

The Wonderful World of Machine Learning

aroura

Lead

Aroua Gharbi

Lead Bio

Aroua Gharbi is a PhD student in Computational Science and Engineering/Aerospace Engineering at Georgia Institute of Technology. Her focus is on leveraging computational capabilities in the systems engineering process of complex systems. She works under the supervision of Professor Dimitri Mavris in the Digital Engineering division of the Aerospace Systems Design Lab (ASDL). Aroua loves camping and is hoping to visit all of Georgia’s state parks before finishing her PhD.

Description

This GIG aims to generate interest and spark curiosity in machine learning (ML) and its vast field of applications. Rather than focusing on a particular type of learning (supervised, deep, unsupervised, etc.), students will be introduced to the various problems that can be solved with ML. The GIG will include a paper, an undergrad poster, a visit of an exhibitor (e.g Google/Tensorflow), a Scientific Visualization and Data Analytics poster, and potentially a Birds of a Feather (BoF) session. The discussions will focus on the generic elements of a machine learning process: data and model. The posters will give a feel of the importance of data exploration and analysis while the paper will rather expose them to some techniques in model building.

Schedule (Eastern Time)

Nov 9 | Kickoff Meeting

  • 2:30 pm: Introduction

Nov 17 | Invited Talk: The Future of AI and HPC – Neuromorphic Computing and Neural Accelerators

  • 9:30 am: Pre-Session Meeting
  • 10 am: Invited Talk
  • 10:45 am: Post-Session Meeting

Nov 17 | Paper: RLScheduler – An Automated HPC Batch Job Scheduler Using REinforcement Learning

Nov 18 | BoF: MLPerf-A Benchmark for Machine Learning

  • 2 pm: Pre-Session Meeting
  • 2:30 pm: BoF Presentation
  • 3:45 pm: Post-Session Meeting

Nov 19 | Poster: A Computer Vision and AI Based Solution to Determine the Change in Water Level in Stream 

GIG 3

Storage and I/O for HPC

jean

Lead

Jean Luca Bez

Lead Bio

Jean Luca is a Ph.D. student in Computer Science at the Federal University of Rio Grande do Sul (UFRGS), Brazil. His main research areas are High-Performance I/O, Parallel I/O, and Parallel File Systems. His research focuses on automatic tuning and reconfiguring the I/O forwarding layer of HPC platforms. He works under the supervision of Prof. Dr. Philippe Navaux (UFRGS) and Prof. Dr. Toni Cortes from Polytechnic University of Catalonia (UPC) and Barcelona Supercomputing Center (BSC), Spain. In his spare time, Jean likes to collaborate in educational projects such as the URI Online Judge platform.

Description

This GIG will explore data storage and its growing importance in High Performance Computing (HPC). Increasingly heterogeneous workloads are entering HPC installations, from the traditionally compute-bound scientific simulations to machine learning applications and Input/Output (I/O) bound big data workflows. While HPC clusters typically rely on shared storage infrastructure, the increasing I/O demands of applications stress this infrastructure, making I/O operations a performance bottleneck. This has the potential to critically impact performance on the next generation of supercomputers. To introduce this area of HPC, the GIG will include paper and poster sessions, an exhibitor’s visit, and a Birds of a Feather (BoF) session where participants will get an overview of the area. The discussions will focus on existing storage systems and technologies. Additionally, we will explore breakthrough research and trends for storage in future supercomputing systems.

Schedule (Eastern Time)

Nov 12 | Kickoff Meeting

  • 9 am: Introduction

Nov 12 | Workshop: Toward On-Demand I/O Forwarding in HPC Platforms (Fifth International Parallel Data Systems Workshop)

Nov 17 | Paper: Improving All-to-Many Personalized Communication in Two-Phase I/O

Nov 17 | Research Posters: 1) State of I/O in HPC 2020 & 2) Understanding I/O Behavior of Scientific Deep Learning Applications in HPC Systems 

Nov 18 | BoF: The IO-500 and the Virtual Institute of I/O

  • 3:15 pm: Pre-Session Meeting
  • 4 pm: BoF Presentation
  • 5:15 pm: Post-Session Meeting

Nov 19 | Panel: Exotic Storage and Data Technology: 2006 to 2020 and Beyond (Optional Session)

GIG 4

High Performance Computing in Medical Sciences and Health Research

sadura

Lead

Sadura Akinrinwa

Lead Bio

Sadura Akinrinwa is a PhD student at the Department of Computer Science, Federal University of Technology, Akure, Nigeria. She is passionate about solving Medical Images Classification Problems using HPCs. She is supervised by Professor Olatunbosun Olabode. Sadura believes that with pursuing a PhD in computing she demonstrates her ability to perform original research for a progressive career in computing. She enjoys participating in activities that present opportunities to share and improve knowledge and gain experience on computational technologies. Her status as a female in PhD studies in computer science not only allows her to improve knowledge in computing, but also enables her to mentor other students in computing.

Description

This GIG seeks to explore the application of High Performance Computing (HPC) to medical sciences and health research.

The role of supercomputers in improving health technologies cannot be overemphasized. HPC is used to devise solutions and advance the bearing of medical research on its capabilities to analyze large datasets in record time. This often leads to accelerated discoveries of life changing breakthrough results and treatments. By bringing HPC and medical sciences together, scientists can explore the use of improved data processing and visualization to create strategies and therapies for better health.

This GIG will lead students in exploring the application of HPC to health through attending of the Computational Approaches for Cancer Workshop, a paper presentation, a poster session, and a visit to the exhibitors; to enable participants understand the application of existing technologies to health researches.

Schedule (Eastern Time)

Nov 12 | Kickoff Meeting

  • 2:30 pm: High Performance Computing in Medical Sciences and Health Research

Nov 13 | Workshop: CAFCW20: Sixth Computational Approaches for Cancer Workshop (Keynote and Panel: HPC, Cancer, and COVID-19)

Nov 13 | Workshop: CAFCW20: Sixth Computational Approaches for Cancer Workshop (Poster)

Nov 13 | Lightning Talk: Urgent HPC

Nov 17 | SciViz Showcase: Visualization of Flow of Circulating Tumor Cells and Blood Cell Suspensions in Microfluidics

  • 10 am: Pre-Session Meeting
  • 10:10 am: Posters Display
  • 10:40 pm: Post-Session Meeting

Nov 17 | Exhibitor Booth Session: Supermicro: Advanced High-Performance Computing Systems to Find a Cure for COVID-19 – The Corona Supercomputer and Mammoth Project

  • 2:10 pm: Pre-Session Meeting
  • 2:30 pm: Booth Session
  • 2:45 pm: Post-Session Meeting
Back To Top Button