Guided Interest Groups (GIGs) are community learning experiences where students shepherd participants through the SC Technical Program.
GIGs are open to all students attending the conference. Priority is given to students participating in the Students@SC cohorts.
Student Networking & Mentoring ChairMichael O. Lam
OCT 5, 2023
Sign-ups Open
NOV 3, 2023
Sign-ups Close
NOV 12, 5:30 pm
Kickoff & Social Event
Participating students are expected to attend sessions chosen by the GIG leaders as well as short pre- and post-session discussions facilitated by the leader. There will also be a GIG kickoff meeting and social event at 5:30 pm on Sunday, November 12.
Once a student has registered and received a registration confirmation for a specific GIG, they will be put in contact with the GIG leader. The leaders will send all meeting invites to the GIG participants.
Nine GIGs are available on a wide variety of HPC topics such as performance analysis and optimization, machine learning, parallel architectures, graph applications, sustainable computing, and software tools.
Have you ever wondered about the advantages of high performance computing (HPC) and whether it aligns with your goals? Why HPC? What is HPC? What can we do with HPC? This GIG will guide you through the world of high performance computing, and how hardware, software, and applications are integrated to get the most out of it.
This GIG will provide an overview of one of the most powerful tools in the scientific world. Multiple paradigms, tools, programming models, and languages are used to generate results and advance science in diverse fields. Activities and discussions will center around understanding how people can implement HPC in their pursuits. One of the main objectives of this GIG is to provide an idea of how HPC can help you advance in your career or push the boundaries of HPC as a scientific discipline.
Diego Andres Roa Perdomo
Diego Andres Roa Perdomo is a Ph.D. student in the Computer Architecture and Parallel Systems Laboratory at the University of Delaware, advised by Dr. Xiaoming Li. His research focuses on the development of Program Execution Models (PXM) for parallel computation based on Dataflow theory. He is implementing a distributed runtime for parallel computation, namely DECARD, based on the Codelet Model.
When Diego is not doing research/reading/programming, he likes to play video-games with his friends, cook, or watch a movie/series. He also enjoys swimming, playing squash, biking, traveling, and 3D-Printing.
Sunday, Nov 12
GIGs Kickoff5:30–7:30 pm MST, Room 201
Tuesday, Nov 14
Panel – Runtimes and Workflow Systems for Extreme Heterogeneity: Challenges and Opportunities10:30 am–2 pm MST, Room 205-207
Wednesday, Nov 15
Paper – Exascale Computing: Experiences Readying Applications for Exascale2–2:30 pm MST, Room 403-404
Thursday, Nov 16
Panel – The Impact of Exascale and the Exascale Computing Project on Industry3:30–5 pm MST, Room 201-203
As scientific computing tackles ever larger problems at ever finer resolutions, it has grown ever more vital that scientific applications are able to fully leverage available high performance computing (HPC) resources. This need has fueled the discovery of innovative techniques for designing, scaling, and optimizing scientific applications.
Over the course of the conference, we will explore the state of the art applications pushing the world’s largest supercomputers to their limits, with a focus on this year’s contenders for the esteemed Gordon Bell Prize. This prize is awarded to a team which develops an application which achieves exceptional performance, scalability, or time-to-solution on an important engineering or scientific problem from a wide variety of domains. In the process, we will discover how these applications work, and the computational techniques which allow them to efficiently solve these complicated problems.
Joy Kitson
Joy Kitson is a PhD student at the University of Maryland, where she is advised by Ahbinav Bhatele. She is also a Department of Energy Computational Science Graduate Fellow, and interned at Oak Ridge National Laboratory over Summer 2022 through that program. She was a virtual intern at Argonne National Laboratory over the Summer of 2020 and graduated from the University of Delaware that Spring with a Bachelors of Science in Computer Science and Applied Mathematics. She worked on the Caliper project at LLNL Summer 2019, and co-presented work done by her team at LANL over Summer 2018 on Effective Performance Portability during SC18. Her current work revolves around optimizing HPC applications, with a current focus on computational epidemiology and computational fluid dynamics.
When not doing research, she loves swing dancing, reading, and playing a variety of games with friends – including D&D, board games, and video games.
ACM Gordon Bell Finalists Presentations 110:30 am–12 pm MST, Room 501-502
ACM Gordon Bell Finalists Presentations 23:30–5 pm MST, Room 501-502
Paper – Molecular Dynamics Applications and Accelerators3:30–4:00 pm MST, Room 301-302-303
This GIG is for anyone who is interested in learning about the latest developments in exascale computing applications and computer architectures at HPC scale. The future in computing covers many different applications from diverse domains, and there are many frontiers delivering performance at scale to new and old problems. Serving many different interests and purposes requires specialized computer architectures that present opportunities and challenges for research and industry alike.
The sessions in this GIG are selected to introduce students to the breadth of opportunities in architecture and applications research for current and future HPC systems. The variety of topics will inspire students to specialize their studies in ways that speak to them individually and connect them to the broader community of HPC research.
Thomas Randall
Thomas Randall is a Graduate Research Assistant studying under Dr. Rong Ge at Clemson University in the School of Computing. His research focuses on performance optimization and autotuning for High Performance Computing (HPC) workloads and applications, especially with GPUs! He is in the fifth year of his PhD program, working towards his dissertation at Clemson. He has also published research with scientists at Argonne National Laboratory and regularly volunteers at the Supercomputing conference series.
Outside of work, Thomas enjoys card and board games with friends as well as expanding his ratings list of 600+ movies.
Monday, Nov 13 (Optional)
Invited Talk – The Legacy of ECP Software Efforts, Realized and to Come9:15–10 am
Paper – GPU-based LU Factorization and Solve on Batches of Matrices with Band Structure11:10–11:30 AM
Paper – Extreme-Scale Applications10:30 am-12 pm MST, Room 403-404
Paper – Exascale Computing1:30–3 pm MST, Room 403-404
Paper – Algorithms on GPUs1:30–3pm MST, Room 301-302-303
Among the computer science disciplines, HPC can feel very detached from the humans which use these systems. Many academic conferences and contributions hedge towards the technical, focusing on scalability, algorithms, performance optimization and computer hardware. Sometimes HPC as a discipline can feel like a train pushing forward progress for progress’ sake.
With this very machine heavy focus, it’s very easy to forget the goal of building HPC centers, the source of the data being used, and who is using these systems. So, in the spirit of re-injecting the human element into HPC, this GIG will take you through a few paper sessions and birds of a feather which balances out the technical discussions with considerations of data sources, accessibility and ethics of how we may use HPC to better serve humanity.
Connor Scully-Allison
Connor Scully-Allison is a PhD student researching Data Visualization for HPC applications at the University of Utah. He has significant experience in data management, data wrangling, UI design and Human Computer Interaction (HCI) methodologies. Overall, Connor is highly interested in focusing the human in our computer-based disciplines and understanding how we can make all of our systems better for end users.
Paper – Sustainable Computing10:30 am–12 pm MST, Room 405-406-407
BoF – With Great Power Comes Great Responsibility: Ethics in HPC12:15–1:15 pm MST, Room 301-302-303
Paper Session 2 – Data Coordination3:30–5 pm MST, Room 301-302-303
BoF – A National Science Data Fabric to Democratize Data Access and Reusability12:15–1:15 pm MST, Room 601-603
As HPC clusters grow in computing power there is an ever-increasing demand for energy. This not only costs money but uses valuable resources. This GIG will serve as an exploration of recent work in addressing the need for more sustainability in the HPC world. We will attend paper sessions, a panel discussion and a BoF event.
Dewi Yokelson
Dewi Yokelson is a PhD candidate in Computer Science at University of Oregon, located in beautiful Eugene, Oregon. Her research is in performance analysis, monitoring, and visualization of high performance computing applications. In her free time she likes to be active outdoors, including hiking, running, skiing, and cycling. Her love for nature has inspired an interest in learning about what can be done to make high performance computing a sustainable endeavor for years to come.
Paper – Sustainable Computing10:30 am–12 pm MST, Room 405-407
Panel – Carbon-Neutrality, Sustainability, and HPC3:30–5 pm MST, Room 205-207
BoF – The Green500: Trends in Energy-Efficient Supercomputing5:15–6:45 pm MST, Room 501-502
This GIG has been crafted with the aim of giving you a comprehensive understanding of the wide-reaching realm of ML (machine learning) applications and scientific computing. High performance computing (HPC) stands as a pivotal force in pushing the boundaries of scientific inquiry. Applications, in turn, strive to harness HPC’s formidable capabilities by employing inventive scientific computing techniques to accelerate data generation and analysis.
Within this GIG, you’ll discover a thoughtfully curated selection of conference events and talks. These have been handpicked to provide students with valuable insights into diverse applications and the computational methods that drive them forward. You’ll also have the unique opportunity to engage in plenary sessions and discussions featuring researchers from various disciplines, all of which will enrich your comprehension of the world of applications and scientific computing.
Shivangi Gupta
Shivangi Gupta is a computer science PhD student at the University of Alabama in Huntsville, where she is advised by Dr. Vineetha Menon. Her research focuses on studying drug-target interactions, using cutting-edge artificial intelligence (AI) and machine learning (ML) techniques. Over the Summer, she interned with the Predictive Biology group at the Lawrence Livermore National Laboratory in California. Her work involved utilizing deep learning models to annotate functions in protein sequences, with the goal of enhancing the protein design process through machine learning techniques. Her research interests include big data analytics, data science, computational biology, and deep learning.
When not immersed in her research, she enjoys hiking, weightlifting, cooking, and watching k-dramas.
Monday, Nov 13
Workshop – Workshop on Artificial Intelligence and Machine Learning for Scientific Applications (AI4S)9 am–12 pm MST (first half of the workshop), Room 501-502
BoF – Mixed Feelings About Mixed Precisions12:15–1:15 pm MST, Room 702
BoF – Machine Learning from the Data’s Perspective: Data-centric AI, AI readiness and AI Reproducibility5:15–6:45 pm MST, Room 501-502
Panel – Scalable and Adaptable Architectures for AI/HPC Advancement1:30–3 pm MST, Room 201-203
As we continue our relentless pursuit of the frontiers of molecular dynamics (MD) with a seamless integration of machine learning (ML) and high performance computing (HPC), we’re set to make the most of the upcoming events. The events have been meticulously chosen to align with our group’s focus, enabling us to gather insights, spark innovations, and reinforce our knowledge.
Harshita Shani
Harshita Sahni is a PhD student at University of New Mexico advised by Dr. Trilce Estrada working on applying machine learning techniques to molecular dynamics simulations. She is working with Los Alamos National Laboratory with Dr. Sandrasegaram Gnanakaran on projects that involve machine learning and analysis of MD simulations, back-mapping of course grained to all-atom simulations. She loves singing and playing piano. She is also involved in sports like swimming and badminton.
Paper – TANGO: Re-Thinking Quantization for Graph Neural Network Training on GPUs12:15–1:15 pm MST, Room 301-303
Paper – NNQS-Transformer: An Efficient and Scalable Neural Network Quantum States Approach for Ab Initio Quantum Chemistry11:30 am–2 pm MST, Room 301-303
BoF – MLPerf: A Benchmark for Machine Learning5:15–6:45 pm MST, Room 601-603
Paper – MST Rapid Simulations of Atmospheric Data Assimilation of Hourly-Scale Phenomena with Modern Neural Networks1:30–2 pm MST, Room 405-407
Are you interested in learning about what performance means in “high performance computing”? Are you interested in state-of-the-art research in this area? This GIG will provide an overview of different notions of performance, as it relates to power, modeling, and benchmarking via various sessions at SC23. These sessions will give nuanced insight into the various aspects of performance and optimization. They will cover major areas of interest, such as power-aware computing, hardware co-design, and future benchmarking strategies.
Daniel Berry
Daniel Berry is a Data Science and Engineering PhD student at the University of Tennessee, Knoxville (UTK), advised by Dr. Jack Dongarra. He does research with the Innovative Computing Lab’s Performance Tools group. Daniel received his Bachelors of Science in Computer Engineering at the University of Tennessee. His research interests include application performance monitoring tools, benchmarking methodologies, optimizing applications, numerical methods for data science, and large-scale data analytics. Daniel has been involved with the HPC community ever since competing with UTK in the 2013 Student Cluster Competition.
Daniel is an avid fan of learning, video games, cooking, and baking.
Paper – DPS: Adaptive Power Management for Overprovisioned Systems2:30–3 pm MST, Room 401-402
Paper – Application Performance Modeling via Tensor Completion4–4:30 pm MST, Room 401-402
BoF – The Future of Benchmarks in Supercomputing5:15–6:45 pm MST, Room 702
Paper – Co-Design Hardware and Algorithm for Vector Search2:30–3 pm MST, Room 401-402
There is no HPC without the word ‘performance’ in it. In this GIG, we visit one of the most crucial topics any HPC application faces, from data movement to machine learning applications. People in this field constantly strive to improve performance and make it as portable as possible.
Radita Liem
Radita Liem is a doctoral candidate and research associate at the RWTH Aachen University, Germany. Her interest is in performance analysis and engineering, especially for large-scale parallel filesystems and I/O. She was involved in various performance analysis and engineering projects such as POP (Performance Optimization and Productivity) Center of Excellence from the European Union Horizon 2020 project, VI-HPS (Virtual Institute – High Productivity Supercomputer), and PERMAVOST (HPDC workshop on performance analysis and engineering).
She enjoys playing games, cooking, and baking in her spare time.
Paper – Prodigy: Toward Unsupervised Anomaly Detection in Production HPC Systems2–2:30 pm MST, Room 401-402
BoF – Navigating Complexity: Achieving Performance Portability in the Evolving Landscape of Heterogeneous HPC Systems12:15–1:15 pm MST, Room 301-303
Paper – Data Flow Lifecycles for Optimizing Workflow Coordination3:30–4 pm MST, Room 301-303
Paper – I/O in WRF: A Case Study in Modern Parallel I/O Techniques3:30–4 pm MST, Room 401-402
Review the GIGs above and submit your selection by November 3.
Contact us if you have questions about student GIGs. We’d be happy to help.
