Argonne training program immerses attendees in extreme-scale computing
Topics included supercomputer hardware, scientific software, artificial intelligence, data analysis and performance tools
image: ATPESC is an intensive two-week training program that includes hands-on sessions using DOE supercomputers.
Credit: (Image by Argonne National Laboratory.)
The annual Argonne Training Program on Extreme-Scale Computing welcomed 75 participants for a crash course in supercomputing.
Gladys Chen, who provides advanced high performance computing (HPC) user support for the Texas Advanced Computing Center (TACC) at the University of Texas, wasn’t in her Austin office for about two weeks. Instead, she was tucked away at a suburban Chicago auditorium, learning about the latest tools and techniques for supercomputers from renowned computing experts.
Her laptop was actively running computational software as she diligently followed along with the presenters, using visualization, profiling and debugging tools to analyze jobs on the Argonne Leadership Computing Facility’s (ALCF) Polaris system. The ALCF is a U.S. Department of Energy (DOE) Office of Science user facility at DOE’s Argonne National Laboratory. Chen wanted to learn as much as she could before returning to Texas to support the diverse community of researchers using TACC supercomputing resources.
“There is something in ATPESC for everyone. It is a unique opportunity to hone your scientific skills. Attendees should come with questions because this is a training program, and it gives you that incredible opportunity to grow.” — Daniel Osei-Kuffuor, computational scientist in the Center for Applied Scientific Computing at DOE’s Lawrence Livermore National Laboratory
The 2024 Argonne Training Program on Extreme-Scale Computing (ATPESC) “was an incredibly enriching experience with top-notch training and networking opportunities,” Chen said. “As an early career professional, I am grateful for the opportunity to learn fundamental skills and concepts that would ultimately pave the foundation for a career in HPC moving forward.”
Chen was among 75 attendees at the 2024 ATPESC, the 12th annual intensive training for next-generation scientists, computer experts, data analysts and others aiming to infuse their computing research with new vibrancy. ATPESC was held July 28 through Aug. 9 at the Q Center in St. Charles, Ill., and featured over 80 expert lecturers from national laboratories, universities and tech companies. They covered a variety of HPC topics, such as programming models, artificial intelligence (AI), software productivity and sustainability, data analysis, performance tools and debuggers.
Seventy-five participants from around the world attended the 2024 training program at the Q Center in St. Charles, Illinois. (Image by Argonne National Laboratory.)
One of those HPC topics, scientific visualization, offers new technologies that take data beyond the typical bar or pie chart. New software emboldens the data with attractive and vibrant images that can explode with color and animation.
Brad Carvey, an Emmy Award-winning scientist, engineer and artist at DOE’s Sandia National Laboratories, also used the power of scientific visualization with his previous work in TV and films, including the intro to the popular movie “Men in Black.”
“We are at the cusp of another revolution … with [artificial intelligence] we can do more incredible work,” said Carvey, one of the featured speakers at ATPESC.
Many attendees hope to use their cache of information from ATPESC to develop new ways of presenting data or scientific discoveries to appeal to a wider audience or to attract new funding. Others sought training to build their skill sets and to contribute more to the scientific community.
Pariksheet Nanda, a systems biologist and postdoctoral researcher at the University of Michigan, is using his ATPESC training to help expand his career options and make more scientific contributions. “This training definitely helps to fill in the gaps,” he said.
Gregory Lemieux, who left the aerospace industry to become a research software engineer developing scientific applications at DOE’s Lawrence Berkeley National Laboratory, works on the Functionally Assembly Terrestrial Ecosystem Simulator (FATES), which is a component of the DOE Energy Exascale Earth System Model (E3SM). It models the climate on exascale machines.
During a tour of Argonne, ALCF’s Bill Allcock (right) gives ATPESC attendees an overview of the hardware that powers the Lab’s Aurora exascale supercomputer. (Image by Argonne National Laboratory.)
“It’s been great to have dedicated time to engage in hands-on work with subject matter experts here, which is difficult to do with a day-to-day workload,” said Lemieux.
Fiona Knoll, an assistant professor of computer science in Maryland, is both passionate about teaching her students and about her own research, which has included distributed matrix multiplication, machine learning and wavelet algorithms.
“This program gives me the ultimate opportunity to learn a plethora of information,” Knoll said. “I especially wanted to learn about the new tools for debugging and internode parallelization. As a teacher, it is important to know current tools.”
It is hard to find experts in the field who are knowledgeable but also have the time to teach the basics and to be mentors, she added.
Daniel Osei-Kuffuor, a computational scientist in the Center for Applied Scientific Computing at DOE’s Lawrence Livermore National Laboratory, was a postdoc when he attended the inaugural ATPESC event in 2013. This year, he lectured on the Krylov solvers and algebraic multigrid methods and participated in a panel discussion on numerical algorithms and software for extreme-scale science.
ATPESC gives attendees an opportunity to gain hands-on experience with the latest supercomputing tools. (Image by Argonne National Laboratory.)
“Coming back was quite rewarding,” said Osei-Kuffuor. “I always have been following the program and recommended it to my interns who have come to the lab and one of my postdocs attended in 2019. He liked it a lot, too. Coming back now as a presenter, it was a pleasure. Participants came from diverse backgrounds, like when I was an attendee. And getting to know them and engage with them was really rewarding.”
ATPESC is designed to provide a well-rounded view of everything from hardware to software development to HPC tools and libraries, he said.
“There is something in it for everyone,” said Osei-Kuffuor. “It is a unique opportunity to hone your scientific skills. Attendees should come with questions because this is a training program, and it gives you that incredible opportunity to grow.”
First-time presenter Patrick Avery, a staff research-and-development engineer for New York-based Kitware, focused on his company’s open-source web framework for visual analytics called Trame.
“Building interactive visualization applications to either explore data or present information from your research is invaluable,” said Avery.
Over the years, more than 800 have attended ATPESC. They have myriad reasons for blocking their schedules for two solid weeks of lectures and presentations. Ultimately, they all return home with new insights into their careers and their contributions to science. In the early years of ATPESC, most attendees were postdocs and early career researchers; very few Ph.D. students attended. But over time there has been a general trend of applicants being accepted earlier in their careers, said Raymond Loy, the ATPESC program director.
“They are meeting the prerequisites earlier due to earlier exposure to HPC in their academic programs as HPC becomes more mainstream,” said Loy, who is also a member of the ALCF performance engineering team. “The trend continues with increasing numbers of Ph.D. students. But this year we also have a larger number of early career researchers compared to recent years. I believe this is due to the penetration of computational science into fields that have not traditionally employed it.”
Attendees have a very broad range of backgrounds, so different topics are more important or interesting to different attendees, Loy said.
“Our goal is to broaden their knowledge base and fill in the gaps in their experience. There is something for everyone,” Loy said.
ATPESC is hosted by the ALCF with support from DOE’s Advanced Scientific Computing Research program.
The Argonne Leadership Computing Facility provides supercomputing capabilities to the scientific and engineering community to advance fundamental discovery and understanding in a broad range of disciplines. Supported by the U.S. Department of Energy’s (DOE’s) Office of Science, Advanced Scientific Computing Research (ASCR) program, the ALCF is one of two DOE Leadership Computing Facilities in the nation dedicated to open science.
Argonne National Laboratory seeks solutions to pressing national problems in science and technology by conducting leading-edge basic and applied research in virtually every scientific discipline. Argonne is managed by UChicago Argonne, LLC for the U.S. Department of Energy’s Office of Science.
The U.S. Department of Energy’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States and is working to address some of the most pressing challenges of our time. For more information, visit https://energy.gov/science.