After an intense four-day flurry of activity (and a cold snap that brought some actual snow flurries), the SC16 show floor closed yesterday (Thursday) and the always-extensive technical program wound down today. This year, 11,000-plus visitors, participants and volunteers from 65 countries came to Salt Lake City for the largest HPC conference in the world to take part in the four-day expo and a six-day technical program.
In an on-site media briefing held Monday, SC16 Chair John West said that this year’s SC provided record space (450,000 square feet) to a record 349 research and industry booths. Attendance however was down from last year, when nearly 13,000, descended into Austin, Texas, a show high.
One of the main events of the show each year is SciNet, which is the largest scientific network in the world for one week. This year SciNet delivered over terabits-per-second of bandwidth (and pushed a record 1.2 terabytes of traffic over the show floor), enough to send about 450 million snaps on Snapshot in 60 seconds, said West, about 850x more snaps than the rest of the world does in that time frame. That’s a lot of selfies, West joked.
West also highlighted the student program. This year there were over 270 students from around the world. Two-thirds of those are attending the conference for the first time and 25 percent are female. “This is where you come to meet the people to shape our industry, and we are looking for future volunteers,” said West.
The conference is non-profit and all volunteer with about 650 volunteers coming together to make it happen every year. Planning for each show begins about three years in advance.
The focus of this year’s conference is on workforce and diversity. Says West, “there aren’t good workforce figures for HPC because it’s just a slice of computing, but the Bureau of Labor Statistics for the United States says that every year, we’re about 200,000 workers short in computing and related fields and if we keep on this trajectory that will be a million workers short by 2022. That’s a big gap, and while HPC is only a fraction of that gap, we are part of it. If you run centers or you work in centers, you know that recruiting HPC is already challenging and it’s going to get worse, so where are we going to find our new talent?
“Two-thirds of the US workforce are women in minorities and less than 20 percent of those folks work in computing, so there’s a big untapped pool that we need to figure out how to engage meaningfully and that’s what SC is working on this year.”
For a deeper look at what SC is doing to support greater inclusivity, read Kim McMahon’s interview with John West and Trish Damkroger, chair of the Diverse HPC Workforce committee. Next year it will continue as the Inclusivity Committee. University of Edinburgh researcher Toni Collis, founder and director Women in HPC, has been tapped to head up the committee. HPCwire interviewed Toni on camera at SC and will make this video available in the coming weeks – this one is not to be missed.
The conference also recognized the 14 recipients of the ACM SIGHPC and Intel Computational and Data Science Fellowship awards. You can learn more about the candidates and the impetus for the program here.
West also shared the results of studies that cite as reasons for women not choosing a computing field the perception that it doesn’t have a direct impact on the things they care about. In response, SC launched a pilot project that gets students hands on using supercomputing to help understand social change. West says that the project is still an experiment at this point, but they are exploring a path to it becoming part of the program in the future.
The SC committee also works with the National Science Foundation and the Department of Energy to support Women in IT Networking at SC (the WINS program) – the goal of this program, now in the second year, is to jump start the careers of women who are engineers by providing them with an intensive hands-on experience in building SciNet.
HPC matters is a recurring thread at SC and one of the most-trending Twitter hashtags coming out of the show. The Monday opening panel has become a way to showcase HPC efforts that are having the biggest impact for societal good. This year’s panel, HPC Impacts on Precision Medicine: Life’s Future–The Next Frontier in Healthcare, described how precision medicine is being used to fight disease, enhance health and lifestyle, extend life, and contribute to basic science along the way.
As always the show had an impressive and packed technical program that made many attendees wish they had parallel bandwidth. Post-Moore’s law technologies featured prominently in the program with panels such as “Post Moore’s Era Supercomputing in 20 Years” and “The End of Von Neumann? What the Future Looks Like for HPC Application Developers” drawing a standing-room only crowd. Popular invited talks included “Memory Bandwidth and System Balance in HPC Systems” and “Beyond Exascale: Emerging Devices and Architectures for Computing” (which we will be covering in a future piece).
Major awards at the conference include the TOP500, which is covered here, with the number one prize going for the second time to the Sunway Taihu-Light system, installed at the National Supercomputing Center in Wuxi. A 12-member Chinese team has also won the 2016 ACM Gordon Bell prize. The research project, “10M-Core Scalable Fully-Implicit Solver for Nonhydrostatic Atmospheric Dynamics,” presents a method for calculating atmospheric dynamics. It’s the first time a Chinese team has won the award and also the first win for a Python-based application.
A Chinese team also took home top honors in the Student Cluster Competition. The University of Science and Technology of China pulled off an SC first, winning both the first place prize for highest overall score and the highest Linpack run, a record 31.5 teraflops. Just ten years ago their system would have been 36th in the TOP500.
Speaking of how much has changed over the last decade, Intersect 360 CEO Addison Snell gives an insightful look back on where HPC has been and where it’s headed in this retrospective piece.
Benchmark winners:
Top500: China’s Sunway Taihu-Light system (93 petaflops)
Green500: Nvidia SaturnV, built with 125 DGX-1 machines (8.17 gigaflops/watt)
HPCG: Japan’s RIKEN K computer
Graph500: Japan’s RIKEN K computer
HPGMG: Argonne Leadership Computing Facility Mira
The HPCwire editorial team would like to congratulate everyone on their achievements this year. And we applaud our HPCwire Readers’ and Editors’ choice award winners, a diverse and exceptional group of organizations and people who are making a difference and advancing the technologies that undergird scientific and technical progress around the world. (The photo gallery of award presentations can be viewed on Twitter.)
We look forward to seeing many of you in June at the International Supercomputing Conference in Frankfurt, Germany, and then in Denver for SC17, November 12-17. The SC17 website is already live and the chairing baton has been handed to Bernd Mohr, member of the Division “Application Support” at the Jülich Supercomputing Centre (JSC) in Germany.
Stay tuned in the coming weeks as we release our SC16 Video Interview series, featuring Women in HPC’s Toni Collis; Dr. Paul Messina, director of the Exascale Computing Project; Addison Snell, CEO of Intersect360 Research, among others.