April 1, 2021 | Written by: Katia Moskvitch
Categorized: AI | COVID-19 | Science
Share this post:
We did it once. We can — and should — do it again.
That’s the main message of an online meeting on 23 March 2021, marking the 1stanniversary of the COVID-19 High Performance Computing Consortium — a global initiative launched to accelerate the work of coronavirus researchers. With industry leaders, prominent academics and members of the US government attending, the session zoomed in on the Consortium’s results of the past year — and expanded them towards new, much wider, frontiers.
Founded in March 2020 just as the pandemic’s wave was starting to wash over the world, the Consortium has brought together 43 members with supercomputing resources. Private and public enterprises, academia, government and technology companies, many of whom are typically rivals. “It is simply unprecedented,” said Dario Gil, Senior Vice President and Director of IBM Research, one of the founding organizations. “The outcomes we’ve achieved, the lessons we’ve learned, and the next steps we have to pursue are all the result of the collective efforts of these Consortium’s community.”
And what a community it is — unique and extraordinary.
It’s the Consortium’s members providing more than 600 petaflops of compute capacity and resources such as additional software and services for scientists worldwide. It’s all the people who enable the Consortium to run so efficiently, including a panel of experts who have so far reviewed 190 research proposals, selecting and matching 98 of them with the available and appropriate high performance computing resources. And it’s of course all the scientists from six continents who have been using the resources for crucial research to help us understand the deadly virus.
A rapid start and great results
Creatng the Consortium, “the collective response of the government, industry and academic partners was rapid, collaborative and pathbreaking – truly a model for the power of science, targeted to rapid response,” said Maria Zuber, Professor of Geophysics and Vice President of Research at MIT, recently appointed co-chair of President’s Council of Advisors on Science and Technology (PCAST) and a former Executive Board member of the Consortium.
The initiative got underway within days, unusually quickly considering the number of high-profile founding partners involved. “Dario called me on a Sunday morning just as I was making brunch,” recalled Zuber. “He told me that I needed to drop everything and get MIT signed on within hours, because of White House press release announcing the consortium was going out the next morning.”
MIT was not the only organization to receive that call, and “we got it done, as did our other founding partners,” added Zuber. “I’m not sure that I’ll ever get an institutional commitment approved that quickly again.”
The many researchers using the Consortium’s computing resources have been focusing on three main areas: the basic science to understand the virus, the exploration of therapeutics and treatments, and ways to enable better patient outcomes. The results shed light on many different facets of the pandemic.
For instance, we now understand how the virus interacts with human receptor proteins, its mutations, and their impact on the creation of vaccines and therapeutics. Researchers using supercomputers via the Consortium have found new molecules and proteins able to fight the coronavirus. Other teams have obtained high resolution simulations of aerosol transport in indoor environments.
Speaking to the event’s participants, Jim Brase, Deputy Associate Director for Data Science at Lawrence Livermore National Laboratory, highlighted a number of important results obtained thanks to the Consortium. One team, led by Jeremy Smith, the Director of the University of Tennessee and Oak Ridge National Laboratory (ORNL) Center for Molecular Biophysics at ORNL, has worked with Google and IBM’s Summit supercomputer at Oak Ridge to screen very large numbers of compounds against a number of viral proteins. The researchers “identified multiple approved drug compounds that can inhibit infectivity and a number of those are now in clinical trials,” Brase said.
And Amanda Randles at Duke University used Microsoft’s facilities to evaluate the potential for splitting ventilators across patients early on in the pandemic, when there was a huge concern about shortage of ventilators. “This was done very quickly and allowed medical providers to be prepared to actually share ventilators, and the fast reaction that she could get through the HPC Consortium enabled the team to get to the FDA authorization very quickly,” said Brase.
Meanwhile, John Towns, Director of Collaborative eScience Programs at the National Center for Supercomputing Applications (NCSA) at the University of Illinois, mentioned that “the epidemiology prediction of spread was often informing the decisions made by leaders in local governments.” Many results obtained thanks to the Consortium were “used as supporting evidence for decision makers, and I think it’s one of our biggest impacts on our management of the pandemic.”
Still in full swing, the Consortium was started during the previous US administration. During her speech, Zuber singled out several key government officials who helped make the Consortium a reality.
She gave a shout out to Kelvin Droegemeier, a research meteorologist at the University of Oklahoma and former Director of The White House Office of Science and Technology Policy (OSTP), and Michael Kratsios, former US Chief Technology Officer at OSTP. She also mentioned former Assistant Director for Quantum Information Science Jake Taylor, former Deputy Chief of Staff Chris Liddell, and Paul Dabbar, former Under Secretary for Science for U.S. Department of Energy, for helping to launch the partnership astonishingly quickly and avoid the typical red tape.
And Zuber praised Gil, Barbara Helland, the Associate Director of the Office of Science’s Advanced Scientific Computing Research (ASCR) program, the rest of the Consortium’s science and computing committees, all the partners, reviewers and researchers, for helping to make the Consortium a success.
“The spirit of teamwork and collaboration among the 43 members was absolutely amazing,” said Mike Rosenfield, Vice President for Data Centric Solutions at IBM Research, who has also played an instrumental leadership role in the creation and operation of the Consortium. “The entire time we just got things done. It has been very impressive.”
Leaping into the future
While the COVID-19 insight obtained because of the Consortium was a crucial part of the anniversary meeting, the participants also discussed the future. “We should use the lessons we’ve learned so far thanks to the Consortium, the knowledge and the experience gained this year to address future global crises,” said Gil. “We should create a broader international organization exactly for that.”
Eventually, Gil thinks, there should be a global body that would deal with various ‘known unknowns’ – large scale emergencies that we could anticipate and prepare for ahead of time. He calls such an organization ‘Science Readiness Reserves’ (SRR), akin to a National Guard ready to spring into action at any point. That’s still in the planning stage, but an intermediate instantiation based on the successes of the Consortium that could eventually morph into the SRR is already underway.
Called the National Strategic Computing Reserve, this public-private coalition of worldwide experts and resource providers spanning government, academia, industry and nonprofits is currently undergoing review in the US Congress. The concept is to enable the world to mobilize computing assets in a global emergency to accelerate the pace of discovery of new materials and therapeutics.
During the panel at the second half of the anniversary event, moderated by Barb Helland, the speakers were in agreement. The Consortium is indeed proof that such private-public partnership approach works – and works well. “It’s clear that the way the HPC Consortium have stood up for the COVID crisis is a wonderful role model,” said Kelvin Droegemeier, one of the panelists. “A crisis showed us that we could come together, we could overcome some of the barriers and challenges and red tape that tended to stand in our way, and we could do the extraordinary.”
A meteorologist, Droegemeier spoke specifically about applying such a model to tackle weather and climate challenges. They are, he said, “among some of the canonical problems which the National Strategic Computing Reserve and the National Science Reserve could be applied to. I think these and other disciplines are better poised now to take advantage of an environment in which we can stand up a resource, and in a day’s time begin using it in a very, very efficient way.”
Another panelist, Pat Falcone of the LLNL, outlined several past big emergencies where a Computing Reserve could have been of use. She mentioned the Deepwater Horizon disaster that resulted in the Gulf of Mexico being covered in 200,000 tons of oil, the devastating Tohuku earthquake of 2011 followed by a deadly tsunami that killed thousands and damaged the Fukushima Daiichi nuclear reactor, and the 2014 Ebola outbreak in West Africa, among other calamities.
“Looking forward to a National Strategic Computing Reserve and working issues to make it successful, we certainly are going to need to be thinking about establishing early connections to the domains and communities and to first responders to really think about the fact that data, software and people are at least as important as the computing resources themselves,” Falcone said.
So what else could be done better next time?
Panelist Geralyn Miller of Microsoft, one of the founding partners, mentioned the importance of the diversity of proposals. “I think this is something we struggled with and that was a little too narrow,” she said. “There is this element of active portfolio management that would be really important going forward, meaning making sure that we are getting a diversity in the types of proposals, and that we always have a lens to what is happening around us so that we can bring in new facets to the portfolio if necessary.”
Speaking last, panelist Manish Parashar of the National Science Foundation neatly summed up the future approach to success: we shouldn’t address a global emergency completely ad hoc again. Longer term planning is crucial – to prepare for the next pandemic, hurricanes, tornadoes, wildfires, or nuclear disasters.
Also, this time “as we shifted resources to conduct a pandemic-related research, we had to delay other science and engineering projects, putting on hold advances in broader research ecosystem,” said Parashar. With the Computing Reserve and the SRR, he added, we need to have structures, processes and resources ready in advance, to mobilize them rapidly and minimize undesirable side effects. “The next time we face such an emergency, we must be prepared.”
Because, as Zuber put it, “when the world needs us, the science and technology community will be there to help. It is one of the many ways that we can address President Biden’s stated intention to rebuild trust inside America to Build Back Better.”
This article originally appeared on the HPC Consortium blog: A year on: From HPC Consortium’s success to National Strategic Computing Reserve
Inventing What’s Next.
Stay up to date with the latest announcements, research, and events from IBM Research through our newsletter.