A Fond Farewell

The National Center for Supercomputing Applications (NCSA) has long been a flagship for research computing. As one of the founding members of the National Science Foundation’s (NSF) ecosystem of high-performance computing resources, NCSA has an important role in the success of research projects nationwide. The Center has been fortunate to have had strong and inspirational guidance over the years, including current Director Bill Gropp.

After eight impressive years as NCSA’s Director, Bill Gropp is announcing his retirement. In addition to his time as the Center’s director, Gropp invested nearly two decades of teaching and research at the University of Illinois (U. of I.). But many might not know his first brush with NCSA came long before that when he was a senior computer scientist at the Argonne National Laboratory.

A Researcher with a Vision

“My first real engagement with NCSA was as co-leader, with Valerie Taylor, of the ‘Performance Engineering Expedition’ as part of the PACI program,” said Gropp. “I saw firsthand the focus at NCSA on both applications and collaborations with research leaders.”

Before becoming director at NCSA, Gropp had already been stacking up accomplishments and accolades. He has long had a keen interest in research computing, beginning as a student when he studied math, physics and computer science. With a particular focus on parallel computing, his educational career led him to a role as a graduate student at the Lawrence Livermore National Laboratory (LLNL), where he had the chance to work with supercomputers.

“I’ve been interested in supercomputing ever since graduate school, where I learned about vector supercomputers during summers at LLNL,” said Gropp. “I saw that many calculations were beyond the capabilities of existing systems and I was excited about developing ways to provide enough computing power to address these challenges.”

This interest led to two of his most significant contributions to HPC. Born out of a desire to help researchers get the most out of computing, Gropp helped create the PETSc parallel numerical library and the Message Passing Interface (MPI) standard – two contributions that helped researchers and computer scientists design software and applications that could utilize the full power of a supercomputer with multiple processing cores. He went even further and led the creation of the MPICH, a free, open-source application that makes MPI portable.

“Before MPI, applications either had to be rewritten for each vendor’s parallel system, or developers had to pick a ‘portability layer’ that was a research project with uneven coverage and implementation, and that did not take the best advantage of each vendor’s hardware,” explained Gropp. “MPI as a standard provides a common programming interface. But for that interface to be useful, there needed to be a high-performance implementation on which developers can depend. What MPICH did was provide an implementation that was designed to make it easy for parallel computer vendors to optimize it for their platform. As an open source implementation, MPICH also made it easy for researchers to explore different ideas, both to improve implementations of MPI and to explore new features to propose for MPI.”

Shortly before he joined U. of I., a revolution had begun taking place in the world of parallel computing. Benchmarks on single-core processors started to ebb. The technology had been pushed to its limits and yielded less with each iteration. Manufacturers began building chips with multiple cores on each processor to improve performance. These changes would make Gropp’s work all the more relevant. While technology sped ahead, software lagged – very little of it was designed to take advantage of these multiple cores. 

Gropp’s work with parallel computing put him in the perfect position to advocate for research computing at U. of I. In a sign of things to come, he founded the Parallel Computing Institute based at U. of I. in response to the parallelism revolution. He continued teaching aspiring computer scientists and collected accolades and achievements before joining NCSA.

When the NCSA director position opened up in 2017, the Center didn’t have to look far to find an inspired choice. With his impressive background and expertise, Bill Gropp was chosen to become the Center’s fifth director.

Changing for the Future

Gropp began his tenure as NCSA director in 2017. At that time, the famous NCSA supercomputer, Blue Waters, was close to the end of its originally planned operation. NCSA was flying high off the incredible research that the fastest supercomputer at any university was helping to produce. Still, Gropp found several areas where the Center could be improved.

He set about reorganizing NCSA to focus more on the last word in it’s name – applications.

“With the rise of cloud computing, there were other options for researchers to get access to computing power,” Gropp said. “However, using that power requires expertise in many areas, and NCSA has always been focused on (super)computing applications. I wanted NCSA to both seize opportunities to run systems to support work that did not fit well on commercial cloud systems and especially to emphasize its expertise in delivering applications.”

Bill Gropp and Dan Katz at NCSA.

“When he became director,” said Daniel S. Katz, NCSA’s Chief Scientist, “Bill recognized the importance and growth of NCSA’s work in research software development and maintenance and created NCSA’s software directorate by promoting a group that was lower in the org chart. This gave software a seat at the table in NCSA decision-making and signaled to our peers, collaborators, and funders that we viewed it as one of our strengths.”

Gropp wanted to build a well-rounded research computing organization. NCSA was more than just a place with the HPC hardware – he wanted to emphasize the full complement of services offered at the Center. New directorates were created to focus on core service and research areas: software and applications, engagement, and outside partnerships that utilized NCSA’s powerful expertise for research consulting.

Bill has always had a strong vision for where research computing is headed. His deep knowledge of HPC led him to usher in the era of GPU computing at NCSA, which led to resources like Delta and DeltaAI. His vision for how NCSA  could support researchers has been invaluable at Illinois, and beyond!

–Susan Martinis, Vice Chancellor for Research and Innovation, University of Illinois Urbana-Champaign

Gropp also pushed to increase the Center’s partnerships with industry. While private corporations have the money to buy supercomputers, places like NCSA have an unparalleled depth of expertise. At that time, the Center had 30 years of experience running state-of-the-art computing facilities. New partnerships with industry would end up being mutually beneficial, as they would bring multitudes of opportunities for new types of research and applications, increasing the Center’s value to researchers with projects that aligned with private enterprise.

During his early years as NCSA’s director, Gropp was also tasked with planning a roadmap beyond Blue Waters. While NCSA experts were able to make such efficient use of the machine that it produced for four years longer than expected, eventually, the famous supercomputer would need to be replaced.

Bill gave software a seat at the table in NCSA decision making, and signaled to our peers, collaborators, and funders that we viewed it as one of our strengths.

-Daniel S. Katz, Chief Scientist, NCSA

With his goal of including more voices from the Center regarding considerable organizational efforts, Gropp brought multiple groups together to form a planning committee that eventually created the Delta proposal. As always, Gropp’s forward-thinking set up NCSA for what would come – Delta, funded by the National Science Foundation,  would be a GPU-based system, something very different from Blue Waters.

Gropp predicted the increasing need for HPC resources that were more efficient for modeling and simulation, and increasingly for artificial intelligence or machine learning research (AI/ML). GPUs offer significant research advantages when certain types of calculations come into play. Because a GPU has an abundance of cores, thousands compared to a CPU’s tens, it can do a lot of simple tasks simultaneously, making it highly efficient when working on science simulations and machine learning.

With the addition of Delta, NCSA’s resource became one of the most requested for AI research in the NSF portfolio. Delta has also been an essential part of the ecosystem of resources available in the National Artificial Intelligence Research Resource (NAIRR) Pilot. In January 2024, NCSA was announced as one of the primary NAIRR partners aiming to develop a national AI research infrastructure that will connect a wide array of researchers and educators with essential AI computing resources.

Much of this was accomplished against the backdrop of the COVID-19 pandemic. While research on NCSA’s resources doesn’t require everyone to be on site, that was not the case for many University of Illinois functions.

Forging a Stronger Partnership

Situated on the north quad at the U. of I. Urbana-Champaign campus, faculty have benefited greatly from their access to NCSA’s expertise and resources. However, during the pandemic, this relationship would truly strengthen, as all of campus shared the burden of meeting the day-to-day demands of running a top-tier university with nearly 60,000 students.

Gropp directed resources to assist with the SHIELD program. The program was designed to be a scalable way to test for COVID-19 infections on a regular basis. The test was a novel use of a saliva test, requiring individuals to simply “spit” to get results. With the program in place, those tested would have very early alerts that they were infected, allowing them to safely quarantine so that everyone else could attend classes in person without fear of being infected.

Bill’s unwavering support and guidance have significantly advanced our collaborative efforts, leaving NCSA much stronger and more connected.

–John Towns, deputy director, NCSA

Inspired by what U. of I. and NCSA accomplished together, Gropp set about increasing the connections between NCSA and campus. Around this time, programs like the Center for Digital Agriculture (CDA), the Astrophysical Sciences Program Office (ASPO) and the Health Innovation Program Office (HIPO) were spearheaded.

NCSA supported each of these new programs through staff and resources. The new programs successfully connected researchers from their respective domains to NCSA’s resources, enabling collaborations across campus that have resulted in remarkable results in cancer research, digital agriculture advancements and astronomical surveys.

“During his tenure as Director, Bill has been a steadfast advocate for fostering new collaborations,” said John Towns, NCSA’s deputy director. “His leadership has been instrumental in strengthening relationships with our sibling centers and spearheading high-level partnerships with various agencies. Bill’s unwavering support and guidance have significantly advanced our collaborative efforts, leaving NCSA much stronger and more connected.”

These collaborative efforts eventually led Gropp to advocate and win funding for an even greater partnership between the U. of I. system and NCSA – Illinois Computes. Through a combined five-year, $50 million commitment from the University of Illinois Urbana-Champaign and U. of I. System, NCSA offers considerable computing and consulting resources to researchers. The program has been wildly popular among Illinois researchers, and NCSA resources have powered many, many projects. The program benefits from the same philosophy Gropp instilled in the Center from the beginning – NCSA offers more than just compute power; Illinois researchers have access to NCSA resources, consultation and expertise all through Illinois Computes.

All these accomplishments only scratch the surface of Gropp’s impact on NCSA. Over the years, he’s been awarded with numerous recognitions from peers and organizations, highlighting his impact in the world of HPC. Among his many recent accolades, the Association for Computing Machinery (ACM) awarded Gropp and five other researchers the 2024 ACM Software System Award for their innovative work on MPICH. He’s also ushered in the deployment of NSF’s DeltaAI, launched the joint effort between U. of I. and NCSA to create the Office of Data Science Research and continues to work tirelessly to forge new partnerships with like-minded organizations throughout the Midwest region.

In the eight years Gropp has led NCSA, he has accomplished a great deal, all while preparing the Center for the future. NCSA’s portfolio of successes – including our growing AI research support, our focus on quantum computing and our work in the nationwide NSF-ecosystem of HPC resources, to name just a few – comes in part from Gropp’s strong leadership of the Center. While staff will be sad to see Gropp retire from the Center, his work with the Computing Research Association (CRA), the National Academies of Sciences, Engineering, and Medicine (NASEM) and the Computing Community Consortium (CCC) will continue to benefit NCSA and organizations like it for a while yet.

Bill Gropp’s legacy at NCSA will have a lasting impact. His expertise and careful planning will mean that even after he’s gone, NCSA will have the tools he left us with to continue on the Center’s path to bringing people, computing and data together to benefit society.

ABOUT NCSA
The National Center for Supercomputing Applications at the University of Illinois Urbana-Champaign provides supercomputing, expertise and advanced digital resources for the nation’s science enterprise. At NCSA, University of Illinois faculty, staff, students and collaborators from around the globe use innovative resources to address research challenges for the benefit of science and society. NCSA has been assisting many of the world’s industry giants for over 35 years by bringing industry, researchers and students together to solve grand challenges at rapid speed and scale.