Homecoming

Larry Smarr, NCSA’s founder, recently came home to NCSA to give a talk on his journey from founding the Center to his work with the National Research Platform (NRP). Smarr was on campus to accept an honorary Ph.D. in computer science at the University of Illinois Urbana Champaign (UIUC) during this year’s commencement. During his talk, he shared stories from the early days of NCSA. This was a time to reminisce for those still at the Center from Smarr’s era. For others, we must go back to the very beginning, before NCSA was even conceived, to fully understand Smarr’s tremendous influence in supercomputing and why those in the world of cyberinfrastructure consider Smarr to be a founding father.

When Larry Smarr joined the physics department at UIUC in 1979, there were no supercomputing centers on university campuses. That didn’t mean research using supercomputers wasn’t being performed, but for professors like Smarr, without this resource at home, he had to travel abroad to use these powerful machines to complete his work in a timely manner. It was one of these trips that spurred Smarr on to send an unsolicited proposal to the U.S. National Science Foundation (NSF) – a proposal that would eventually lead to the creation of NCSA and four other founding supercomputing centers across the country.

According to an interview with Smarr for SC18, after he visited Germany in 1981 to make use of one of their supercomputers, he was questioned by fellow astrophysicist Karl Heinz Winkler about his trip. Winkler asked why America, a country that was arguably a leader in supercomputing development, didn’t have these machines for use among the elite academic institutions. “Winkler essentially shamed the U.S. and that it was ‘mooching off of Germany’ for supercomputer access,” Smarr said at the time. “For me, this was an existential moment. I felt there was something terribly wrong with the country, and I resolved to not give up on having open academic access to supercomputing in the U.S.”

Spurned into action in 1982, Smarr went about surveying colleagues at UIUC, asking them what compute power would be beneficial and, in some cases, necessary to their work. The information he gathered was used to write a paper with the attention-grabbing title, “The Supercomputing Famine in American Universities.”

In the paper, which can still be found in the original NSF panel report for the national supercomputing program, Smarr made a compelling argument for why supercomputers were not only important to researchers but an essential component of their research. “The current distribution of supercomputers … virtually excludes American universities,” Smarr wrote. He raised the alarm that other countries, like Japan, would soon eclipse the U.S. in scientific research if the problem wasn’t remedied quickly, as those countries had made supercomputing resources available to their scientists.

“Clearly, it would be much better to have the supercomputer (or access to one) at the home institution of the scientist.” he wrote. “Then he could spend all year in ‘compute, think, read, talk, recompute’ mode. That is, he could do a calculation, produce output graphically, thoughtfully look through it, realize he needed different functions plotted, compare his work with previous efforts, discuss his ideas with other American-based colleagues, do an analytic side calculation, understand more deeply what the physics is, and then decide what to compute next. He has access to a friendly and comfortable local computing system instead of having to learn an entirely new, foreign system in a few days’ time. He has graduate students, secretaries, his own office and library, and a familiar circle of scientific colleagues to talk with. All of these factors are important for good scientific research.”

The point is that the scientist should be able to choose the tool which best enables him to do his science on his timescale. Currently, the answer to this choice is being masked by the physical difficulty of gaining access to supercomputers.

–from “The Supercomputing Famine in American Universities” by Larry Smarr (pg. 150)

Not long afterward, Smarr worked on an unsolicited proposal to the NSF called “A Center for Scientific and Engineering Supercomputing,” though it was more commonly known as the “Black Proposal” due to the color of the cover. The proposal kicked off a flurry of events, but eventually, Smarr’s work yielded results. The NSF decided to fund and support five supercomputing centers located on campuses across the country. UIUC was chosen as the site for one such center, NCSA, and Larry Smarr was appointed as its first director in 1985. The creation of academic supercomputing centers was just the beginning.

Smarr continued as NCSA’s director for 15 years. In that time, he helped spearhead a number of influential technologies and ideas. Smarr’s interest in cyberinfrastructure led him to push for machines that were the most performant for researchers. Jay Alameda, a senior technical program manager at NCSA, remembers Smarr’s focus on testing out the limitations of these massive computers.

“I arrived at NCSA in 1990,” Alameda said. “Although NCSA started with Cray vector computers, Larry was not content to rest there at all – we went through an array of systems, from the massively parallel Connection Machine CM-2 (which was really quite immature in its system maturity at the time of our deployment) and the Connection Machine CM-5. The biggest transformation, though, was the shift to microprocessor-based machines, ultimately landing in the world of systems built using commodity processors – which built a solid link from the desktop computer one would use to the largest scale supercomputer that one could use to advance their science. Our big change was symbolized by the adoption of large SGI shared memory systems. This dramatic change – which forced us to migrate applications, users and our entire experience base – helped accelerate progress as there was no fallback on the traditional Cray vector machines. Larry forced this transition, even if it wasn’t uniformly popular, to help us continue to lead the charge in high-performance computing.”

Smarr was influential in creating what would later become the Internet as we know it today. His advocacy for a high-speed network connecting the supercomputing centers eventually prompted the creation of NSFNET, a backbone network allowing high-speed data transfer between centers. According to the NSF, the NSFNET backbone became the first national 45-megabits-per-second Internet network in 1991.

Smarr was also deeply supportive and passionate about the field of scientific visualization. Donna Cox, a professor at UIUC and someone with her own list of impactful accomplishments, worked closely with Smarr all the years he was at NCSA. In fact, the first visualization of the NSFNET backbone was developed by Cox and Bob Patterson at NCSA. This visualization went on to become the icon for the burgeoning internet of the 90s.

Scientific visualization became a major innovation out of early NCSA. Smarr enabled this by investing in young faculty and industry professionals and establishing several research and Hollywood-style production visualization groups. 

“Larry Smarr has the uncanny ability to discover creative people and technological trends,” said Cox “He has been an outstanding innovator through his extraordinary ability to spot, support and promote humanistic creativity. My 36-year tenure at NCSA, helping to lead scientific visualization and develop the field, was made possible through Larry Smarr. He could foresee the future and how the nascent 90’s visualization and virtual reality technologies pioneered at NCSA heralded the future. Many of these technologies were so ahead of their time, they are only now coming to market.”

Colleen Bushell, director of the Health Innovation Program Office at NCSA, also recalled those early years learning about scientific visualization techniques. “I came to NCSA in 1986, shortly after the Center began,” Bushell said. “Larry and Donna Cox were initiating a Scientific Visualization effort and began hiring several people from backgrounds in software development, computer animation, design and data representation to help scientists understand the results from simulations run on the supercomputers. I was in the right place at the right time and launched my career in this area. This was a very unique initiative and became another important example of Larry’s dedication to pushing the envelope and supporting new directions that advanced computational research.”

Smarr’s desire to support scientific visualization advancements was there before NCSA existed. Bob Haber, a former mechanical science and engineering professor at UIUC, began working with NCSA in the early days of the Center, conducting research simulations. “I first met Larry Smarr, pre-NCSA, in the early 1980s, when I joined the VAX and Image Processing (VIP) facility that he and Bob Wilhelmson had created to explore visualization in scientific computing,” said Haber. “That theme was absorbed into the NCSA’s program at a scale perhaps only Larry might have foreseen.”

Without a doubt, my interactions and collaborations with Larry and NCSA set high-performance computing and visualization as two of the major themes in my research career.

–Bob Haber, mechanical science and engineering professor emeritus, UIUC

NCSA staff who worked with Smarr often describe him as a source of inspiration and vision. “We had so much innovation in the early 90s,” Alaina Kanfer, assistant director for strategic partnership development at NCSA, recalled. “People always talk about the innovations, but that was just part of it. All these projects were emerging out of Joe Hardin’s group – he led the software development group at the time. But it was Larry who understood their immense potential for social impact and invested significant personal energy into making sure innovations like NCSA Mosaic were recognized and adopted internationally.  He started small, setting up meetings with community leaders to share his vision and bring in support. I remember him saying during one of these presentations, ‘Imagine going online and booking your tickets to the Assembly Hall in real-time.’ That was Larry – he sees all these pieces and has this vision, but even more than that, he’s able to communicate it in a way that also helped everyone listening understand the potential. It wasn’t just the innovation itself that brought success to NCSA. It was Larry’s understanding of that innovation and how it could be used in every sector of society. ”

Smarr always wanted applications to be one of the main features of NCSA – thus the prominent location of the word Applications in the Center’s title. “Larry recognized from the beginning that applications – using computing to solve problems that couldn’t be solved in any other way – were the reason the nation needed to invest in supercomputing,” said NCSA Director Bill Gropp. “It shows in our name – the National Center for Supercomputing Applications. That focus on applications has kept NCSA at the forefront of using advanced computing in research. Larry’s vision set NCSA on the course that we are still following – use computing to solve the most challenging problems society has today.”

“Larry built the Center with a strong application focus,” said Alameda, “which had at its core a dynamic applications group comprised both of Illinois faculty and NCSA staff representing a broad range of domains. This nucleus was the place Larry would sound out ideas, listen to concerns brought to the table by the research scientists, and learn where the fields were going with their research, especially in the realm of their supercomputing needs and growth.”

Software innovations became a hallmark of NCSA work. One of these innovations was Mosaic. Smarr’s ability to create an environment that encouraged and supported innovation led to not only a reliable and intuitive web browser, but expanded to showing novel ways to use the internet.

Not enough credit goes to those who design the software we use, but Smarr knew that user design and experience were just as essential to get right as the programming. He made sure top talent was involved to achieve a look and feel that would draw people in. Bushell recalls using what she’d learned through her work in visualization to help design what browsers and websites would look like for a long time, proving that Smarr’s championing of interdisciplinary collaboration was a large part of the success of the Center’s software output. “I was able to apply visualization concepts to the design of the first Mosaic interface,” explained Bushell. “One of my early projects shortly after Mosaic was to create a virtual office space for Al Gore’s performance review team. Our goal was to show how a website could be more than an online brochure, which was the predominant type of website in the very early days. When Larry first brought the idea to me, he said it was for the vice president. I just assumed it was a vice president of one of our Industry partners… until I saw one of the emails was from albert.gore. I should have known he meant THE Vice President. Larry always had grand visions and knew how to make them happen.”

Scott Lathrop, New Frontiers Initiative and Blue Waters training and outreach coordinator, has recollections of Smarr that date back to the very beginning of NCSA. “I joined NCSA very early, after listening to Larry’s vision to advance education and research in all fields of study while he led the development of the winning proposal for the formation of NCSA.”

Like many others, Lathrop remembered Smarr as a visionary figure during the nascent days of NCSA. “Larry continually motivated and energized people to accelerate innovations and foster inter-disciplinary collaboration,” Lathrop said. “Larry’s approach to engaging academia, government and industry has become a model for advancing science and engineering research, education and development.”

Lathrop credited Smarr with more than just a vision – Smarr also had a talent for recruiting and creating a team capable of fulfilling his vision. “His leadership brought together talented individuals from all sectors,” Lathrop said. “That led to significant applications of computational science, high-speed networking, data analytics, visualization, educational resources and collaboration technologies. It was Larry’s visionary leadership that led to NCSA becoming internationally renowned.” This ability to find new talent is why NCSA still thrives today, with many of those who came to NCSA as students now at the many helms of the Center.

In 1997, Smarr became the Principal Investigator (PI) on a project called the National Computational Science Alliance. The goal of the alliance was to provide a nationwide network of supercomputing resources that enabled “advances to be achieved by a new level of national-scale multidisciplinary collaborative research.” According to the NSF award for the National Computational Science Alliance, the principal mission was “the integration of the many computational, visualization and information resources into a national-scale ‘Grid.’” If this sounds familiar, that’s because the alliance was the precursor to TeraGrid, which then became XSEDE, and is now ACCESS. These programs have helped open the door to researchers far and wide, democratizing the research process and instilling the idea of the bigger mission – to support computational and data-enabled research on a much larger scale. 

When Smarr left NCSA in 2000, he left a network of individuals he mentored and trained. John Towns, NCSA’s deputy director, was just a graduate student when he first met Smarr. 

“I recall when I first saw Larry in the spring of 1989,” said Towns. “As a young graduate student looking to find an advisor, I attended a seminar he presented in the physics department. He wove a story of advancing science through the use of computing and visualization to advance the field of gravitational physics – my chosen field of interest. Having been an avid tinkerer with early personal computers and the like, this brought together my two intellectual passions.” 

Towns told a story that exemplifies Smarr’s hyper-awareness of potential not just in ideas and hardware but in people as well. “Larry was looking to identify two to three new graduate students for his group. This led to my first meeting with him. I was incredibly young, impressionable and a bit timid. Going to that meeting, I was excited but quite nervous. Larry immediately made me feel at ease and, as I found he tended to do with everyone, he took a genuine interest in me and my background. By the fall of 1989, I was a member of his research team and, unbeknownst to me, was launching a career influenced and aided by Larry’s mentorship.”

Smarr’s ability to find the people he knew could help make his vision a reality was only part of the story, as Towns explained. “While others have commented on Larry’s ability to identify talent and pull the talented individuals together into cohesive teams, it is sometimes lost that he put significant time and effort into developing those he gathered together. As a graduate student and then staff at NCSA, I was inspired by his leadership and visions, but I also benefited greatly from his ‘taking me under his wing’ and helping me to see and then help realize his vision. I played a supporting role in a number of efforts, from leading technical support for the Mosaic browsers to playing a key role in developing and executing on the National Computational Science Alliance proposal.”

Smarr’s instinct in recruiting and mentoring Towns was spot on. Eventually, Towns took on many of the responsibilities Smarr originated and became a leader in his own right. Towns has been instrumental in the continued success of the national supercomputing networks in the form of TeraGrid, XSEDE and now ACCESS.

“Larry helped me understand how I could bring my own understanding of computational science – a term not actually used until years later – to not only advance my area of interest but how to develop resources and services that meet researcher needs and further the national research agenda. This led to my leadership in securing grants for supercomputers: playing a key role in the TeraGrid project and leading it in its last two years, and leading the XSEDE project.”

NCSA may be part of Smarr’s origin story, but his accomplishments aren’t over yet. He continues to innovate and explore the limitless possibilities of supercomputers. He’s been featured a number of times for his work using supercomputers to monitor his health and his efforts to increase cross-disciplinary collaboration. In Smarr’s talk at NCSA, he shared his recent accomplishments as well as his past achievements, and the work largely remains the same. Smarr continues to build and support the infrastructure researchers need to optimize their output through endeavors like the NRP. But for many at NCSA, he is recalled foremost as the founding mentor of the Center, and his ideas remain the guiding principles that it strives for – to bring people, computing and data together to benefit society.

You can view Larry Smarr’s full talk at NCSA here: From NCSA to the National Research Platform.

Smarr also gave a talk at the Carl R. Woese Institute for Genomic Biology while visiting the UIUC campus. You can view that presentation here: Discovering Human Gut Microbiome Dynamics.

---

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.