As chief technical officer (CTO) of Livermore Computing (LC), Bronis R. de Supinski is responsible for formulating, overseeing, and implementing LLNL’s large-scale computing strategy, requiring managing multiple collaborations with the HPC industry and academia.
Early in de Supinski’s career, mentor Jim McGraw inspired him, saying, “If you want to be a world-leading researcher, LLNL is a great place to do that.” When asked how he set about achieving that, de Supinski says, “I just looked for interesting problems and solved them. I had the chance to come up with lofty theories as a grad student, but I quickly realized that I preferred to create software—and later systems—that help people solve real-world problems.” Laughing, he adds, “And I like to solve problems that seem unsolvable. I really like to prove people wrong.”
Bringing a world-class exascale system like the forthcoming El Capitan to LLNL might seem like an insurmountable task, but de Supinski has undertaken the challenge with gusto.
“Right now, it’s all El Capitan all the time. It’s going to be a transformative system, and I don’t use that word lightly,” he says. “Our current advanced technology system, Sierra, has made 3D simulations tractable, but we expect El Capitan to make them almost mundane. That’s why code teams and scientists have been using the word transformative. What could be more exciting than that?”
Not only will El Capitan be the first National Nuclear Security Administration (NNSA) machine of exascale magnitude, but it will also have some bleeding edge components: APUs—a tightly coupled CPU-GPU node element, and Rabbits—an innovative near-node storage solution among them. de Supinski’s leadership in procuring these technologies for LLNL means that El Capitan is expected to have a vastly easier to use memory system and unprecedented input-output performance.
de Supinski has also been a proponent of open source for HPC and has gathered a team that believes in its importance. As such, El Capitan will be notable for its software stack as well: the Livermore-led TOSS system software stack, Flux workload manager, and Spack package manager will be just three of many open source software packages that make El Capitan the most open-source-run exascale system, increasing its lifespan and maintainability by staff already familiar with these tools.
de Supinski earned his Ph.D. in computer science from the University of Virginia and joined LLNL’s Center for Applied Scientific Computing (CASC) in the late 1990s. He led several research projects and served as a group leader for numerous groups in CASC before moving over to Livermore Computing when he accepted the CTO role. He was named a Distinguished Member of Technical Staff in 2015, LLNL’s highest technical job classification, saying: “I am honored to be included in the ranks of LLNL's distinguished scientists. The Lab's commitment to high-quality science in general and high performance computing in particular creates an exciting work environment that has provided me with an excellent position from which to influence supercomputing."
de Supinski has also served as Livermore’s lead on the Exascale Computing Project’s (ECP’s) PathForward element as well as its precursor FastForward and Design Forward projects that, similarly to ECP were co-funded by the NNSA’s Advanced Simulation and Computing (ASC) program and the Office of Science. He previously co-led ASC’s Application Development Environment and Performance Team (ADEPT), which is responsible for the development environment, including compilers, tools, and runtime systems on LLNL’s large-scale systems. He is the chair of the OpenMP Language Committee.
de Supinski won the prestigious Gordon Bell Prize in 2005 and 2006. He also has won two R&D 100 awards, including one for leading a team that developed STAT, a novel scalable debugging tool. He was named to the 2021 IEEE class of fellows and served as general chair of the ACM/IEEE 2021 International Conference for High Performance Computing, Networking, Storage and Analysis (SC21). Most recently, he was named a 2022 ACM Fellow.
In addition to his work at the Lab, de Supinski is a Professor of Exascale Computing at Queen’s University of Belfast. He loves walking by the river and that city is one of his favorite places to visit. For fun, de Supinski plays in two weekly bridge games, one at the Lab and one with several former colleagues that included Jim McGraw, his early mentor, before he moved away from the Tri-Valley area. While the time comes to retire, de Supinski imagines spending more time in Northern Ireland. He also hopes to play more bridge and advance from his current level of Sectional Master in the American Contract Bridge League to Life Master.
“Throughout my career, I have had the pleasure not only of standing on the shoulders of giants but also of working shoulder-to-shoulder with many outstanding computer scientists and computational scientists—many of whom still work at LLNL, while some now work (or always have) at other institutions. LLNL has provided me the opportunity to pursue interesting work of importance to the nation.”