February 18, 2014: People of ACM: William D. Gropp
Tuesday, February 18, 2014
William D. Gropp holds the Thomas M. Siebel Chair in Computer Science at the University of Illinois at Urbana-Champaign. He is the Deputy Director for the Research Institute for Advanced Computing Applications and Technologies, and the Director of the Parallel Computing Institute, both at the University of Illinois.
Gropp helped to create the Message Passing Interface, also known as MPI, and the Portable, Extensible Toolkit for Scientific Computation, also known as PETSc, one of the leading packages for scientific computing on highly parallel computers. He was awarded the IEEE-CS Sidney Fernbach Award in 2008, "for outstanding contributions to the development of domain decomposition algorithms, scalable tools for the parallel numerical solution of Partial Differential Equations, and the dominant HPC communications interface."
He has shared R&D 100 awards for both PETSc and MPICH, a leading implementation of MPI. In 2010, he was elected to the National Academy of Engineering, "for contributions to numerical software in the area of linear algebra and high-performance parallel and distributed computation." He is a fellow of ACM, IEEE-CS, and SIAM.
Gropp received a B.S. degree in Mathematics from Case Western Reserve University, an M.S. degree in Physics from the University of Washington, and a Ph.D. degree in Computer Science from Stanford. He has coauthored several books, including Parallel Multilevel Methods for Elliptic Partial Differential Equations with Barry Smith and Petter Bjorstad and Using MPI, with Ewing Lusk and Anthony Skjellum.
As Chief Applications Architect and co-Principal Investigator of the Blue Waters sustained petascale computing project, what is your view of its impact on addressing national priorities in light of the burgeoning interest in Exascale research?
Blue Waters is first and foremost an instrument for advancing science through computations that require the most powerful computer available. The National Science Foundation has selected nearly 30 projects that cover a wide range of research, from understanding the standard model of physics, the behavior of biomolecules or earthquakes, to the formation of the universe. Early results have included the chemical structure of the HIV's capsid, the shell that protects the virus. In addition, Blue Waters is a balanced system — offering outstanding performance for data as well as numerical computing.
Like all supercomputing systems, many high-quality proposals for use of the system can't be granted because the need greatly outstrips the supply. In this context, Exascale research is important for two reasons: first, the technology needed to deploy an Exascale system will greatly increase access to petascale systems, allowing scientists and engineers to routinely tackle problems that only a few can do now. Second, one thing we've learned is that there are always even harder problems that will require more computing — for example, we're already looking at one at our new Department of Energy-supported Center for the Exascale Simulation of Plasma-Coupled Combustion. Without Exascale computing, the solutions to important problems will remain beyond our reach.
Is the HPC community doing enough to convince the nation's technology leaders that sustained investment in Exascale research will bring tangible returns to science, industry, society and the economy?
No. I define HPC (high performance computing) as any computational approach where achieving high performance is necessary. Viewed this way, HPC is everywhere—used to design consumer products and advance medical science as well as an essential tool for basic science and national security. And if we focus only on what I'll call supercomputing, the use of the most powerful machines to solve the most difficult problem, we have even more to do. We need to explain why some problems can't be solved on mere clusters or even large cloud systems, and we need to explain why both HPC and supercomputing are essential for important problems in big data and real time sensors.
What country currently has a valid claim to having the world's fastest supercomputer, and is this claim still a meaningful measure of achievement?
How do you define fast? Computers are tools that we use to solve problems, so I define fast in terms of the problems that I want to solve. Different problems need different capabilities in a computer, and thus different computers may be considered "fastest." For Blue Waters, we defined a "sustained petascale performance" metric that is based on measuring the performance on a representative subset of our applications. For this set, we believe we have the world's fastest machine. Others will have different problems that they consider the most important. What's critical is not who has the fastest computer by some arbitrary measure but which countries are investing in building systems capable of solving the hardest problems. China, Japan, the European Union, and the US all have strong programs to build and use supercomputers more capable than any that exist today. Many others are investing in HPC to support their scientific research and commercial development.
As General Chair of SC13, what advice would you give to young people considering careers in high performance computing?
HPC is exciting! HPC is about solving problems and understanding the world around us. It has been and remains one of the core elements of computing — and it is intertwined with all other areas of computing, including hardware architecture, programming languages, algorithms, networks, data and I/O, and applications. HPC as a field is healthy and growing — we had over 10,000 attendees at SC13, representing vendors, researchers, and users. New this year, SC13 highlighted some of the uses of HPC in our HPC Impact Showcase, from designing faster Formula One racers to better pharmaceuticals; just browsing the list of presentations gives you a hint at the breadth of impact of HPC. For those who want to learn more, SC offers a number of programs to introduce students to HPC. One example is the SC student volunteer program, which gives students a good way to sample the field of HPC while working with some of the leaders of the field. If you want to be part of changing the world for the better, consider HPC. And come to SC14 in New Orleans!