Congratulations to Robert D. Falgout, the 2022 recipient of the SIAM Activity Group on Supercomputing Career Prize, and Piyush Sao, Xiaoye S. Li, and Richard Vuduc, the 2022 recipients of the SIAM Activity Group on Supercomputing Best Paper Prize. Both prizes will be presented at the SIAM Conference on Parallel Processing for Scientific Computing (PP22).
Robert D. Falgout
Robert D. Falgout is the 2022 recipient of the SIAM Activity Group on Supercomputing Career Prize. The award will be presented at the SIAM Conference on Parallel Processing for Scientific Computing (PP22), to be held virtually February 23 – 26, 2022. Falgout will give a talk at the conference titled “Scaling in Space and Time” on Friday, February 25, 2022 at 9:25 a.m. PT.
The prize was awarded to Falgout for his contributions to theory, algorithms, and open-source software for efficient parallel algebraic multigrid solvers and multigrid schemes in space and time.
The SIAM Activity Group on Supercomputing awards this prize every two years to one outstanding senior researcher who has made broad and distinguished contributions to the field of algorithms research and development for parallel scientific and engineering computing.
Robert D. Falgout is a computational mathematician in the Center for Applied Scientific Computing (CASC) at Lawrence Livermore National Laboratory (LLNL). He is the project leader for the hypre scalable linear solvers project and the XBraid parallel time integration project. Falgout earned his Ph.D. in Applied Mathematics at the University of Virginia in 1991 under James Ortega and joined LLNL as a postdoc that same year. He worked initially on concurrent BLAS algorithms and later served as the technical leader for the ParFlow groundwater flow project. Falgout was named LLNL Distinguished Member of Technical Staff in 2018 and 2021 SIAM Fellow.
Q: Why are you excited to receive the SIAG/SC Career Prize?
A: My entire career has been devoted to developing algorithms for large-scale parallel computing, so it’s awesome to be recognized by the community for that work. When I saw the names of the five previous awardees, all I could think was, “Wow, this is special!”
Q: Could you tell us a bit about the research that won you the prize?
A: Multigrid methods solve systems of equations with a computational complexity that scales linearly with problem size, hence they have good potential for scaling with the next generation of parallel computers and high-fidelity simulations. But there is no single algorithm that works for all applications, and in some cases effective methods don’t yet exist. My research focuses on the development of theory, algorithms, and open-source software for these methods.
Q: What does your work mean to the public?
A: The solution of large sparse linear systems is an essential ingredient in many scientific simulation codes and multigrid methods are among the fastest techniques for solving them. Multigrid can also be applied to the time dimension (or simultaneously in space-time) to provide significant speedups in time-dependent simulations, where traditional serial time stepping is becoming a bottleneck on today’s parallel computers.
Q: What does being a member of SIAM mean to you?
A: SIAM’s journals and conferences have been key resources throughout my career, so it’s important to me to help support the society through membership and volunteer activities.
Piyush Sao, Xiaoye S. Li, and Richard Vuduc
Piyush Sao, Xiaoye S. Li, and Richard Vuduc are the 2022 recipients of the SIAM Activity Group on Supercomputing Best Paper Prize. The award will be presented at the SIAM Conference on Parallel Processing for Scientific Computing (PP22), to be held virtually February 23 – 26, 2022. The recipients will give a talk at the conference titled “Optimizing Communication-Avoiding Sparse Lu Factorization on Multi-Gpu Clusters” on Friday, February 25, 2022 at 8:30 a.m. PT.
The prize was awarded to Sao, Li, and Vuduc for their paper “A communication-avoiding 3D algorithm for sparse LU factorization on heterogeneous systems,” trading-off memory to minimize communication to significantly increase strong scaling of sparse direct solvers.
The SIAM Activity Group on Supercomputing awards this prize every two years to the author(s) of the most outstanding paper, as determined by the selection committee, in the field of parallel scientific and engineering computing published within the four calendar years preceding the award year.
Piyush Sao is a research scientist at Computer Science and Mathematics Division (CSMD) in the Oak Ridge National Laboratory, where he works on scalable numerical and discrete algorithms for scientific computing and data analytics. He received his Ph.D. from the Georgia Institute of Technology and B.Tech. from IIT Madras in India. He leads the development effort for the Snapshot Project that was a finalist for the Gordon-Bell prize in 2020 and the R&D 100 award in 2021. He was a part of the team that placed the Summit supercomputer 3rd on the Graph500 list.
Xiaoye S. Li is a Senior Scientist in the Computational Research Division, Lawrence Berkeley National Laboratory. She has worked on diverse problems in high-performance scientific computations, including parallel computing, sparse matrix computations, high precision arithmetic, and combinatorial scientific computing. Xiaoye is the lead developer of SuperLU, a widely used sparse direct solver, and has contributed to the development of several other mathematical libraries, including ARPREC, LAPACK, PDSLin, STRUMPACK, and XBLAS. She has collaborated with many domain scientists to deploy advanced mathematical software in their application codes, including those from accelerator engineering, chemical science, earth science, plasma fusion energy science, and materials science. She earned her Ph.D. in Computer Science from UC Berkeley and B.S. in Computer Science from Tsinghua Univ. in China. Xiaoye has served on the editorial boards of
SIAM Journal on Scientific Computing, and
ACM Transactions on Mathematical Software, as well as many program committees of the scientific conferences. She is a 2016 SIAM Fellow.
Richard (Rich) Vuduc is a Professor at the Georgia Institute of Technology. He works in the School of Computational Science and Engineering, a department devoted to the study of computer-based modeling, simulation, and data-driven analysis of natural and engineered systems. His research lab, The HPC Garage, is interested in high-performance computing, with an emphasis on algorithms, performance analysis, and performance engineering. He received his Ph.D. in Computer Science from the University of California, Berkeley, and was a postdoctoral scholar in the Center for Advanced Scientific Computing at the Lawrence Livermore National Laboratory.
Q: Why are you all excited to receive the SIAG/SC Best Paper Prize?
A: First, it’s a truly humbling honor to receive the prize. When you’ve thought about a fun problem for a long time and finally managed to chip away at it, it’s always exciting to have a chance to share what you did with others. So, the prize gives us a once-in-a-lifetime opportunity to do that. We hope someone who might not otherwise had a look at our paper now does so and learns something new or unexpected from it.
Q: Could you tell us a bit about the research that won you the prize?
A: For any algorithm running on a parallel computer, what always slows you down is communication, that is, the time needed to send data from one processor to another. For linear algebra computations, there has been a lot of work on communication-avoiding methods, which try to reduce communication by possibly trading it off for more computation or storage, with many great results in the case when the matrices are dense. But sparse matrices dominate a lot of real-world science and engineering applications. In our paper, we focus on the problem of solving a sparse system of linear equations by Gaussian elimination (a.k.a., sparse LU factorization) and show how to adapt the ideas for the dense case to the sparse one. Our method involves a “three-dimensional” distribution of data and judicious duplication of data between processes to effectively reduce communication by up to several orders of magnitude, depending on the input problem. We also show how to engineer this algorithm so that it runs effectively on modern supercomputers, including those with graphics co-processors (GPUs).
A random fun fact for us is that this collaboration is a “family affair.” Sherry and Rich are academic siblings (same advisor), who first met over 20 years ago, and Piyush is Rich’s academic child (and so Sherry’s academic nephew). It’s amazing that we get to keep working together and says a lot about the importance of your academic family.
Q: What does your work mean to the public?
A: Systems of linear equations are everywhere, from high school textbooks to high-fidelity simulation codes for weather prediction and climate modeling. Our algorithms can speed up many such computations by a lot on modern parallel computers. This capability will result in faster and higher-resolution predictions, facilitate a better understanding of many scientific phenomena, and expedite many scientific discoveries.
After the paper was published, we also put a lot of effort into solidifying the software. We are very excited to report that earlier this year, this algorithm was released in the SuperLU sparse direct solver library (SuperLU_DIST v7.0), which is a widely used library in both academic and commercial applications. We expect this software to make it easier to put our ideas into practice.
Q: What does being a member of SIAM mean to you?
A: Piyush: As an early career researcher, I find SIAM conferences are among the best places for presenting my work, for keeping up with the latest advances, and meeting and networking with other leading researchers in my research area.
Sherry: SIAM has been the most important professional society that brings together applied mathematics and computing to solve many real-world problems. I have been a SIAM member for over 20 years, and have benefited immensely from reading SIAM journals, attending SIAM conferences, meeting, and collaborating with the other SIAM members on numerous occasions.
Rich: Speaking of family, SIAM, and especially SIAG/SC, is like a big family that you want to hug and keep close (the pandemic notwithstanding). I don’t think I would have stayed in supercomputing had it not been for all the great times and amazing people I’ve met through SIAM.