7/10/2024 Bruce Adams
Laxmikant (Sanjay) Kale, Paul and Cynthia Saylor Professor Emeritus of Computer Science, received the 2024 Achievement Award in High Performance Distributed Computing. He delivered a keynote address and was recognized at the ACM International Symposium on High-Performance, Parallel and Distributed Computing 2024 conference on June 3 in Pisa, Italy.
Written by Bruce Adams
Laxmikant (Sanjay) Kale, Paul and Cynthia Saylor Professor Emeritus of Computer Science, has received the 2024 Achievement Award in High Performance Distributed Computing. He delivered a keynote address and was recognized at the ACM International Symposium on High-Performance, Parallel and Distributed Computing 2024 conference on June 3 in Pisa, Italy.
Kale said, “I was honored to receive the award in Pisa. The citation says: ‘For pioneering development of task-based adaptive parallel programming models and runtime systems, leading to a new category of highly scalable scientific applications.’ While I am grateful for the award, I believe it belongs to my 50-ish PhD students and a similar number of other students and co-workers. In my talk I traced the evolution of the ideas in the migratable-objects programming model (as embodied in Charm++, Adaptive MPI, Charm4Py..). I clearly see the imprint of each generation of graduate students, along with the influence of each co-designed application code in the conceptual evolution of the model. I am very proud of the incredible Charm++ software stack that we have developed and continue to maintain, advance, and distribute, and several highly impactful scalable applications developed using it.”
Kale’s career exemplifies the cross-disciplinary spirit operative at the University of Illinois Urbana-Champaign. Around 1991, he started collaborations in Fluid Dynamics with professor Surya Pratap Vanka and with biophysicist the late Klaus Schulten at the Beckman Institute, using his nascent parallel programming system, Charm++, which he had developed earlier for use in combinatorial search applications. Since then, his work on parallel programming systems has been influenced by the real-world needs of research scientists. Charm++ developed via a process of accretion he described as “co-development and co-design.” Kale noted, “We’d discover a need for a particular language feature while we were co-developing an application and then hone the feature in the context of other applications before incorporating it in the system.” Students from his teams proceeded to work for academia, government laboratories, and large technology companies like Google, Intel, or NVIDIA, often where the performance component of parallel computing is crucial.
Charm++ enabled the development of many highly scalable parallel applications. For example, Kale worked with University of Washington astronomist Thomas Quinn on ChaNGa, an application in cosmological and planet formulation simulations.
Kale was involved in creating the NIH-funded Macromolecular Modeling and Bioinformatics Center at the University of Illinois in 1992 with the first academic multi-petaFLOP/s computer. This center has had what he called “an unbelievable thirty-year run.” The NAMD program for biomolecular simulations he co-developed there remains an open-source application and won the prestigious Gordon Bell award in 2002. NAMD was used to visualize the COVID-19 virus during the pandemic, coming into the view of mass media audiences.
As the director of the Parallel Programming Laboratory, he led the development of several languages and systems for parallel programming, most notably the Charm++ system for adaptive task-based programming. Kale stated, “Having multiple applications co-developed with Charm++ has been beneficial overall.” Charm++ allows programmers to divide computation into a large number of entities, which an intelligent runtime system maps to the available processors. This model enabled many automatic runtime optimizations, leading to improvements in both productivity and performance in parallel programming.
As of 2024, his work has been cited over 42,000 times and has an h-index of 69. Kale has also been recognized as a fellow of the ACM and the IEEE and received the Gordon Bell Prize for outstanding achievement in high-performance computing in 2002 and the Sidney Fernbach Award for outstanding contributions in applying high-performance computers in 2012. He retired from teaching in 2019, continuing an active research program with his research group as a research professor in the Siebel School of Computing and Data Science.