System and Application Performance Modeling and Simulation in the AI Era
The increasing complexity and heterogeneity of systems at large scale, combined with challenging characteristics of applications driven by data, and dominated by adaptivity and irregularity poses the need for fundamental re-thinking and re-tooling of Modeling and Simulation (ModSim) for systems and applications. In this presentation, after an overview of the state-of-the-art for traditional methodologies and tools, we will discuss new ideas related to Machine Learning both as an increasingly important application workload, and as a methodology for ModSim. In the context of mapping such applications to state-of-the-art systems we will analyze the need for “dynamic performance modeling” as an actionable way to effectively optimize for performance during execution.
Adolfy Hoisie is a lead Computer Scientist at the Brookhaven National Laboratory where he directs the Computing for National Security Department. Before joining Brookhaven, he was with the Pacific Northwest National Laboratory as Director of the Advanced Computing, Mathematics, and Data Division, and a Laboratory Fellow. Prior to PNNL, he served in a variety of scientific and leadership positions at Los Alamos National Laboratory, including as leader of the Computer Science for High-Performance Computing Group and its world-renowned Performance and Architecture Laboratory. Adolfy is an internationally recognized expert in performance analysis, modeling, and engineering of extreme-scale parallel computing systems and applications and system architecture. Adolfy is a past winner of the Gordon Bell Award and has served extensively in the high-performance computing community in various capacities, including conference organizer, editorial board member, committee and panel member, and on advisory boards. He has published extensively in peer-reviewed literature and co-authored three books.