Exascale Computing and Its Impact on Multiphysics Simulations: Current Status and Future Directions

Argonne National Laboratory (ANL), USA

^* Corresponding author: siegela@mcs.anl.gov

Published online: 15 October 2024

Abstract

This talk provides a detailed analysis of the current landscape and future directions in high-performance computing (HPC) systems, with a particular focus on the impact of evolving hardware trends on complex, multiphysics simulations. The recent deployment of the Frontier Supercomputer at Oak Ridge National Laboratory marked the beginning of the exascale computing era, closely followed by the introduction of the Aurora system at Argonne National Laboratory. Their deployment enabled the successful completion of the U.S. Department of Energy’s Exascale Computing Project (ECP)—a complementary, high-risk effort in re-designing software and scientific applications to take efficient use of these systems. Both Frontier and Aurora utilize specialized accelerators to achieve the power performance necessary for practical exascale deployment. This shift toward GPU-based systems has strongly influenced the HPC community’s approach to algorithms, impacting everything from implementation specifics to numerical strategies and mathematical modeling. Looking ahead, further specialization is inevitable, driven largely by the need for energy efficient neural network training and inference. This talk addresses the potential impact of these trends on multiphysics simulations, discussing both conventional and more innovative approaches to progress.