Analysis of Vector Particle-In-Cell (VPIC) memory usage optimizations on cutting-edge computer architectures

https://doi.org/10.1016/j.jocs.2022.101566Get rights and content
Under a Creative Commons license
open access

Highlights

  • •

    Half-precision and fixed-point storage optimizations enable larger PIC simulations.

  • •

    Grid resolution mitigates accuracy loss from using lower precision particle storage.

  • •

    Cutting-edge CPU and GPU architectures enable high performance.

  • •

    Data structure format optimizations reduce data movement and improve performance.

Abstract

Vector Particle-In-Cell (VPIC) is one of the fastest plasma simulation codes in the world, with particle numbers ranging from one trillion on the first petascale system, Roadrunner, to ten trillion particles on the more recent Blue Waters supercomputer. As supercomputers continue to grow rapidly in size, so too does the gap between computing capability and memory capability. Current memory systems limit VPIC simulations greatly as the maximum number of particles that can be simulated directly depends on the available memory. In this study, we present a suite of VPIC memory optimizations (i.e., particle weight, half-precision, and fixed-point optimizations) that enable a significant increase in the number of particles in VPIC simulations. We assess the optimizations’ impact on memory and runtime performance for a suite of cutting-edge computer architectures such has the NVIDIA V100 GPU, the IBM Power9, and the Fujitsu A64FX architectures. Our optimizations enable a 31.25% reduction in memory usage and up to 40% increase in the number of particles. This paper extends our work on developing particle storage format optimizations Tan et al. (2021) [1].

Keywords

Memory and runtime performance
Mixed-precision
Fixed-point numbers
Plasma physics
High performance computing

Cited by (0)

Nigel Tan is a Ph.D. student in Computer Science under Dr. Michela Taufer at the University of Tennessee, Knoxville. He earned his B.S. in both Computer Science and Applied Math at the University of California Merced before earning an M.S. in Computational and Applied Math at Rice University.

Nigel’s research interests lie in high performance computing with an emphasis on performance portability and optimization across multiple architectures.

Robert Bird is a Computational Scientist at Los Alamos National Laboratory, in the Applied Computer Science group. His research interests are performance-portability and low-level code optimization. He is the computer science lead for the VPIC project. Dr. Bird received his Ph.D in Computer Science from the University of Warwick, England.

Guangye Chen received his B.S. and M.S. degrees in Aerospace Engineering from the Beijing University of Aeronautics and Astronautics in 1999 and 2002, respectively, and a Ph.D. degree in Aerospace Engineering from U. of Texas at Austin in 2008. He is currently a staff scientist in the Theoretical Division at the Los Alamos National Laboratory. His research interests include machine-learning, computational plasma physics, and high-performance computing.

Scott V. Luedtke is a Postdoc Research Associate at Los Alamos National Laboratory in the Plasma Theory and Applications group. His research interests include high energy density physics, high intensity short-pulse laser–plasma interactions, and large-scale physical simulation. Dr. Luedtke received his Ph.D.\ in physics from the University of Texas at Austin in 2020.

Brian Albright (Senior Member, IEEE) is a Fellow of the American Physical Society and a Laboratory Fellow in the X-Theoretical Design Division of Los Alamos National Laboratory. His research interests include plasma and high energy density physics, high performance computing, numerical methods, and defense applications. Dr. Albright received his Ph.D. in physics from UCLA in 1998.

Michela Taufer (Senior Member, IEEE) is an ACM Distinguished Scientist and holds the Jack Dongarra professorship in high performance computing with the Department of Electrical Engineering and Computer Science, University of Tennessee Knoxville. Her research interests include high-performance computing, volunteer computing, scientific applications, scheduling and reproducibility challenges, and in situ data analytics. Dr. Taufer received her Ph.D. in Computer Science from the Swiss Federal Institute of Technology (ETH) in 2002.

© 2022 The Authors. Published by Elsevier B.V.