LAMMPS-KOKKOS: Performance Portable Molecular Dynamics Across Exascale Architectures

TL;DR

This paper discusses how LAMMPS, a molecular dynamics code, has been adapted for performance portability across diverse exascale architectures using Kokkos, demonstrating strong scaling and performance analysis on multiple supercomputers.

Contribution

The paper introduces the integration of Kokkos into LAMMPS for performance portability and evaluates its performance across various hardware architectures and supercomputers.

Findings

Achieved performance portability of LAMMPS on GPUs from different vendors.

Demonstrated strong scaling on multiple exascale supercomputers.

Analyzed performance trends related to FLOPS, memory bandwidth, and cache capabilities.

Abstract

Since its inception in 1995, LAMMPS has grown to be a world-class molecular dynamics code, with thousands of users, over one million lines of code, and multi-scale simulation capabilities. We discuss how LAMMPS has adapted to the modern heterogeneous computing landscape by integrating the Kokkos performance portability library into the existing C++ code. We investigate performance portability of simple pairwise, many-body reactive, and machine-learned force-field interatomic potentials. We present results on GPUs across different vendors and generations, and analyze performance trends, probing FLOPS throughput, memory bandwidths, cache capabilities, and thread-atomic operation performance. Finally, we demonstrate strong scaling on three exascale machines -- OLCF Frontier, ALCF Aurora, and NNSA El Capitan -- as well as on the CSCS Alps supercomputer, for the three potentials.