Optimizing GPU Performance - Case Study Using Chain Benchmark in LAMMPS

Explore GPU performance optimization techniques through a case study using the Chain Benchmark in LAMMPS in this 44-minute conference talk presented by Stan Moore from Sandia National Laboratories. Delivered at the Institute for Pure & Applied Mathematics (IPAM) workshop on Co-design for the Exascale and IPAM Hackathon, this presentation delves into strategies for enhancing computational efficiency in molecular dynamics simulations. Gain insights into the challenges and solutions associated with optimizing GPU performance for complex scientific applications, specifically focusing on the LAMMPS software package and its Chain Benchmark. Discover how these optimization techniques can be applied to improve the performance of large-scale simulations in various scientific domains.