Berk Hess

Davood Saffar Shamshirgar, Berk Hess, Anna-Karin Tornberg

The smooth particle mesh Ewald (SPME) method is an FFT based method for the fast evaluation of electrostatic interactions under periodic boundary conditions. A highly optimized implementation of this method is available in GROMACS, a widely used software for molecular dynamics simulations. In this article, we compare a more recent method from the same family of methods, the spectral Ewald (SE) method, to the SPME method in terms of performance and efficiency. We consider serial and parallel implementations of both methods for single and multiple core computations on a desktop machine as well as the Beskow supercomputer at KTH Royal Institute of Technology. The implementation of the SE method has been well optimized, however not yet comparable to the level of the SPME implementation that has been improved upon for many years. We show that the SE method is very efficient whenever used to achieve high accuracy and that it already at this level of optimization can be competitive for low accuracy demands.

Petter Sandås, Sergio Iserte, Íñigo Aréjula-Aísa et al.

Static resource allocations in high-performance computing (HPC) lead to inefficiencies for time-varying workloads, causing idle resources, queue delays, and higher node-hour costs. The Dynamic Management of Resources (DMR) middleware enables MPI process malleability in Slurm via a simple API decoupled from scheduler internals. In this work, we integrate DMR into the GROMACS molecular dynamics engine to obtain a malleable variant that can dynamically adapt its MPI process count by combining communication-efficiency-aware reconfiguration with GROMACS' native checkpoint/restart mechanism. We evaluate this design on the MareNostrum~5 supercomputer, comparing dynamic runs against static executions and quantifying reconfiguration overheads, time-to-solution, and node-hour savings for bursty GROMACS workloads.