More bang for your buck: Improved use of GPU nodes for GROMACS 2018

• Published in: Journal of computational chemistry Vol. 40; no. 27; pp. 2418 - 2431
• Main Authors: Kutzner, Carsten, Páll, Szilárd, Fechner, Martin, Esztermann, Ansgar, Groot, Bert L., Grubmüller, Helmut
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 15.10.2019; Wiley Subscription Services, Inc
• Subjects: benchmark; Central processing units; Computer simulation; computer simulations; CPUs; CUDA; Energy costs; energy efficiency; GPU; Graphics processing units; GROMACS; Hardware; high throughput MD; Molecular dynamics; Nodes; parallel computing; performance to price; CUDA; energy efficiency; computer simulations; molecular dynamics; performance to price; GPU; GROMACS; high throughput MD; parallel computing; benchmark
• DOI: 10.1002/jcc.26011
• ISSN: 0192-8651; 1096-987X

We identify hardware that is optimal to produce molecular dynamics (MD) trajectories on Linux compute clusters with the GROMACS 2018 simulation package. Therefore, we benchmark the GROMACS performance on a diverse set of compute nodes and relate it to the costs of the nodes, which may include their lifetime costs for energy and cooling. In agreement with our earlier investigation using GROMACS 4.6 on hardware of 2014, the performance to price ratio of consumer GPU nodes is considerably higher than that of CPU nodes. However, with GROMACS 2018, the optimal CPU to GPU processing power balance has shifted even more toward the GPU. Hence, nodes optimized for GROMACS 2018 and later versions enable a significantly higher performance to price ratio than nodes optimized for older GROMACS versions. Moreover, the shift toward GPU processing allows to cheaply upgrade old nodes with recent GPUs, yielding essentially the same performance as comparable brand‐new hardware. © 2019 Wiley Periodicals, Inc. Hardware that is optimal to produce MD trajectories on Linux compute clusters with the GROMACS 2018 simulation package is identified. The trajectory output per invested Euro is extremely dependent on the type of acquired hardware. Nodes with inexpensive consumer GPUs produce up to six times as much trajectory per Euro as CPU nodes or nodes with Tesla GPUs. Upgrading old nodes with modern GPUs yields even higher performance to price ratios.