CP2K biomolecular QM/MM benchmarking data

<p>Dataset containing&nbsp;raw CP2K output logs produced by running benchmarks from the BioExcel QM/MM benchmark suite (<a href="https://doi.org/10.5281/zenodo.6591692">https://doi.org/10.5281/zenodo.6591692</a>), provided as part of BioExcel-2 project deliverable D1.6.&nbsp;</p> <p>An analysis of (a subset of)&nbsp;these benchmark results can be found in&nbsp;<a href="https://doi.org/10.5281/zenodo.6591574">https://doi.org/10.5281/zenodo.6591574</a>&nbsp;</p> <p>&nbsp;</p> <p><strong>Machines used for benchmarking:</strong></p> <p>Cirrus@EPCC (HPE SGI ICE XA):</p> <p><a href="https://www.cirrus.ac.uk/">https://www.cirrus.ac.uk</a></p> <ul> <li>Infiniband interconnect</li> <li>CPU compute nodes: <ul> <li>2 x 18-core Intel Xeon (Broadwell) E5-2695, 2.1 GHz</li> <li>256GB RAM</li> </ul> </li> <li>GPU compute nodes: <ul> <li>2 x NVIDIA Tesla V100 (Volta) SXM2-16GB</li> <li>2 x 20-core Intel Xeon (Cascade Lake) Gold 6248, 2.4 GHz</li> <li>384GB RAM</li> </ul> </li> </ul> <p>ARCHER2@EPCC (HPE CRAY EX):</p> <p><a href="https://www.archer2.ac.uk/">https://www.archer2.ac.uk/</a></p> <ul> <li>Interconnect: HPE Cray Slingshot</li> <li>Compute nodes: <ul> <li>2 x 64-core AMD EPYC (Zen2 Rome) 7742, 2.25GHz</li> <li>256GB RAM</li> </ul> </li> </ul> <p><strong>Benchmarking protocol</strong></p> <p>Repeated runs to rule out machine noise variability were performed for each benchmark on each machine for both 1 MD step and 6 MD steps. Subsequent analysis and visualisation of parallel scaling of average runtime per MD step and subroutine-level profiling was performed using our analysis script available from:</p> <p><a href="https://doi.org/10.5281/zenodo.6591681">https://doi.org/10.5281/zenodo.6591681</a></p> <p>Results on Cirrus CPU nodes are for CP2K release version 8.1, whilst results on Cirrus GPU nodes and ARCHER2 are with CP2K version 8.2.</p>