On Wednesday 11 March, just before the COVID-19 lockdown in Belgium, the nVidia DGX Station for the SSAVE project arrived at the Royal Military Academy. It is normally intended to be used to analyze mathematical models to quantify the reliability and accuracy of the data that is exchanged between sensors of different maritime platforms and incorporate these models into decision making tools. In the meantime, the COVID-19 crisis struck hard, and the Folding@Home project announced that it is launching computations in support of the search for drug binding sites on the new virus.
We decided to immediately deploy the DGX station, together with our existing computing cluster and to donate spare computing cycles to this project. Needless to say, the system is configured in such a way that computations required by our own research take immediate priority.
From the first calculations for Folding@Home, we can already make a relative estimate of the performance of our new GPU machine compared to the rest of the cluster: one Tesla card is about 4.4 times faster than a “regular” 80-core compute node, so extrapolating this, the shiny new toy is about three times as fast as the whole regular computing cluster for the Folding@Home workload.
You can track the RMA team’s progress here.