HPC Cluster – HERMES

The Institute of Crystallography hosts HERMES (High-performance Environment for Research in MolEcular and Structural Biology), a high-performance computing cluster designed to provide support to the scientific community for CryoEM, BioSAXS, X-Ray data storage and processing, as well as to enhance theoretical and computational research in Structural Biology and Drug Discovery. Specifically, the system consists of 5 CPU compute nodes and 4 CPU/GPU compute nodes, each with 4 NVIDIA H100 GPUs capable of delivering high computational power for applications such as image processing, artificial intelligence model training, and biological big data processing. Additionally, the system features a Scale-Out NAS storage system capable of supporting the storage of large amounts of data.

HPC Cluster details

• N. 5 Compute nodes Dell PowerEdge R760: 2x Intel Xeon Gold 6448Y 2.1G, 32C/64T, 16GT/s, 60M Cache; 1024GB RAM 4800MHz DDR5
• N. 4 GPU nodes Dell PowerEdge R760XA: 4x GPU Nvidia H100 94GB NVLINK Bridge
• N. 2 Login/service nodes Dell PowerEdge R760xs
• Storage: Scale-Out NAS 4x Dell PowerScale Isilon A3000 nodes: 1280 TB raw space

Software Stack

• Compilers and libraries: C/C++, Fortran, Python, Perl, Julia, MPI, OpenMP, CUDA
• Molecular Dynamics: Gromacs, NAMD, LAMMPS
• Structural Biology: MX packages, BioSAXS software, Cryo-EM software, AlphaFold2, Imaging
• Computational Chemistry: CP2K, Nwchem, Quantum Espresso, DeepMD
• AI and Deep Learning: Tensorflow, Keras
• Statistics: R
• Drug discovery: Schrödinger Suite
  1. Docking and Virtual Screening
  2. Molecular Dynamics Simulations
  3. Homology Modelling
  4. QSAR and cheminformatics models