Equipment

  • 266 bi-socket nodes Intel Xeon E5-2860v3 2.5GHz 12 coeurs, 14 of them for visualization with 28 GPUs on NVIDIA K80 cards.
  • A total of 36 000 Gigabytes of system memory, 5.33 GB per core.
  • A high-speed interconnect network InfiniBand FDR (56 GBps) between nodes.
  • +900TB High Performance Distributed and Parallel Storage Spectrum Scale.
  • Compute nodes are exclusively cooled with warm water using the BULL’s Direct Liquid Cooling concept reducing by third the electrical consumption.
  • Facilities are located in the Nantes Central Campus secured by survey cameras and limited nominative card access and personal pin.

Services

  • Compute OnDemand
  • HPC and 3D Remote Visualization optimized through a web portal BULL XCS.
  • Submissions batch system with SLURM.
  • Technical Support done by HPC and system experts.
  • Training Courses OnDemand; Linux, HPC, parallel programming, etc.
  • Courses and Tutorials are freely available online as Open Licence.

Example of MRE applications

  • Monolithic approach and fluid coupling (air) / fluid (water) / structure (wind)
  • Immersed Volume Methods
  • Adaptive multiphase flow model and discrete by finite elements
  • Development of a direct model, without simplifications
  • Multiphase CFD massively parallel code, highly scalable and associated to the Pays de la Loire Area SuperComputing Facilities.
  • Wind turbines and offshore wind turbines simulations, simulation based on actual data (laser, satellites, sonar, etc.), aerodynamic simulation on submersive objects and simulations of swells, currents and seabeds.