Technological Platform High Performance Computing
HPC is a key step in present development of both new technologies and fundamental science but is also essential for the analysis of large amount of data. The PTCI platform proposes to the user an access to high-performance computing (HPC) equipment combined with a high level technical support. Our service includes the estimation of the hardware requirements and an assistance to optimize the software performances on our facilities. Our cluster is dedicated to general purpose job and handle many concurrent jobs from several users with varying requirements.
The PTCI is part of the shared infrastructure of the Fédération Wallonie-Bruxelles CÉCI (Consortium des Équipements de Calcul Intensifs).
The PTCI has an expertise for the management of HPC equipment and for the installation and optimization of codes dedicated to particular applications, including the license management. The fields of expertise of the University research teams involved in the PTCI are very broad:
Chemistry, mathematics, physics, economy, biology, geography, computer science, astronomy, exoplanets
Big data, complex systems, networks, optimization, control
Materials, spectroscopies, optics, numerical simulations
Drugs, photovoltaïc, biomimetism, polymers, molecular electronics
Supercalculator, parallel computation, compiler
Supramolecules, optics, metal, semi-conductors, growth, genetic algorithm, efficiency, solar cells, superabsorber
Tensor calculations, waves, image processing
The main cluster (Hercules) comprises more than 900 cores spread across 65 compute nodes including multi-core nodes (from 12 to 24 cores) with 24 to 128 GB of RAM and GPGPU. All the nodes are interconnected by a Gigabit Ethernet network and have access to several shared storage systems for a total capacity of 100 TB. Two hardware accelerated interactive visualization nodes provide a way to remotely run 3D accelerated application.
More than 100 softwares are installed, including most of the general programming languages, home-made codes developed for particular applications, major scientific codes under license or visualization tools. As part of the CÉCI shared infrastructure, the PTCI also gives access to larger HPC facilities.
For technical details about the cluster hardware, please refer to the Hardware page.
The Hercules Cluster
Hercules is primarily designed for users who need to run either shared memory parallel jobs (OpenMP or Pthreads), or multiple resource-intensive sequential jobs.
This system has 1000 cores, 4.0 TB of total memory, and 70 TB of shared storage. He is capable of 14 TFLOP/s theoretical peak performance. Two GPU enabled nodes are available for general use.
Users with an account can connect to one of the Hercules login nodes using SSH. The main login node is hercules.ptci.fundp.ac.be. See the How to connect page for details of how to configure your SSH client and transfer files to and from your local computer.
Want to use the cluster? Check the PTCI usage policies page.