Skip to main content

Getting started

MeosBFC is a new computing cluster based on modern CPU and GPU architectures.
It is intended for deep-learning, machine-learning and for applications requiring newer Operating System.

Main features​

  • GPU computing. Only for deep-learning and machine-learning applications
  • GPU are based on NVIDIA A100 technologies
  • Parallel MPI and SMP based applications
  • Interactive sessions for post-processing and graphical applications
  • Resources management are based on Slurm
  • High speed network (Intel Omni-Path and Mellanox InfiniBand) 100G/s
  • Parallel file system based on BeeGFS
  • Software and packages are installed using Spack package management tool
  • Operating System is Rocky Linux 8.7

HPC workflow​

A typical use case for an HPC cluster is as follows:

  1. Connect to the cluster
  2. Transfer your files to the cluster
  3. Compile your code and test it or load installed software using module
  4. Create a job script
  5. Submit the job
    1. job gets into the queue
    2. job gets executed
    3. job finishes
  6. Move your results

In this documentation, we will learn each of these points.