Linux servers for complex computing

Introduction

MFCF manages many Linux systems with different configurations and OS distributions. The general-use machines allow interactive login without job control, but the specialty machines are controlled by the Slurm workload manager. These pages describe the specialty machines and the use of Slurm.

Machine list and hardware specifications - HPC clusters and GPU servers

• HPC clusters offer MPICH and OpenMPI for parallel programming
• GPU servers offer CUDA, TensorFlow and Keras, Theano, PyTorch

Access to specialty machines

File systems and file transfer

The Slurm workload manager

Skip straight to the examples if you need just a quick reminder.

• Managing jobs with Slurm
• Slurm partitions
• Environment modules
• Slurm accounts
• Slurm resource limits
• Slurm usage examples
  • Job submission
  • Workflow