Bede Documentation

Bede is a supercomputer (otherwise known as an HPC system) run by the N8 group of research intensive universities in the north of England, on behalf of EPSRC, and hosted at Durham University.

The system has a specialist architecture, optimised for large memory problems on GPUs and for multi-node multi-gpu programs. This has a range of applications in Machine Learning and Imaging.

The site encourages user contributions via GitHub. Feel free to add items of concern to the Github issues section.

Please note that the system is still under active development, and so some functionality may temporarily break.

Site Contents

• Hardware
• Usage
  • Registering
  • Login
  • File Storage
  • Running Jobs
    • Requesting resources
      • Part of, or an entire node
      • Multiple nodes (MPI)
      • Multiple nodes (IBM PowerAI DDL)
• Software
  • Environments
    • Spack
    • Easybuild
  • Compilers
    • GCC
    • LLVM
  • BLAS/LAPACK
  • MPI
  • HDF5
  • NetCDF
  • PyTorch Quickstart
  • PyTorch and TensorFlow: IBM PowerAI and Watson Machine Learning Community Edition (wmlce) [Possibly Out of Date]
• Profiling
  • NVIDIA Profiling Tools
    • Compiler settings for profiling
    • Nsight Systems and Nsight Compute
      • More info
      • Cluster Modules
    • Visual Profiler (legacy)
      • Documentation
      • Cluster Modules
  • NVIDIA Tools Extension
    • Documentation
• Useful Training Material
  • Profiling Material
  • General Training Material