skip to content

Research Computing Services

On 30th March 2017 the University of Cambridge Research Computing Service, in partnership with EPSRC and STFC, announced a co-investment of £14M for the creation of the Cambridge Service for Data-Driven Discovery (CSD3). The EPSRC element of the service is a consortium led by the University of Cambridge in partnership with the Universities of Bristol, Leicester, Southampton and Oxford, Kings College London, UCL and Imperial College. CSD3 includes a GPU-accelerated cluster (Wilkes), in addition to a pure CPU cluster (Peta4).

At the time of installation, Wilkes2 was the largest GPU enabled supercomputer for academic research in the UK, comprised of 90 nodes each with 4 Nvidia P100 GPUs.

In November 2021 Wilkes2 was succeeded by Wilkes3, based on Nvidia A100 GPUs. Wilkes3 scored third position in the June 2021 Green500 list and was the highest scoring UK academic machine on the June 2021 Top500 list.

System Features of Wilkes3

Nationally available infrastructure for GPU compute via EPSRC, STFC and Industry.

80 nodes based on Dell PowerEdge XE8545.

Each node contains:

  • 2x AMD EPYC 7763 64-Core Processor 1.8GHz (128 cores in total)
  • 1000 GiB RAM
  • 4x NVIDIA A100-SXM-80GB GPUs
  • Dual-rail Mellanox HDR200 InfiniBand interconnect

Connects to the 5PB/s LustreFS shared with Peta4.

Additional Information

A non-exhaustive list of research areas expected to particularly benefit from Wilkes3 includes:

  • Turbulence simulations (relevant to the aeronautical industry).
  • Combustion simulations (relevant to the automotive industry).
  • Analysis of large data sets from the Internet of Things (relevant to smart cities research).
  • Materials modeling research.
  • Medical applications, including:
    • Analysis of large medical image data sets.
    • Population-scale gene variant analysis.
    • Real-time medical analytics for use in hospital decision support systems.