Skip to main content Skip to navigation

About Sulis

Sulis is a tier 2 HPC platform focussed on high throughput and ensemble computing with the following principle objectives.

  • Deliver substantial HPC capacity targeted specifically at data-intensive high-throughput workloads. This will facilitate ensemble computing, i.e. workflows involving thousands of small (workstation scale) calculations, replicated to concurrently sample or generate data over a space of inputs or model parameters. Such workloads underpin Uncertainty Quantification (UQ), training of proxy models from traditional simulation engines and other data-intensive areas of computational science and engineering.
  • Provide users with access to modern technologies appropriate to high-throughput computing. We will enable rapid migration to Sulis by adopting the latest HPC software deployment and containerisation technologies, allowing scale-up from desktop to Tier-2 with minimal code modification. The hardware will provide users with access to high core-count x86 servers, high speed SSD storage for storing intermediate data, and to the latest GPU architecture for post-processing and training of model parameters from simulated/sampled data.
  • Develop Research Software Engineering training which recognises that traditional large-scale MPI applications are not always the most appropriate or efficient route to scale-up of scientific computing research. It will explicitly address barriers to deploying data-intensive workflows on traditional HPC platforms, specifically software availability, IO limitations and walltime limits. We will train users in the use of emerging software frameworks for simple user-level implementation and management of task-queue based parallelism, containerisation technology, use of databases and user-level checkpointing/restart functionality.

The Sulis hardware is funded by a £3M grant from the Engineering and Physical Sciences Research council, and is managed by Warwick on behalf of the HPC Midlands+ consortium of universities.