DiRAC Training Skills
DiRAC Training and Skills is a coordinated programme that helps researchers across the STFC communities—astronomy, cosmology, particle and nuclear physics—make the most of the DiRAC high‑performance computing ecosystem. It supports users at every stage, from onboarding and good practice for new PhD students to advanced optimisation for experienced developers, building a shared culture of efficient, reproducible, and secure computing across CPU and GPU systems. The aim is to translate compute allocations into faster science, better software, and a sustainable pipeline of computational expertise.
The programme offers a mix of courses and hands‑on activities covering core HPC topics—parallel programming (MPI/OpenMP), GPU acceleration (e.g., CUDA, OpenACC, SYCL), performance profiling and optimisation, job scheduling with Slurm, workflow orchestration, and containerisation—alongside software engineering practices such as version control, testing, documentation, and reproducibility. It also runs domain‑focused masterclasses (for example, N‑body and hydrodynamics, lattice QCD, inference and pipelines) and data skills for analysis and curation. Delivery blends webinars, workshops, summer schools, hackathons, and code clinics, with access to DiRAC research software engineers for targeted consultancy. Materials are made openly available where possible to maximise reuse across projects and institutions.
Impact is seen in more productive users, faster and more portable codes, and stronger communities of practice that span universities and facilities. Training is aligned with evolving architectures and methods so teams are ready for next‑generation DiRAC upgrades and exascale‑class workflows. Researchers can engage via the DiRAC website’s training calendar and mailing lists, with events open to DiRAC users and, where feasible, the wider UK community. The programme emphasises accessibility and inclusion, offering recorded sessions and clear learning pathways so individuals and groups can plan development over the lifetime of their projects.