Chris Grave, Martin Wolstencroft and Dave Love from the Platforms team in Research IT recently spent a week at Durham University for their HPC Days 2026 event and it’s fair to say it was a packed (and very worthwhile) few days. The event brought together a range of people working across HPC, from Research Technical Professionals (RTPs) and Research Software Engineers (RSEs) through to leading academics, funders, and industry vendors. That mix made for a highly engaging atmosphere, with plenty of opportunities to share ideas across different parts of the community.

The programme covered a wide range of topics, all focused on HPC. Workshops ranged from hands-on sessions like “Build Your Own HPC,” where Raspberry Pis were used to build a Slurm cluster from the ground up, through to more operational discussions, such as sessions focused on closing the EDI implementation gap in HPC. These drew on lessons from EuroHPC, UKRI, and beyond.

The keynote talks were a particular highlight. Prof. Alice-Agnes Gabriel (Scripps Institution of Oceanography, UC San Diego, California) presented her work on how supercomputing is transforming earthquake and tsunami science, including her 2025 ACM Gordon Bell Prize–winning work on a real-time tsunami digital twin for Cascadia. At the other end of the spectrum, Dr Luke Davis (UKRI) provided an update on the UK’s national HPC landscape, including UKRI’s plans for the Next National Supercomputing Service (NNSS) and the new National Compute Resources (NCRs), building on the UK Compute Roadmap (July 2025) vision for a world-class, federated compute ecosystem.

There was also a strong technical focus throughout the week. Tutorials included sessions from NVIDIA on model parallelism for training large neural networks, alongside a more traditional HPC optimisation topic from AMD using OpenMP for GPUs to accelerate portable scientific applications. Other sessions covered everything from user-driven, large-scale federated data movement to computational engineering.

A couple of personal highlights stood out. Dr Andy Turner (EPCC) delivered an excellent presentation on Green HPC and responsible computing—particularly the idea of developing “responsible computing plans” (think Data Management Plans, but for computing). This tied in nicely with local efforts in the Green Compute for Research @UoM CaDiR group and felt especially relevant as we consider sustainability and efficiency in our own environments. Another standout was Dr Albert Reuther’s (MIT Lincoln Laboratory Supercomputing Center) keynote on interactive, on-demand HPC at MIT Lincoln Laboratory, which offered an interesting look at user expectations around interactivity and on-demand compute services.

Alongside the talks and workshops, there was also a vendor exhibition where companies showcased the latest hardware and software developments shaping the future of HPC. As always, it was useful to see where the industry is heading and how those trends might map onto our own platforms and services.