Skip to content

Case StudiesΒΆ

Deploying a shared JupyterHub on the EOSC EU Node

Shared computational research infrastructure matters.

JupyterHub has become a standard piece of infrastructure in data-intensive research across many disciplines. Although this work was initially developed for a STEM project, its value for arts and humanities work is no different, illustrating how DRI initiatives like CCP-AHC allow for cross-discipline knowledge sharing. When a research team needs a common environment (the same software, the same data access, the ability to hand work off between colleagues without a reinstall), a managed JupyterHub is one of the most effective answers available. Every user gets their own notebook server, isolated from their colleagues but pulling from a shared image; a PI can open a colleague's notebook and run it without ceremony.

Archaeological rendering at 2,600 models per hour: HPC for digital heritage collections

Rendered 3D models of Stone Age handaxes

When you have 2,637 three-dimensional high-resolution scans of Stone Age handaxes and need thumbnail images for each, you have a few options. You could render them one by one on your workstation, leaving your desktop running for 38+ hours and hoping for the best. Or, if you happen to have access to a supercomputer, you could have them all done in about twenty minutes.

This post describes how we used Hamilton8, Durham University's HPC cluster, to batch-render thumbnail images for a collection of photogrammetric 3D models at the heart of an AHRC-funded research project. It's a small but concrete example of how HPC infrastructure can quietly transform what's practical in arts and humanities research.