Theme: AI and simulations - We all need HPC

Workshop website: https://edin.ac/3DbxZlv

As we continue to progress further into the exascale era it is becoming clearer that, despite people's claims, there is really no gap between HPC and AI. Both these areas require high-performance, rely on the same hardware technologies and often require staff with similar skillsets. Consequently, HPC can be thought of as underpinning both simulation of scientific computing workloads and AI, and we are all in the same boat. However, different communities have different needs around how the HPC infrastructure is used and provided, which then also affect each other. Following this “difference but common interests” theme, we aim to bring disciplines to Durham which have typically been underrepresented in previous years. There are many challenging issues shared between communities but are potentially solved in different ways, and hence we aim for HPC Days 2025 to explore these, identify synergies and shared solutions.

To be held at Durham University, Department of Computer Science, Durham, UK on 2nd June - 7th June 2025

Call for presentations

We invite proposals for talks, typically 15-25 minutes long, which will be embedded into a wider programme hosting invited speakers and vendor presentations. Suggested topics of interest for this conference include (but are not limited to):

• Advanced in AI- and simulation-driven domains enabled through HPC installations
• Optimisation of HPC, AI and ML workloads to deliver increased efficiency and scaling on HPC machines
• Use of novel technologies, techniques and architectures to deliver improved energy efficiency
• Software, programming, and algorithmic approaches for reducing power consumption by codes
• The convergence of AI/ML and HPC applications and algorithms
• Success stories around efforts to encourage Diversity, Equity and Inclusion (DEI) in the HPC workforce
• Approaches to user training
• How we are able to share hardware designed for AI/ML with HPC and vica-versa
• Going beyond the batch system, for instance the use of interactive computing such as Jupyter notebooks, on traditional HPC machines
• HPC centre policy to enable non-traditional workloads
• Lessons learnt from energy efficient HPC centre operation

As HPC is a large community that encompasses many different roles, we welcome both technical and non-technical contributors at all career levels to submit their work in order to broaden the knowledge of all those who attend our event. We also invite colleagues to get in contact with the organisers if they want to organise a minisymposium (typically 4 or 8 talks).

Submission details

We request submission of an abstract, approximate length 300 words which will be peer reviewed, via the EasyChair system at https://easychair.org/conferences/?conf=hpcdays25 .Successful presentations will be required to register and attend the conference in-person.

Committee

Conference Committee

• Marion Weinzierl, ICCS, University of Cambridge
• Eleanor Broadway, EPCC, University of Edinburgh
• Nick Brown, EPCC, University of Edinburgh
• Tobias Weinzierl, IDAS/CS, Durham University
• Alastair Basden, DiRAC/ICC, Durham University

Contact

The conference is organised by Durham's Institute for Data Science. Enquiries can be sent to idas.director@durham.ac.uk.