The International Conference on Computational Science (ICCS) brings together researchers and scientists from mathematics and computer science as basic computing disciplines who are pioneering computational methods in sciences such as physics, chemistry, life sciences, and engineering, as well as in arts and humanitarian fields, to discuss problems and solutions in the area, identify new issues, and shape future directions for research.
This year ICCS was held in Málaga, Spain, and we had an ExCALIBUR day to focus on how ExCALIBUR activities and outcomes can be leveraged to address the challenges of the ICCS audience. This was a resounding success and showcased the transformative potential of exascale computing to a wider audience, fostering collaboration, generating significant interest, and providing a platform for insightful discussions.
A Global Gathering of Exascale Experts
The event drew in a diverse mix of in-person and online attendees, demonstrating its global reach. The visibility gained from being part of ICCS 2024 generated significant interest in ExCALIBUR throughout the week, with researchers from diverse disciplines and continents expressing a keen interest in learning more about the project’s goals, achievements, and potential applications of exascale computing.
A Hub for Exascale Innovation and Collaboration
ExCALIBUR Day brought together researchers from various ExCALIBUR projects, including PAX-HPC, SEAVEA, CompBioMedX, CompBioMedEE, QEVEC, EXA-UQ and EXAMPLER, fostering collaboration and showcasing the latest advancements in exascale computing. Presentations covered a wide range of applications, from simulating human anatomy to developing real-time policy evaluation tools.
The event featured a series of engaging discussion panels and sessions, covering a wide range of topics related to exascale computing. These sessions provided opportunities for researchers to share their insights, ask questions, and engage in lively debates.
Key Presentations and Discussions
Gary Polhill (EXAMPLER) raised some legitimate concerns about the realism of having to explicitly predict runtimes for individual jobs as part of ARCHER applications, even when the codes involved have strong non-deterministic components (as is often the case with ABMs).
Peter Challenor (EXA-UQ) provided insights into uncertainty quantification and its importance in exascale simulations, while Arindam Saha from SEAVEA then continued the discussion by demonstrating the effectiveness of Gaussian process emulation as a fast and computationally inexpensive way to approximate simulation results.
Diana Suleimenova showcased the SEAVEA toolkit for actionable and VVUQ-evaluated exascale applications, which includes components for VVUQ workflows, automation, surrogate models, and more. Meanwhile, Yehor Yudin presented pioneering work on aleatoric uncertainty in sensitivity analysis procedures, extending the SEAVEA toolkit.
Peter Coveney introduced the CompBioMedX project, which focuses on deploying high-fidelity human-specific models of the multiscale human anatomy on exascale architectures. Xiao Xue, also from CompBioMedX, then introduced the HemeLB software for modelling the human vascular system, with applications in stroke, aortic stenosis, and AAA studies. Followed up by Jon McCullough (CompBioMedX) who discussed the process of preparing the HemeLB blood flow simulation code for exascale computing by porting it to GPU execution while maintaining portability on multiple hardware types. Jon’s presentation provided practical guidance for researchers looking to optimise their own codes for exascale performance.
Agastya Bhati (IMPECCABLE) presented their integrated modelling pipeline for accelerating drug discovery by combining machine learning and physics-based methods. Andrea Townsend-Nicholson (CompBioMedEE) highlighted their efforts to promote and support the use of computational biomedical modelling and simulation at the exascale. Andrea’s presentation showcased their successful efforts to educate and empower new audiences in this field, emphasising the importance of bridging the gap between exascale research and other fields, and providing practical strategies for overcoming common barriers.
Ava Dean (PAX-HPC) discussed performance aspects for particle simulations across various scales, both on the largest and smallest scales. Ava also presented the PAX-HPC project’s work on parallelism in particle-based simulations, focusing on high-throughput, large-scale simulations, and coupling different simulation methods. Omer Rathmore from the QEVEC project, explored quantum algorithms for accelerating CFD applications, focusing on scalability and practical strategies for harnessing a quantum advantage.
Two lively panel discussions allowed experts to explore the technical challenges and potential benefits of exascale computing, highlighting the need for innovative approaches and solutions and the role of heterogeneous computing, combining CPUs, GPUs, and potentially QPUs, in maximising the performance and capabilities of exascale systems.
An Opportunity for Future Collaboration
The event laid the groundwork for future collaborations and partnerships. The insightful discussions and networking opportunities fostered at the event provided a fertile ground for researchers to connect, exchange ideas, and explore potential collaborations.
ExCALIBUR Day was a testament to the project’s success. The event highlighted the transformative power of exascale computing and fostered a collaborative environment among researchers. Looking forward, ExCALIBUR is poised to continue driving innovation and addressing complex challenges in the field.