SciComp Members Present at PPAM 2026 in Poznan

September 2, 2026.

SciComp members participated in the International Conference on Parallel Processing and Applied Mathematics (PPAM 2026) held in Poznan from 30 August – 2 September 2026, where three members of the group presented their latest research on task-based parallelism, GPU performance optimisation, and dynamic load balancing for scientific computing.

SciComp researchers contributed a keynote and two presentations highlighting the group’s work on developing efficient computational methods that exploit modern heterogeneous hardware and address the challenges of large-scale scientific simulations.

Three members of the group delivered presentations:

• Tobias Weinzierl delivered a keynote titled “Modern Tasking Approaches to Simulate Black Holes (and Other Interesting Phenomena): How Can We Make Them Fit to Modern Hardware?”. The keynote explored how task-based parallelism and modern runtime systems can be used to develop scalable simulation approaches for complex scientific problems, with a particular focus on black-hole simulations and the challenges posed by evolving hardware architectures.
• Mladen Ivkovic presented “Memory Layouts for GPU-Data Transfer Buffering in SPH”. His work investigates how memory layouts can be designed to improve data transfer and buffering efficiency in GPU-based Smoothed Particle Hydrodynamics (SPH) simulations, with the aim of improving performance on modern accelerator architectures.
• Chung Ming Loi, PhD student supervised by Dr. Anne Reinarz, presented “Dynamic Load Balancing for Uncertainty Quantification with Applications in Bayesian Inversion”. His presentation addressed dynamic load-balancing strategies for uncertainty quantification workloads, with applications to Bayesian inversion. The work considers how computational resources can be distributed efficiently when the underlying workload varies, an important challenge in large-scale scientific inference and simulation.