PAX-HPC: Particles at Exascale

PAX-HPC combines expertise in Massively Parallel Particle Hydrodynamics (MPPH) and Materials & Molecular Modelling (MMM), including many community partners such as the Materials Chemistry Consortium (MCC) and UK Car-Parrinello Consortium (UKCP).  It focuses on particle-based simulations across many length- and time- scales, from atoms to galaxies. 

New methods for new hardware: 

  • Performance-portable programming methods (e.g., SYCL, OpenACC) 
  • Massively-parallel methods for common operations (e.g., matrix diagonalisation) 
  • Explore alternatives to common bottlenecks (e.g., parallel FFTs) 
  • Exploit emerging network technologies, such as DPUs and GPU-aware communication layers 
  • Work with hardware vendors toco-design new hardware solutions 

 New parallelisation strategies: 

  • Exploit parallelism at all levels 
  • High-level additional parallelism, e.g., 
    •    Task-farming 
    •    Parallel-in-time methods 
  • Low-level task parallelism: 
    •    Break operations into small tasks 
    •    Task graphs describe dependencies 
    •    Task engine assigns tasks to hardware 

Complex workflows: 

  • Coupling codes to deliver new scientific capabilities 
  • Multiscale, multiphysics models by coupling codes together in a single workflow 
  • Focus computational effort where it is most needed 
  • FAIR data enables novel data science 

Example applications from MPPH and MMM: 

  • Electrochemical cells have many applications, from energy storage to catalysis 
  • Electron transfer and chemistry at the electrodes requires quantum mechanics to model accurately 
  • Multi-body multi-scale fluid dynamics simulations have many applications from offshore renewable energy converters to galaxy formation. 

Latest news