White Paper: The EXCELLERAT Best Practice Guide – Introduction and Overview
)
The guide brings together lessons learned during the first phase of EXCELLERAT and provides practical recommendations for executing engineering applications efficiently on modern HPC systems. Its objective is to help users, application developers and HPC experts prepare engineering workflows for the exascale era.
The document is structured around the typical workflow of scientific computing applications in engineering:
- Pre-processing
This step includes the preparation of input data, model setup, meshing, data transfer to HPC sites, and all actions required before simulation. The guide addresses challenges such as secure and efficient data transfer, preparation of computational grids, and usability for non-HPC experts. - Simulation
The simulation step focuses on executing engineering codes on large-scale HPC resources. Topics include scalability, adaptive mesh refinement, discretisation methods, domain decomposition, solution strategies for linear systems, load balancing, validation, regression testing and efficient I/O strategies. - Post-processing
As engineering simulations produce increasingly large volumes of data, post-processing becomes a key challenge. The guide presents best practices for data reduction, compression, high-performance data analytics and visualisation methods to extract knowledge from simulation results. - Simulation workflow and result feedback
This step covers automated workflow processing, result feedback, uncertainty quantification and methods that support iterative engineering design cycles. - Workflow overarching activities
The guide also addresses cross-cutting activities that are relevant across the full engineering workflow, including performance engineering, co-design, porting to new architectures and efficient implementation.
The Best Practice Guide supports knowledge transfer by providing practical descriptions of methods, approaches and lessons learned. It helps academic users, application developers and engineering communities identify starting points, understand technical bottlenecks and apply HPC more effectively in research and development.
By sharing these best practices, EXCELLERAT contributes to the uptake of HPC and HPDA as tools for faster, more efficient and more sustainable engineering innovation in Europe.
)
