Modelling the aerodynamic performance of race car geometries forms a large part of the computational workload of Formula 1 teams. The aim of a Formula 1 is to design a car with the best possible aerodynamic performance, and computational modelling of the car geometry is a key aspect here. The car geometry is represented as a complex mesh with many millions of cells – generating this mesh, and partitioning it to be ready for use by the parallel computational fluid dynamics (CFD) solver, is a difficult, and often poorly performing, aspect of the overall workload.
This project will investigate how to improve the overall performance of the “meshing” components of the computational toolchain. The primary focus is on improving the time to solution and the scalability – ultimately, the aim is for the meshing components of the workload to complete in as short a time as possible. Areas for investigation might include offloading work to accelerators, bespoke vectorisation, optimising data layouts and memory footprints, and improving parallel efficiency through better load balancing. AI and Machine Learning techniques could form part of the solution.
A secondary aspect of the project is performance portability. The aim is to support a wide range of hardware architectures whilst maintaining good performance. Achieving performance portability means that applications can cope with hardware changes that are introduced over time. Starting from a baseline of portable performance, applications can be fine tuned to deliver optimal performance for a given hardware architecture.
The student will be based at EPCC, the supercomputing centre at the University of Edinburgh. The project will be run in close collaboration with Renault Sport Racing. The student will have access to the wide range of HPC system architectures at EPCC.
This project is available with a start date between 11 January and 31 May 2021. Applications for a start date between 01 June and 31 July 2021 may also be considered. Please note your preferred start date in any correspondence and your application.
Student Requirements:
Minimum: A UK 2:1 honours degree in a relevant subject such as computer science and informatics, physics, mathematics, or engineering. You must be a competent programmer in at least one of C, C++, or Fortran. Standard University English language requirements. For more information see: https://www.ed.ac.uk/studying/postgraduate/degrees/index.php?r=site/view&edition=2021&id=855Edinburgh.
Student Recommended/Desirable Skills and Experience
The following skills are a bonus, but they are not essential for a student to be considered eligible:
- Understanding of HPC programming concepts;
- Experience of using HPC systems;
- Application benchmarking and optimisation;
- Interest in CFD and meshing applications.