Events

Peter Dueben will talk on MAELSTROM – Scalable Machine Learning for Weather and Climate Modelling

To develop Europe’s computer architecture of the future, MAELSTROM will co-design bespoke compute system designs for optimal application performance and energy efficiency, a software framework to optimise usability and training efficiency for machine learning at scale, and large-scale machine learning applications for the domain of weather and climate science. This talk will provide an overview on challenges and opportunities for machine learning in weather and climate modelling and explain how some of these challenges will be addressed by MAELSTROM, potentially in collaboration with Time-X.

Maria Garzaran will talk on oneAPI and the Aurora Exascale Supercomputer

Developers need to write code that execute on a variety of hardware, as different algorithms run more efficiently on different architectures. However, while programmers need to tune their codes to select the best data layout, the data movement across different memories and/or devices, and how to best map the parallelism to the target device, they also need a programming environment that allows them to write portable codes that run across different architectures and vendors.

In this talk, I will introduce oneAPI, a cross-industry, open, standards-based unified programming model that allows programmers to use the same language and tools across different hardware architectures. I will also introduce Aurora, the Intel-HPE exascale supercomputer for Argonne National Laboratory, built to support the next science and AI discoveries.

Jemma Shipton will talk on ExCALIBUR – time-parallelism, and symbolic manipulation for rapid prototyping of new algorithms

This talk will have two parts. Firstly, I will give a high-level overview of the ExCALIBUR programme, a UK research programme to that aims to deliver the next generation of high-performance simulation software for the highest priority fields in UK research. I will describe the aims and structure of the programme and the project that I am leading: Advanced Parallel in Time Algorithms for Partial Differential Equations (APinTA PDEs). I will then introduce Gusto, the compatible finite element dynamical core toolkit, which provides stable, accurate, scalable methods for solving geophysical fluid dynamics equations. Within Gusto we have implemented symbolic manipulations of the spatial discretisation of these equations to enable rapid prototyping of complex timestepping algorithms, including time-parallel algorithms. I will demonstrate how this works and show our latest results.