+++ using Dates
title = "A EuroHPC Success Story" date = Date(2024, 05, 13) reading_time = "3-minute read"
tags = ["amdgpu", "julia", "eurohpc"] +++
Significant advances running FastIce.jl on LUMI-G reflected in an interview with EuroHPC about the STREAM project.
Ludovic Räss
Our on-going ice flow modelling efforts using AMD GPUs on the European flagship supercomputer LUMI are now featured in an online interview with EuroHPC about Improved modelling of ice flow to better anticipate sea-level rise.
The STREAM research team believes that new techniques in high-performance extreme-scale GPU computing will revolutionise ice flow modelling in the future and provide more accurate predictions. The team tackled the complexities of ice flow dynamics by employing a computational approach that involves solving partial differential equations (PDEs) on GPUs.
Amongst the latest result, we are now working towards resolving the Vavilov ice cap in order to understand the processes that lead to the rapid destabilisation of ice and the formation of an ice stream.
These preliminary results already suggest an acceleration of flow in the region of the ice flow instability.
<center>
<img src="../../assets/images/vavilov_3d.png" title="Ice flow velocity on Vavilov" alt="Ice flow velocity on Vavilov" width="40%">
<img src="../../assets/images/vavilov_slice.png" title="Ice flow on Vavilov, cross section" alt="Ice flow on Vavilov, cross section" width="40%">
</center>
In the coming steps, we will refine the numerical resolution and activate the coupling to perform time-dependant simulations.