This use case tests the scalability and performance of RI-SCALE’s Data Exploitation Platform (DEP) on EuroHPC systems using large datasets and AI models from Destination Earth (DestinE). It aims to ensure that DEPs can support compute-intensive research at exascale, helping Research Infrastructures integrate with EuroHPC and DestinE for next-generation digital twin and AI applications.
DEP Scalability on EuroHPC with DestinE
Challenge
As the Destination Earth (DestinE) initiative continues to expand, both the volume of data and complexity of AI models involved are growing rapidly. At the same time, the EuroHPC ecosystem is offering increasing computational power to support large-scale AI and data analytics. To ensure that the Data Exploitation Platform (DEP) can keep pace, its performance and scalability must be tested under the demanding conditions that DestinE and EuroHPC are expected to present.
Approach
-
Deploy the full DEP technology stack on multiple (pre-)exascale EuroHPC machines through the DestinE allocation
-
Use the AIFS machine learning model and digital twin datasets from DestinE
-
Access data via the DestinE DataBridge, a cloud stack co-located with EuroHPC systems
-
Measure scalability and performance under real workload scenarios
-
Optimise the DEP’s handling of compute-intensive AI applications
Goal
To validate the ability of the DEP to operate efficiently at scale within a EuroHPC environment, using large datasets and AI models provided by Destination Earth.
Relevance
-
Research Infrastructures (RIs) that want to integrate or re-use data from Destination Earth
-
RIs aiming to deploy or scale their DEPs on EuroHPC systems for large AI/ML workloads
-
Stakeholders interested in AI-enabled Earth system modelling and the next generation of digital twins.
Impact
This use case will demonstrate how RI-SCALE’s DEP technology can scale to meet the requirements of high-end AI applications and massive data volumes. It will offer valuable insights to RIs planning to connect with DestinE or to leverage EuroHPC infrastructures for their research, contributing to the broader uptake of AI in environmental and Earth system science.
Partners Involved
