Within ESiWACE3, Key Exploitable Results (KERs) include a portfolio of software tools, model codes, and applications developed to advance performance assessment and optimisation for Earth system modelling. These results support researchers and practitioners in analysing computational efficiency, testing scalability, and improving portability across emerging HPC architectures.
This section highlights these KERs and the results they generate, showcasing practical solutions built around realistic climate and weather workloads. By focusing on representative scientific use cases rather than synthetic metrics such as High Performance Linpack (HPL), they provide actionable insight into how computing systems perform for real modelling and forecasting applications.
HPCW
High Performance Climate and Weather benchmarking suite. HPCW isolates key elements in the workflow of weather and climate prediction systems to improve performance and to allow a detailed performance comparison for different hardware platforms thus fostering co-design with vendors and technology providers.
AUTO-RPE
AutoRPE is an automated framework designed to implement mixed-precision configurations in large-scale Fortran models by emulating reduced numerical precision through custom data types. It allows researchers to significantly improve computational efficiency and reduce memory usage by identifying and optimizing code segments that do not strictly require double-precision arithmetic.
Automatic Performance Profiling (APP)
Automatic Performance Profiling (APP) provides an automated workflow for profiling Earth System Models, generating detailed performance reports that help identify bottlenecks and improve scalability and efficiency on HPC systems.
Docker EC-EARTH
Integration of the EC-Earth container with Autosubmit
Autosubmit is a lightweight workflow manager designed to meet climate research necessities. It integrates the capabilities of an experiment manager, workflow orchestrator and monitor in a self-contained application. Integrating the EC-Earth container with Autosubmit provides a highly portable solution for performing climate simulations on different HPC systems while also simplifying the management and control of the entire simulation workflow."
NEMO on GPU
porting / integration on MN5 and LUMI
FDO4Climate: Initial Evaluation
Evaluation of the readiness of published climate simulation output for automated analysis. High potential because automated analysability of climate data has not been evaluate before and because specifications to enhance/enable machine actionability can be deduced from this work.
Field Compression Library
The field compression benchmark provides a highly-scalable environment to benchmark and analyse the performance of different data compression methods on a large and diverse set of weather and climate datasets. The Field Compression Benchmark Suite laid the technical groundwork for the ClimateBenchPress benchmark project, by Tim Reichelt (Embed2Scale) with Juniper Tyree (University of Helsinki-ESiW3), Peter Dueben (ECMWF-ESiW3) and Sara Faghih-Naini (ECMWF -ESiW3) as co-authors.
PSyclone
Use of PSyclone in the NEMO build system. PSyclone is also used by the official NEMO release and it is part of the NEMO build system. According to partner: STFC, Crown, Met Office, NERC and CMCC (as a contributor)
FVM
FVM is developed at ECMWF as a new dynamical core for weather and climate simulations. The model can simulate global or local domains at various levels of resolution from hectometric limited area simulations to km-scale global simulations. The work in ESiWACE is focussing on the porting of FVM to GPU hardware and heteoregeneous hardware via the GT4Py domain specific language.
Containerization of EC-Earth
Containerisation is one potential approach to port EC-Earth to new systems, which includes pre-exascale systems. Given that access to tier 0 systems usually comes with severe time constraints and strong focus on actual production runs, container based porting combined with the use of the Autosubmit workflow manager can help to better utilise the assigned resources.
Online Laboratory for Climate Science and Meteorology
The Online Laboratory for Climate Science and Meteorology provides a JupyterLab-like environment that has common climate and meteorology packages and libraries preinstalled. Provides researchers an installation-free quick-to-launch Jupyter environment in their web browser (runs entirely inside the user’s webbrowser using WebAssembly) that can be used for running small exper.
Compression Laboratory
The (field) compression laboratory builds on ECMWF’s field compression library. It provides several Jupyter notebooks that explore data compression for weather and climate data. The notebooks can be run locally or in the Online Laboratory.
Kernel Tuner Ecosystem
The Kernel Tuner Ecosystem is a suite of tools and libraries aimed at maximizing the speed and energy efficiency of high-performance GPU software.