The Netherlands eScience Center and Atos-Bull have granted four proposals within Service 1 of the EU-funded project ‘Center of Excellence in Simulation of Weather and Climate in Europe’ (ESiWACE2).
The ESiWACE2 Services enable short collaborative projects that offer guidance, engineering, and advice by the ESiWACE2 consortium partners to support exascale preparations for weather and climate models in Europe. Service 1 targets model developing groups and assists with porting the codes to new hardware platforms, such as GPUs.
After a year of successful collaborations in accelerating atmospheric and sea ice-ocean modelling with scientists, we are excited to continue our efforts to prepare the weather and climate modelling communities in Europe for the upcoming exascale revolution.
These are the four awarded projects:
Wageningen University – Chiel van Heerwaarden
RTE+RRTMGP C++ (rte-rrtmgp-cpp)
RTE+RRTMGP (Radiative Transfer for Energetics + Rapid Radiative Transfer Model for GCMs – Parallel) is a set of codes for computing radiative fluxes in planetary atmospheres. The goal of this project is to bring the C++/CUDA implementation of those codes to maximum performance, using the Kernel Tuner as developed by the Netherlands eScience Center, and integrate machine learning-based replacements for part of the radiation parametrizations.
Accurate radiative transfer modelling is essential for credible climate simulations and for weather forecasts, in particular those made for solar energy production.
The ambition is to build the fastest available general purpose radiative transfer solver for climate, weather, and large-eddy simulation codes. Radiative transfer models are a significant computational bottleneck, and the aim is to alleviate this. The project will serve as a stepping stone to more general and advanced machine learning-based parametrizations.
Alfred Wegener Institute – Natalja Rakowsky
Finite-volumE Sea ice Ocean Model, Version 2.0 (FESOM2)
FESOM2 is a global sea-ice ocean circulation model based on unstructured meshes. It allows one to simulate the global ice-ocean system at extremely high resolution in the regions of interest at an affordable computational cost. The broad spectra of FESOM2 applications includes several climate models and standalone sea ice-ocean configurations (including 1 km Arctic configuration and very high resolution global meshes).
The aim is to consolidate the advances made in the current ESiWACE Service 1 project for FESOM2. The port of the tracer transport module should be finalised with CUDA, and an OpenACC version should follow. Both approaches will be benchmarked on current architectures. An assessment of the cost of the implementation in relation to acceleration and portability will guide the decision which of the approaches will be implemented in the main branch of FESOM2.
Delft University of Technology – Fredrik Jansson
DALES – the Dutch Atmospheric Large Eddy Simulation
DALES is a large-eddy simulation code designed for studies of the physics of the atmospheric boundary layer. DALES, for example, is used in the Ruisdael project to perform high-resolution simulations over the Netherlands of all relevant atmospheric processes, such as clouds, convection, precipitation and radiation, and their effects on air quality and transport of emitted greenhouse gases.
The main goals are to improve the scaling of DALES to many nodes (scaling within a single node is already reasonably good, but MPI over network is a bottleneck for good scaling); and to improve single-threaded performance through more cache-friendly data-access patterns, vectorization, exploring switching from double to single precision calculations, and improved numerical algorithms.
Inria – Laurent Debreu
AGRIF: Adaptive Grid Refinement In Fortran
AGRIF is the tool proposed to the ocean modelling community to solve a variety of problems in which the local resolution (e.g. 1 km) is central. Its use could benefit to any application other than pure academic test cases with simplified geometry. AGRIF will specifically be useful in global and expensive simulations used for example in operational oceanography (for shelf seas and fractal coast geometry) or in climate research simulations, where nominal resolution is limited by the complexity of the earth system models and the length of the simulated period (centuries).
The goal of this collaboration is to establish a comprehensive performance profile of AGRIF high-resolution nested models in simulations of the NEMO ocean model to quantify the performance impact of the library on NEMO simulations, identify bottlenecks, and propose solutions to alleviate them, including the potential use of GPUs.
For more information visit the ESiWACE2 website.
ESiWACE2 has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 823988.