Understanding cloud and precipitation processes is very crucial in order to reduce significantly the uncertainty of climate change projections. Within the HD(CP)2 project an LES model based on ICON (ICOsahedral Non hydrostatic GCM) has been developed. The ICON-LES model aims at resolving cloud and precipitation processes using grids with a resolution of 10000x10000x400 grid elements and a grid spacing of 100m. Such simulations are computationally and data intensive. For this it is vital to be able to exploit the hardware resources of Exascale-HPC systems in an optimal way. Current and future HPC systems are massively parallel computers consisting of hundreds of thousands of cores and a good scaling behaviour of ICON is the key to use such architectures efficiently. Furthermore, taking into account the time needed for writing the simulation results out on the file system (time to solution), makes the scalability of the model a big challange.
In order to address this issue a major refacotring of the code has been undertaken. Thereby, all the global fields were substituted with distributed data structures and the corresponding algorithms were parallelized. In this talk the results of the ongoing efforts towards a high scaling ICON model, which is able of running experiments with a grid resolution of 120m on up to 458752 cores, will be presented.