Global Cloud Resolving Atmospheric Modeling on Exascale Computers
Mark Taylor, Sandia National Laboratories
I will first give an overview of modern Earth system models and how they are used to study the Earth's climate. I will then describe the DOE's Energy Exascale Earth System Model (E3SM) project, with a focus on Sandia's role in developing the atmosphere component model. The atmospheric model solves the differential equations of motion in the Earth's atmosphere over the entire globe, which is then coupled to a suite of parameterizations modeling the many unresolved processes. I'll go through several of the numerical design choices made in global atmospheric models including choice of grids, equation formulation and discretization methods.
The E3SM Project's computational mission is to ensure the model can make efficient use of DOE's present and future Exascale systems. To support these new architectures, E3SM is rewriting much of the existing Fortran code in C++ with the Kokkos library used for on-node parallelism. We have found the C++/Kokkos approach to be more robust and better supported then a Fortran+directives based approach across several different GPUs (NVIDIA, AMD and Intel). I'll give some highlights from this work, which include record setting performance of E3SM's "cloud resolving" model running on Frontier which is currently ranked #1 in the Top500 list of supercomputers.