TERRA-NEO - Integrated Co-Design of an Exascale Earth Mantle Modeling Framework

点击翻译按钮获取中文摘要

Much of what one refers to as geological activity of the Earth is due to the fact that heat is transported from the interior of our planet to the surface in a planetwide solid-state convection of the Earth's mantle. Linking mantle processes to their surface manifestations is seen widely today as one of the most fundamental problems in the Earth sciences, while being at the same time a matter of direct practical relevance through their impact on the evolution of sedimentary basins and their paramount economical importance.The sheer magnitude of the computational challenge of earth mantle simulations on modern super- computer architectures can be grasped easily from the dramatic growth of disciplinary complexity. Dominating roadblocks in Geophysics are model complexity and uncertainty in parameters and data, e.g., rheology and seismically imaged mantle heterogeneity, as well as the enormous space and time scales one must resolve. In Algorithmics, the traditional view of optimality is more and more obsolete and must be replaced by new HPC performance oriented metrics. In Exascale Computing, the disruptive transition from modest concurrency to billions of threads leaves the software devel- opment trailing behind. This consortium has already demonstrated the potential of interdisciplinary research by a series of joint publications, see Part B Subsection 1.1, and is fully committed to the cross-disciplinary collaboration that is necessary for creating TERRA-NEO as new community-usable, sustainable exascale simulation framework.The new community code TERRA-NEO will be based on a carefully designed multi-scale finite ele- ment discretization using an icosahedral mesh with block-wise refinement for the non-linear trans- port processes in the Earth mantle. This permits the construction of exascale solvers with maximal scalability and optimal efficiency. We have already demonstrated the computation of flow fields on petascale supercomputers with a resolution of globally 1km that would have been unthinkable until recently since it requires the solution of indefinite systems of more than 10 to the 12 unknowns per time step.Advancing to exascale, we will develop communication-avoiding, asynchronous solution techniques that alleviate the traditional over-synchronization of hierarchical iterative methods, enabling the solution of inverse problems and the quantification of uncertainties. Advanced resiliency techniques will be supported on the algorithmic level. TERRA-NEO will be developed specifically for the upcoming heterogeneous exascale computers in an architecture-aware design process that is guided by performance models, leading to a holistic co-design of the data structures and algorithms.