Snapshot of a SeisSol simulation of the 2019 Ridgecrest Earthquake (extracted from https://www.youtube.com/watch?v=JvMuQN4FuvU).
Led by the Texas Advanced Computing Center, results of the Ridgecrest simulations were presented as an SC23 Scientific Visualization & Data Analytics Showcase at SC23: The International Conference for High Performance Computing, Networking, Storage, and Analysis.
Dynamic rupture simulation of the 2018 Palu, Sulawesi earthquake and tsunami (Krenz et al., 2021); the respective SC21 paper also motivated the scenario for the Mystery Application for the SC22 Student Cluster Competition).
SeisSol is a high-performance computational seismology software to simulate complex earthquake scenarios. It supports various rheologies (elastic, anisotropic-elastic, acoustic, viscoelastic, poroelastic), boundary conditions and dynamic rupture laws. SeisSol uses high-order discontinuous Galerkin discretisation with an Arbitrary DERivative (ADER) local time stepping on unstructured adaptive tetrahedral meshes.
SeisSol follows an MPI+X parallelisation, supporting OpenMP and CUDA (SYCL upcoming) for heterogeneous compute nodes, and exhibits excellent performance and scalability on a suitable range of supercomputing platforms.
Per-node performance on SuperMUC (M), Shaheen II (S), and Cori (C).
High parallel efficiency and fast time-to solution using local time-stepping (LTS).
Meet the cheesequake challenge (video on YouTube)
SeisSol offers example workflows based on the Gmsh and Simmetrix mesh generation packages. See the SeisSol Tutorials and Cookbook for examples.