About this Abstract |
Meeting |
2024 TMS Annual Meeting & Exhibition
|
Symposium
|
Algorithm Development in Materials Science and Engineering
|
Presentation Title |
Enabling Materials Science Simulations with the Cabana Library |
Author(s) |
Sam Reeve, Kwitae Chong, John Coleman, Steve DeWitt, David Joy, Austin Isner, Matt Rolchigo, Pablo Seleson, Stuart Slattery, Jamie Stump |
On-Site Speaker (Planned) |
Sam Reeve |
Abstract Scope |
We present Cabana, a new performance portable library for building scientific particle applications. Cabana has been developed throughout the U.S. Department of Energy Exascale Computing Project to enable simulation on current and future exascale supercomputers, all the way down to individual laptops. Cabana implements particle (and structured grid) parallelism, data structures, algorithms, and communication, building on top of Kokkos and MPI. We will discuss how to create Cabana-based materials science applications, including mesh-free molecular dynamics (MD) and peridynamics; hybrid particle-mesh material point method (MPM); and mesh-based phase field, cellular automata, and finite difference heat transfer. Across each method we will highlight Cabana features, critical kernels for performance and scalability across system sizes, and discuss benchmark test problems. |
Proceedings Inclusion? |
Planned: |
Keywords |
Computational Materials Science & Engineering, Modeling and Simulation, |