Re-Implementation of the Shallow Water Solver LANL/CLAMR using Kokkos, Cabana, and Cajita.
- Properly functions using
- Serial
- OpenMP
- Cuda
- Serial + MPI
- OpenMP + MPI
- Cuda + MPI
- Able to write to Silo files using PMPIO (Tested on Mac in Serial, OpenMP, Serial + MPI, OpenMP + MPI)
- Can Visualize results using VisIt
- Fix Cuda MPI issue in use of Cajita Halo gather (currently functioning with a work around using CudaUVM and a custom halo exchange). It might be an issue with how ExaCLAMR is using the Cajita gather?
- Get Silo to build on Wheeler and Xena and get it working with the CudaUVM case
- Template the mesh and problem manager classes for regular grids, AMR grids, and space-filling curves (see Jered branch)
- 2-D Hilbert Layout
- Investigate using Cabana AoSoA
- Add particle physics
git clone https://github.com/CUP-ECS/ExaCLAMR.git
cd ExaCLAMR
bash scripts/build_wheeler.sh
mkdir -p data
mkdir -p data/raw
- Note: You have to build on a compute node
git clone https://github.com/CUP-ECS/ExaCLAMR.git
cd ExaCLAMR
bash scripts/build_xena.sh
mkdir -p data
mkdir -p data/raw