OceanHPCFrame is a parallel framework for accelerating ocean simulation models. It includes four progressively optimized versions to evaluate the performance impact of vectorization, multithreading, and multi-node parallelism.
Use the make command to compile each version:
make v1: Baseline implementationmake v2: Vectorized versionmake v3: Multi-threaded versionmake v4: Fully optimized version
Use the corresponding script to run:
./run_v1:./run_v2:./run_v3:./run_v4 <n>
- Simulation parameters are configured in
nwamctl.ini, including grid resolution (GRDSZX,GRDSZY) - Input topography data is provided in
etop5.nc
This project was tested on the LS Pilot System with the following software:
- MPI v4.1.6
- netCDF v4.9.2
- Clang v17.0.6
- Flang v17.0.6
- Sdma v1.0
- Memkind v1.0
Please make sure your environment includes these or compatible versions for execution.
Feel free to open an issue if you have any questions.