SPH

CUDA and OpenMP versions of SPH (Smoothed Particle Hydrodynamics) serial algorithm.

Dependencies

The GCC compiler is used to build the code.

Build

For this project are available:

• 4 OpenMP parallelized versions
• 2 CUDA parallelized versions.

Note

The make command must be run in the src folder.

OpenMP versions

To build all OpenMP versions: make openmp

To build them indipendently:

    gcc -std=c99 -fopenmp -Wall -Wpedantic omp-sph.c -o omp-sph -lm
    gcc -std=c99 -fopenmp -Wall -Wpedantic omp-sphv2.c -o omp-sph -lm
    gcc -std=c99 -fopenmp -Wall -Wpedantic omp-sphv3.c -o omp-sph -lm
    gcc -std=c99 -fopenmp -Wall -Wpedantic omp-sphv4.c -o omp-sph -lm

CUDA versions

To build all CUDA versions: make cuda

In alternativa:

    nvcc cuda-sph.cu -o cuda-sph
    nvcc cuda-sphv2.cu -o cuda-sphv2

Results

See the Report.

Authors

Marco Sangiorgi
2023©