Skip to content

Latest commit

 

History

History
68 lines (41 loc) · 2.18 KB

README.md

File metadata and controls

68 lines (41 loc) · 2.18 KB

contributors

  • matteoalberti Matteo Alberti
  • davydenk Irina Davydenkova
  • ElenaDep Elena De Paoli
  • NataNazar Natalia Nazarova

This package contains simplified MD code with multi-threading parallelization for simulating atoms with a Lennard-Jones potential.

requirements:

python 3.6.5 openmpi version 3 or higher

Compiling and usage instructions:

!!! always make clean before anything

to use python:

  • make shared to create shared libraries for serial version
  • make shared_mpi to create shared libraries for mpi version
  • make shared_openmp to create shared libraries for openmp version
  • make shared_mpi_openmp to create shared libraries for mixed version

Then the go to the examples folder (or any subfolder where you place .inp files) and run pytho3 ../ljmd.py < INPUT_FILE to execute or mpirun -np N pytho3 ../ljmd.py < INPUT_FILE if you are using the mpi version

to use non-python version:

  • make - will create the simplest possible executable (currently the default target)
  • make mpi - the executable for shared_mpi
  • make openmpi - the executable with openmp enabled
  • make mpi_openmp - the executable with both mpi and openmp enabled

Then go to the examples folder (or any subfolder where you place .inp files) and run ../ljmd.x < INPUT_FILE to execute for non-mpi.

regression checks

!!! make clean before anything and make ONE OF THE ABOVE OPTIONS to create your libraries/executables

  • make check to check python (after make shared or make shared_openmp)
  • make mpi_check to check python (after make shared_mpi or make shared_mpi_openmp)
  • make check_c to check c code (after make or make openmp)
  • make mpi_check_c to check c mpi code (after make mpi or make mpi_openmp)

Reports

MPI Implementation

A complete report is available inside /mpi_report folder.

We want to report here just the Strong Scaling plots

OPENMP Implementation

A complete report is availabe inside openmp_report folder