3.4. Basic build options

The following topics are covered on this page, for building with both CMake and make:

3.4.1. Serial vs parallel build

LAMMPS is written to use the ubiquitous MPI (Message Passing Interface) library API for distributed memory parallel computation. You need to have such a library installed for building and running LAMMPS in parallel using a domain decomposition parallelization. It is compatible with the MPI standard version 2.x and later. LAMMPS can also be built into a “serial” executable for use with a single processor using the bundled MPI STUBS library.

Independent of the distributed memory MPI parallelization, parts of LAMMPS are also written with support for shared memory parallelization using the OpenMP threading standard. A more detailed discussion of that is below.

-D BUILD_MPI=value        # yes or no, default is yes if CMake finds MPI, else no
-D BUILD_OMP=value        # yes or no, default is yes if a compatible compiler is detected
-D LAMMPS_MACHINE=name    # name = mpi, serial, mybox, titan, laptop, etc
                          # no default value

The executable created by CMake (after running make) is named lmp unless the LAMMPS_MACHINE option is set. When setting LAMMPS_MACHINE=name the executable will be called lmp_name. Using BUILD_MPI=no will enforce building a serial executable using the MPI STUBS library.

MPI and OpenMP support in LAMMPS

If you are installing MPI yourself to build a parallel LAMMPS executable, we recommend either MPICH or OpenMPI which are regularly used and tested with LAMMPS by the LAMMPS developers. MPICH can be downloaded from the MPICH home page and OpenMPI can be downloaded correspondingly from the OpenMPI home page. Other MPI packages should also work. No specific vendor provided and standard compliant MPI library is currently known to be incompatible with LAMMPS. If you are running on a large parallel machine, your system admins or the vendor should have already installed a version of MPI, which is likely to be faster than a self-installed MPICH or OpenMPI, so you should study the provided documentation to find out how to build and link with it.

The majority of OpenMP (threading) support in LAMMPS is provided by the USER-OMP package; see the USER-OMP package page for details. The USER-INTEL package also includes OpenMP threading (it is compatible with USER-OMP and will usually fall back on styles from that package, if a USER-INTEL does not exist) and adds vectorization support when compiled with compatible compilers, in particular the Intel compilers on top of OpenMP. Also, the KOKKOS package can be compiled to include OpenMP threading.

In addition, there are a few commands in LAMMPS that have native OpenMP support included as well. These are commands in the MPIIO, SNAP, USER-DIFFRACTION, and USER-DPD packages. In addition some packages support OpenMP threading indirectly through the libraries they interface to: e.g. LATTE, KSPACE, and USER-COLVARS. See the Packages details page for more info on these packages and the pages for their respective commands for OpenMP threading info.

For CMake, if you use BUILD_OMP=yes, you can use these packages and turn on their native OpenMP support and turn on their native OpenMP support at run time, by setting the OMP_NUM_THREADS environment variable before you launch LAMMPS.

For building via conventional make, the CCFLAGS and LINKFLAGS variables in Makefile.machine need to include the compiler flag that enables OpenMP. For GNU compilers it is -fopenmp. For (recent) Intel compilers it is -qopenmp. If you are using a different compiler, please refer to its documentation.

OpenMP Compiler compatibility

Some compilers do not fully support the default(none) directive and others (e.g. GCC version 9 and beyond, Clang version 10 and later) may implement strict OpenMP 4.0 and later semantics, which are incompatible with the OpenMP 3.1 semantics used in LAMMPS for maximal compatibility with compiler versions in use. If compilation with OpenMP enabled fails because of your compiler requiring strict OpenMP 4.0 semantics, you can change the behavior by adding -D LAMMPS_OMP_COMPAT=4 to the LMP_INC variable in your makefile, or add it to the command line while configuring with CMake. LAMMPS will auto-detect a suitable setting for most GNU, Clang, and Intel compilers.

3.4.3. Build the LAMMPS executable and library

LAMMPS is always built as a library of C++ classes plus an executable. The executable is a simple main() function that sets up MPI and then creates a LAMMPS class instance from the LAMMPS library, which will then process commands provided via a file or from the console input. The LAMMPS library can also be called from another application or a scripting language. See the Howto couple doc page for more info on coupling LAMMPS to other codes. See the Python page for more info on wrapping and running LAMMPS from Python via its library interface.

For CMake builds, you can select through setting CMake variables between building a shared or a static LAMMPS library and what kind of suffix is added to them (in case you want to concurrently install multiple variants of binaries with different settings). If none are set, defaults are applied.

-D BUILD_SHARED_LIBS=value   # yes or no (default)
-D LAMMPS_MACHINE=name       # name = mpi, serial, mybox, titan, laptop, etc
                             # no default value

The compilation will always produce a LAMMPS library and an executable linked to it. By default this will be a static library named liblammps.a and an executable named lmp Setting BUILD_SHARED_LIBS=yes will instead produce a shared library called liblammps.so (or liblammps.dylib or liblammps.dll depending on the platform) If LAMMPS_MACHINE=name is set in addition, the name of the generated libraries will be changed to either liblammps_name.a or liblammps_name.so, respectively and the executable will be called lmp_name.

Additional information

Note that for creating a shared library, all the libraries it depends on must be compiled to be compatible with shared libraries. This should be the case for libraries included with LAMMPS, such as the dummy MPI library in src/STUBS or any package libraries in the lib directory, since they are always built in a shared library compatible way using the -fPIC compiler switch. However, if an auxiliary library (like MPI or FFTW) does not exist as a compatible format, the shared library linking step may generate an error. This means you will need to install a compatible version of the auxiliary library. The build instructions for that library should tell you how to do this.

As an example, here is how to build and install the MPICH library, a popular open-source version of MPI, as a shared library in the default /usr/local/lib location:

./configure --enable-shared
make install

You may need to use sudo make install in place of the last line if you do not have write privileges for /usr/local/lib or use the --prefix configuration option to select an installation folder, where you do have write access. The end result should be the file /usr/local/lib/libmpich.so. On many Linux installations the folder ${HOME}/.local is an alternative to using /usr/local and does not require superuser or sudo access. In that case the configuration step becomes:

./configure --enable-shared --prefix=${HOME}/.local

Avoiding to use “sudo” for custom software installation (i.e. from source and not through a package manager tool provided by the OS) is generally recommended to ensure the integrity of the system software installation.

3.4.4. Including or removing debug support

By default the compilation settings will include the -g flag which instructs the compiler to include debug information (e.g. which line of source code a particular instruction correspond to). This can be extremely useful in case LAMMPS crashes and can help to provide crucial information in tracking down the origin of a crash and help the LAMMPS developers fix bugs in the source code. However, this increases the storage requirements for object files, libraries, and the executable 3-5 fold.

If this is a concern, you can change the compilation settings or remove the debug information from the LAMMPS executable:

  • Traditional make: edit your Makefile.<machine> to remove the -g flag from the CCFLAGS and LINKFLAGS definitions

  • CMake: use -D CMAKE_BUILD_TYPE=Release or explicitly reset the applicable compiler flags (best done using the text mode or graphical user interface).

  • Remove debug info: If you are only concerned about the executable being too large, you can use the strip tool (e.g. strip lmp_serial) to remove the debug information from the executable file. Do not strip libraries or object files, as that will render them unusable.

3.4.5. Build LAMMPS tools

Some tools described in Auxiliary tools can be built directly using CMake or Make.

-D BUILD_TOOLS=value         # yes or no (default)
-D BUILD_LAMMPS_SHELL=value  # yes or no (default)

The generated binaries will also become part of the LAMMPS installation (see below).

3.4.6. Install LAMMPS after a build

After building LAMMPS, you may wish to copy the LAMMPS executable of library, along with other LAMMPS files (library header, doc files) to a globally visible place on your system, for others to access. Note that you may need super-user privileges (e.g. sudo) if the directory you want to copy files to is protected.

cmake -D CMAKE_INSTALL_PREFIX=path [options ...] ../cmake
make                        # perform make after CMake command
make install                # perform the installation into prefix

During the installation process CMake will by default remove any runtime path settings for loading shared libraries. Because of this you may have to set or modify the LD_LIBRARY_PATH (or DYLD_LIBRARY_PATH) environment variable, if you are installing LAMMPS into a non-system location and/or are linking to libraries in a non-system location that depend on such runtime path settings. As an alternative you may set the CMake variable LAMMPS_INSTALL_RPATH to on and then the runtime paths for any linked shared libraries and the library installation folder for the LAMMPS library will be embedded and thus the requirement to set environment variables is avoided. The off setting is usually preferred for packaged binaries or when setting up environment modules, the on setting is more convenient for installing software into a non-system or personal folder.