![]() |
MAiNGO
|
There is also a parallelized version of MAiNGO, which allows the use of multiple processors via MPI. Its main use is for large problems where the B&B algorithm needs a long time to converge. The parallel version can be accessed and compiled the same way as the sequential version. In order to generate a project of the parallel MAiNGO version please set the CMake variable MAiNGO_use_mpi=true
and make sure you have an MPI library installed. Unfortunately, the parallel version is not yet supported when using MAiNGO from Python.
To run MAiNGO using MPI, type
where <mpiexec>
is the correct mpiexec executable of your MPI library.
It may happen that MPI asks you for an username and a password. If you type in a wrong password or a wrong username resulting in MPI not allowing you to proceed. In this case you can reset it via <mpiexec> -remove
.