This branch was developed because of the following issue: differences in architecture on different nodes of the supercomputer requires the FMM C++ code to be compiled on the node in question. This becomes annoying when attempting to run jobs with sbatch, as the user does not always know which node with the desired memory requirements will be next available.
A build_tmp.sh script is provided that allows the user to build FLOWExaFMM and store the shared library with a random tag, making it possible to rebuild FLOWExaFMM many times for running jobs and various nodes simultaneously. If the environment variable FMM is set, FLOWExaFMM uses the shared library at the path specified by FMM. Otherwise, it defaults to the old behavior.
This branch was developed because of the following issue: differences in architecture on different nodes of the supercomputer requires the FMM C++ code to be compiled on the node in question. This becomes annoying when attempting to run jobs with sbatch, as the user does not always know which node with the desired memory requirements will be next available.
A build_tmp.sh script is provided that allows the user to build FLOWExaFMM and store the shared library with a random tag, making it possible to rebuild FLOWExaFMM many times for running jobs and various nodes simultaneously. If the environment variable FMM is set, FLOWExaFMM uses the shared library at the path specified by FMM. Otherwise, it defaults to the old behavior.