bug with high num_threads (1/2): simulation not deterministic anymore
I've encountered severe issues, when increasing num_threads to higher orders e.g. 24. An ini-file with no changes at all including constant seed does not yield reproducible results. See screenshot.
@Arne @OSchmidts @chraibi Does someone of you have an idea about that? FYI: The Hermes Cluster is available again and could be used for reproducing that. I've attached exemplary files in #199 (closed).