[trans(sonic)_pride]

Dear readers,


I have been running OpenFoam simulations for some time on my cluster, and I have now just realised it might be operating under its capabilities. On my nodes, I set up the CPUs number on slurm to match the number of threads of the machine, since slurm says that it is one of the valid configurations, so the slurm.conf file shows:
"CPUs=128 Boards=1 SocketsPerBoard=2 CoresPerSocket=32 ThreadsPerCore=2"
I am now thinking that every time I submit a job, it might be using multi-threading on the cores, as the number of CPUs on slurm is equal to the number of threads instead of the number of physical cores. I recall (from various sources) that using multi-threading is detrimental to simulation performance, so, the question is: is my slurm config optimal? if not, how can I fix it? my systems are EPYC Rome based.


Thank you