USPEX + VASP: All Jobs Running on One Node Despite Multiple Nodes Allocated

[Tahir Wahab]

Dear all,

I am using USPEX_2023.0.2 with VASP_6.5.0 with a SLURM job asking for 4 nodes. During the run, USPEX calls 4 VASP calculations at the same time, each of them with the command line 

mpirun -np XX vasp_std_6.5.0

where XX is the number of cores on the cluster for one node. The problem is that those calculations are executed in the same node where USPEX is running, while the other node remains empty. My question is, is there any command-line option for mpirun or any other command alternative to mpirun, like srun, to have a proper distribution of the VASP calculations over the allocated nodes?


Thanks
Tahir