I suggest approaching this from two aspects:
1.Check whether the filename.inp can run normally on other computers.
2.Clusters typically distinguish between management nodes (running the graphical interface) and compute nodes (dedicated to heavy computations). If you directly start calculations via the graphical interface on the management node, the tasks will execute on the management node itself, which usually has limited resources and poorer performance. To optimize resource usage, submit computational tasks to compute nodes using cluster management tools (e.g., job schedulers like SLURM or PBS). This requires coordination with your cluster administrator to ensure proper configuration and access permissions.