Closed MikeAxtell closed 9 years ago
Performance of 3.0 still needs testing on HPC clusters to see if this has been solved.
Per our testing at least on Penn State's HPC compute clusters, we don't see this issue with 3.0, 3.1, or 3.2 (at present 3.2 is latest release), so I am closing this issue.
Correspondence with user found that the same run can take more than a week on a cluster, just takes a few hours on a laptop. Likely because the speed at which certain cluster configurations can read and write to the users working space can be slow. Users should be advised in future FAQ that, for most jobs, running on a cluster is not necessary and can in some cases really slow things down.