Hi,
I’m migrating my Telemac setup to a different cluster and I have noticed that using the same model, the memory usage is larger but the model runs faster.
This is seff’s output from the original cluster:
Nodes: 4
Cores per node: 24
CPU Utilized: 4-19:08:13
CPU Efficiency: 99.74% of 4-19:26:24 core-walltime
Job Wall-clock time: 01:12:09
Memory Utilized: 46.24 GB (estimated maximum)
Memory Efficiency: 19.59% of 236.00 GB (59.00 GB/node)
For the same model, on the new cluster, I’ve got:
Nodes: 4
Cores per node: 24
CPU Utilized: 2-01:29:18
CPU Efficiency: 99.03% of 2-01:58:24 core-walltime
Job Wall-clock time: 00:31:14
Memory Utilized: 95.99 GB (estimated maximum)
Memory Efficiency: 40.67% of 236.00 GB (59.00 GB/node)
What could be the reason for these changes? Could this affect the results from the model?
So far, I haven’t compared the models’ results since I was focused on runtime.
[name removed]