On an inherited cluster that runs torque/moab scheduler, I find that the nodes:ppn requests are not properly honored. The nodes are allocated as per the value of ppn regardless of the value of node. For instance, nodes=8:ppn=16 will allocate only 4 nodes (each node has 32 cpu cores). I am looking to find possible configuration setting to remedy the situation.
Asked
Active
Viewed 84 times
4
Ketan Maheshwari
- 159
- 1
- 1
- 9