Essentially, there are two compute clusters that can be chosen to run my program, one used for students and one for researchers. The code I am working on is used to diagnose experiment results, so I need to be able to work with both. The grad cluster uses slurm and the research cluster uses qsub.
First Problem:
The grad cluster is not uniform in its make up, some of the nodes have vastly different processors. For my particular use I need Fuse Multiply Add. How do you request cpus with certain properties/instructions in slurm (I don't need this for qsub but it would be useful to know)
Second Problem:
The script I am using to test my program currently calls a command (pytest) which is set up to require the input of the number of nodes and cpus being used. I would like to be able to get the number of nodes and the number of cores each node has so that the use doesn't have to edit the script when they run a test. How can I get this info from qsub and from slurm? Currently, I do not have a work around for slurm and for qsub I have a python script that gets a line containing the info which is parsed for the info.
Thanks!