I'm trying to run large Abaqus jobs on a Linux cluster using parallelization.
I'm running into an issue where if I set (for example) my Abaqus job to use 80 cores, and submit a batch file requesting 2x nodes, with 40x cores each, the job will run on only 1x node and not utilize both. It will successfully run, but only using the 1x core (making it too slow). I have the MPI module loaded.
Do you have any ideas or suggestions? Maybe you've seen this before? I
Let me know what you think, any help is appreciated.
I'm running into an issue where if I set (for example) my Abaqus job to use 80 cores, and submit a batch file requesting 2x nodes, with 40x cores each, the job will run on only 1x node and not utilize both. It will successfully run, but only using the 1x core (making it too slow). I have the MPI module loaded.
Do you have any ideas or suggestions? Maybe you've seen this before? I
Let me know what you think, any help is appreciated.
