Integration with Platform's Load Sharing Facility (LSF)
The Load Sharing Facility (LSF) scheduler is only supported on Linux. Jobs may be submitted in any of the following ways:
- Job Submission GUI
- Using LSF commands (qsub, etc.)
See the Ansys Electromagnetics Suite 2024 R2 Linux Installation Guide for additional information on supported schedulers.
If a temp directory is setup by the LSF cluster administrator, analysis engines use this temp directory, overriding the setting in the Ansys EM product.
LSF Job Management
You can use Ansys Electronics Desktop to submit batch jobs to LSF and monitor those jobs.
This involves the following steps:
- Use Tools > Job Management > Select Scheduler to select LSF as the scheduler.
- Use Tools > Job Management > Submit Job to submit a batch job to LSF.
LSF-specific Settings:
- On the Compute Resources tab, click the Resource Selection Parameters ellipses button (...) to specify the following:
- Queue – a drop-down menu lets you select Normal, chkpn_rerun_queue, idle, license, night, normal_allow_excl, owners, priorituy, or short.
- MinCoresPerNode – the minimum number of cores allowed on a node to be eligible for selection; translates to bsub -R select[ncpus>=N]
- MaxCoresPerNode – the maximum number of cores allowed on a node to be eligible for selection; translates to bsub -R select[ncpus<=N]
- MinMemoryPerNode – the minimum amount of physical memory (specified in integer GigaBytes) allowed on a node to be eligible for selection; translates to bsub -R select[maxmem>=M]
- MaxMemoryPerNode – the maximum amount of physical memory (specified in integer GigaBytes) allowed on a node to be eligible for selection; translates to bsub -R select[maxmem<=M]
If you do not specify parameters, the scheduler does so.
For LSF, the only non-automatic method of Resource Selection is Number of Tasks and Cores. You can specify the number of tasks, whether they are for exclusive use by the job, cores per distributed task,
- On the Compute Resources tab, click the Resource Selection Parameters ellipses button (...) to specify the following:
- Use Tools > Job Management > Monitor Job to monitor the job's progress.
LSF Job Submission Guidelines
Before submitting an LSF job, ensure the following are true:
- The project is available in a shared drive that is accessible to all machines in the cluster.
- The project is available using the same path on all machines of cluster.
- There is sufficient space in the project directory and temp directories.
- There is sufficient memory per engine.
- The number of compute resources (Distributed Analysis machines and Multi Processing cores) will achieve the desired scale factor and effective resource utilization.