Learning Forum

[nir.levanon]

How are distributed HPC jobs to be run on multiple hosts submitted using LSF?

Is it possible to execute distributed simulation jobs without mpirun? 

If MPICH2 can be used to 'Run CPi distributed between two computers' , how can that be combined with submitting an LSF job via the bsub command?

Thanks,

Nir

[Lito Yap]

@nir.levanon,

Just request the number of nodes, number of tasks and CPUs per tasks on the cluster and run using the number of tasks you requested. 

Example LSF submission script:

!/bin/bash 
#BSUB -W 01:00              #1 hr wall time
#BSUB -n 64                 #64 processes/cores
#BSUB -R "span[ptile=4]"    #4 nodes
#BSUB -M 128000             #memory in MB =128gb 
#BSUB -J lumtest            #job name=lumtest
#BSUB -e %J.err             #job error log file
#BSUB -o %J.out             #job output log file
#change directory to the location of your simulation files
cd /path_to/simulationfiles
#run 1 FDTD simulation using the bundled MPICH2 (default install path shown) using all requested resources
/opt/lumerical/v221/mpich2/nemesis/bin/mpiexec -n 64 /opt/lumerical/v221/bin/fdtd-engine-mpich2nem -t 1 simulationfile.fsp

Run your submission script on the cluster.