CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   STAR-CCM+ (https://www.cfd-online.com/Forums/star-ccm/)
-   -   running star-ccm+ on more compute nodes (https://www.cfd-online.com/Forums/star-ccm/239422-running-star-ccm-more-compute-nodes.html)

pi120 November 4, 2021 11:41

running star-ccm+ on more compute nodes
 
Hi,

Is it possible to run star-ccm+ on a cluster with more than one compute node? If so, can you share the command that allows to do so?

At the moment I am running it with 1 compute node, 32 cores and 124gb of memory. What I'd like is to use for example 4 compute nodes, each with 32 cores and 124gb.

Here's the command I tried to use with no success:

starccm+ -batch -load -v -machinefile $PBS_NODEFILE -np #nodes*#cores path_to_file/file.sim

I'm quite ignorant on how to write these commands so very glad if you have any suggestion.

LuckyTran November 4, 2021 15:41

Is it possible? Absolutely. People do this day in and day out.

All you do in Star is properly list the names of the nodes in the file of machine names. If that file is correct and it still doesn't work, then you have a configuration issue either on your end or your cluster's end. Or you don't know how to type a proper path.

Make sure $PBS_NODEFILE points to the machine file and that machine file looks like it should.

wcj1n18 November 5, 2021 10:40

Hi, are you submitting jobs in batch? If so, I may be able to help, otherwise, not sure how you do this

pi120 November 8, 2021 11:42

Hi Will yes I am submitting in batch mode. Ever done something similar?

wcj1n18 November 8, 2021 12:54

Yes, I am doing something similar, so have some experience but not a huge amount. Are you using a Linux cluster? And are you using a remote computing software such as Mobaxterm? This is the case for me. I'm using the following script which i submitted to the computer cluster using SBATCH "SCRIPT_FILE_NAME"

#!/bin/bash
#Script to run a Starccm+ job
#set default resource requirements for job
#SBATCH --nodes=2
#SBATCH --time=01:00:00
#SBATCH --ntasks=64
#SBATCH --mail-type=ALL
#SBATCH --mail-user="insert_email_address"

#use starccm module to set up the environment for starccm+
module load starccm/15.06
module load openmpi/3.1.4/gcc

# Create our hosts file ala slurm
NODEFILE="$(pwd)/slurmhosts.$SLURM_JOB_ID.txt"
srun hostname -s &> $NODEFILE

#Surface Wrapper and Mesh Domain
starccm+ -np $SLURM_NTASKS -machinefile $NODEFILE -load simulation_file.sim -batch > output_file_name

Hopefully something similar works for you


All times are GMT -4. The time now is 04:58.