Differences
This shows you the differences between two versions of the page.
| Both sides previous revision Previous revision Next revision | Previous revisionLast revisionBoth sides next revision | ||
| doku:ansys [2024/04/25 08:12] – [Single node script] amelic | doku:ansys [2024/06/17 09:21] – [Multiple node script] amelic | ||
|---|---|---|---|
| Line 72: | Line 72: | ||
| ===== Job script ===== | ===== Job script ===== | ||
| + | When executing Slurm jobs for Fluent, it's essential to consider specific parameters within the command, including: | ||
| + | |||
| + | * 2d: Indicates a job for 2D Fluent simulations. | ||
| + | * 2ddp: Denotes 2D Fluent jobs in double precision. | ||
| + | * 3d: Specifies a job for 3D Fluent simulations. | ||
| + | * 3ddp: Represents 3D Fluent jobs configured for double precision. | ||
| + | |||
| + | These parameters play a crucial role in defining the type and precision of the Fluent simulations run through Slurm. | ||
| + | |||
| ==== Single node script ==== | ==== Single node script ==== | ||
| - | When executing Slurm jobs for Fluent, it's essential to consider specific parameters within the command, including: | ||
| - | |||
| - | 2d: Indicates a job for 2D Fluent simulations. | ||
| - | 2ddp: Denotes 2D Fluent jobs in double precision. | ||
| - | 3d: Specifies a job for 3D Fluent simulations. | ||
| - | 3ddp: Represents 3D Fluent jobs configured for double precision. | ||
| - | These parameters play a crucial role in defining the type and precision of the Fluent simulations run through Slurm. | ||
| A script for running Ansys/ | A script for running Ansys/ | ||
| Line 102: | Line 104: | ||
| </ | </ | ||
| - | ==== Multiple node script ==== | ||
| - | A script for running Ansys/ | ||
| - | |||
| - | < | ||
| - | #!/bin/sh | ||
| - | #SBATCH -J fluent_1 # Job name | ||
| - | #SBATCH -N 2 # Number of nodes | ||
| - | #SBATCH --partition=mem_0096 # mem_0096, mem_0384, mem_0768 | ||
| - | #SBATCH --qos=mem_0096 | ||
| - | #SBATCH --ntasks-per-node=48 # number of cores | ||
| - | #SBATCH --threads-per-core=1 # disable hyperthreading | ||
| - | |||
| - | |||
| - | module purge | ||
| - | module load ANSYS/ | ||
| - | |||
| - | unset SLURM_JOBID | ||
| - | export FLUENT_AFFINITY=0 | ||
| - | export SLURM_ENABLED=1 | ||
| - | export SCHEDULER_TIGHT_COUPLING=1 | ||
| - | FL_SCHEDULER_HOST_FILE=slurm.${SLURM_JOB_ID}.hosts | ||
| - | /bin/rm -rf ${FL_SCHEDULER_HOST_FILE} | ||
| - | scontrol show hostnames " | ||
| - | |||
| - | JOURNALFILE=fluent_transient # .jou file | ||
| - | |||
| - | if [ $SLURM_NNODES -eq 1 ]; then | ||
| - | # Single node with shared memory | ||
| - | fluent 3ddp -g -t $SLURM_NTASKS -i $JOURNALFILE.jou > $JOURNALFILE.log | ||
| - | else | ||
| - | # Multi-node | ||
| - | fluent 3ddp -platform=intel -g -t $SLURM_NTASKS -pib.ofed -mpi=intel -cnf=${FL_SCHEDULER_HOST_FILE} -i $JOURNALFILE.jou > $JOURNALFILE.log | ||
| - | fi | ||
| - | </ | ||
| - | |||
| - | This job script allows a variable definition of desired configuration. You can manipulate the number of compute nodes very easily and the job script generates the appropriate command to start the calculation with Fluent. | ||
| - | |||
| - | ---- | ||
| ===== License server settings ===== | ===== License server settings ===== | ||