Learning Forum

[singh.112]

Hi, 

I am trying to run one of my fluent cases using GPU on our institute HPC server. 

I do submit my jobs using SLURM job scheduler on normal HPC system.

I just wanted to know what are all the necessary changes has to be made in job scheduler file to run it on GPU. 

This is the sample HPC job scheduler file- 

"

SLURM FILE

#!/bin/bash

#SBATCH --job-name=DiscCounter     # Job name

#SBATCH --nodes=1                    # Run all processes on a single node

#SBATCH --ntasks=16          # Run a single task

#SBATCH --ntasks-per-node=16           # Number of CPU cores per task

#SBATCH --mem=32gb                    # Total memory limit

#SBATCH --time=12:00:00              # Time limit hrs:min:sec

#SBATCH --partition=test   # Queue size small/medium/large

#SBATCH --output=first_%j.log     # Standard output and error log

date;hostname;pwd

module load icc/2023.1.0 mpi/2021.9.0

/scratch/apps/packages/ansys/ansys_2024/v241/fluent/bin/fluent -g 3ddp -t16 $SBATCH_NTASKS -i transient_journal.jou -ssh -mpi=intel -cnf=$SBATCH_NODEFILE

"

[MangeshANSYS]

Hello

For Fluent refer to the gpu command lie flag documented here
https://ansyshelp.ansys.com/public/account/secured?returnurl=/Views/Secured/corp/v251/en/flu_ug/flu_ug_startramp.html%23flu_ug_sec_startup_option 

For requesting GPU option resources from slurm, you will need to contact your slurm administrator on any specific syntax and guidelines to get the GPU resources assigned

Let me know if this is helpful