General computational fluid dynamics solver (cell-centered FVM). GPUs are supported.
General Information
To obtain and checkout a product license please read Ansys suite first.
Example Jobscripts
The underlying test case described here can be downloaded here: NaturalConvection_SimulationFiles.zip
Job for 2 nodes with 40 tasks (on 40 cpu-cores) per node
#!/bin/bash #SBATCH -t 00:10:00 #SBATCH --nodes=2 #SBATCH --ntasks-per-node=40 #SBATCH -L ansys #SBATCH -p medium #SBATCH --mail-type=ALL #SBATCH --output="cavity.log.%j" #SBATCH --job-name=cavity_on_cpu module load ansys/2019r2 srun hostname -s > hostfile echo "Running on nodes: ${SLURM_JOB_NODELIST}" fluent 2d -g -t${SLURM_NTASKS} -ssh -mpi=intel -cnf=hostfile << EOFluentInput >cavity.out.$SLURM_JOB_ID file/read-case initial_run.cas.h5 parallel/partition/method/cartesian-axes 2 solve/initialize/initialize-flow solve/iterate 100 exit yes EOFluentInput echo '#################### Fluent finished ############'
Job for 2 nodes running with 4 GPUs per node and 4 host tasks per node
#!/bin/bash #SBATCH -t 00:10:00 #SBATCH --nodes=2 #SBATCH --ntasks-per-node=4 #SBATCH -L ansys #SBATCH -p gpu-a100 ### on emmy -p is simply called gpu #SBATCH --output="slurm.log.%j" #SBATCH --job-name=cavity_on_gpu echo "Running on nodes: ${SLURM_JOB_NODELIST}" srun hostname -s > hostfile module load ansys cat <<EOF > fluent.jou ; this is a Ansys journal file aka text user interface (TUI) file parallel/gpgpu/show file/read-cas initial_run.cas.h5 solve/iterate 100 file/write-case-data outputfile ok exit EOF fluent 2d -g -t${SLURM_NTASKS} -gpgpu=4 -mpi=intel -cnf=hostfile -i fluent.jou >/dev/null 2>&1 echo '#################### Fluent finished ############'
Your job can be offloaded if parallel/gpgpu/show denotes the selected devices with a "(*)".
Your job was offloaded successfully if the actual call of you solver prints "AMG on GPGPU".
In this case, your .trn output file contains device_list and amgx_and_runtime, respectively.
Ansys only supports certain GPU vendors/models:
https://www.ansys.com/it-solutions/platform-support/previous-releases
Look here for the PDF called "Graphics Cards Tested" of your version... (most Nividia, some AMD)
-gpgpu must be an integer multiple of the tasks per node (ntasks-per-node)