Aries is a 176-GPU cluster, containing AMD GPUs (courtesy of AMD).

Jobs are allocated by nodes, not GPUs. Since there are 8 GPUs per node, each submitted job must be able to use 8 GPUs. Remember to set properly in your script --gres=gpu variable, it is not automatically. Below, we provide examples for how to effectively use this resource.

Since each job submission on Aries runs exclusively on an entire GPU node, the user must launch 8 parallel GPU processes. It is ideal that all 8 parallel runs to have similar run time to maximize efficiency. Make sure the 8 parallel runs have different names for the output file generated.

All Aries users should have access to the data storage directory /work/cms16/. Create a sub-directory within /work/cms16/ with your "username".

mkdir /work/cms16/$USER

Recommended to store all generated data to your directory in /work/cms16/username and not in your /home. Your /home directory will be able to store around 20 GB for you, while /work partition has a total of 100 TB. 

Partition commons

This partition contains 22 GPU nodes (gn01-gn22). Each node contains 8 AMD Vega20 GPUs, 96 CPUs and 512 GB memory. Expect for the last three nodes (gn20-gn22) with each having 128 CPUs and 8 AMD MI100 GPUs. 

Partition highmem

This partition contains 2 CPU nodes (hm01-02) with each node having 64 CPUs and about 4 TB of RAM. These nodes don't have GPUs.

Singularity Container

For submitting OpenMM jobs, a singularity container with OpenMM pre-installed is available.

container=/home/pcw2/bin/openmm-ctbp.sif

This container does not have OpenSMOG and needs to be installed with pip3. 

$pip3 install OpenSMOG

Usage example with bash submission script, openmm run python script, and input files can be downloaded below.

aries_example.tar.gz

#!/bin/bash -l
#SBATCH --job-name=ctbpexample
#SBATCH --nodes=1
#SBATCH --cpus-per-task=96        #set to 96 if not using MPI (OpenMM does not use MPI)
#SBATCH --tasks-per-node=1
#SBATCH --export=ALL
#SBATCH --mem=0                   #each GPU assigned 32 GB by default
#SBATCH --gres=gpu:8 
#SBATCH --time=1-00:00:00         #max run time is 1 day

Launcher_GPU

For submitting jobs, another possibility is to use the Launcher_GPU module and bind each run/task to one gpu. For this, remember to set the '--cpus-per-task' as the total number of cpus divided by the number of '--tasks-per-node'. The example bellow, runs 8 jobs in parallel. For this, OpenMM was installed locally via conda environment. 

#!/bin/bash -l

#SBATCH --job-name=my_job
#SBATCH --account=commons
#SBATCH --partition=commons
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=8
#SBATCH --cpus-per-task=6
#SBATCH --threads-per-core=2
#SBATCH --mem-per-cpu=3G
#SBATCH --gres=gpu:8
#SBATCH --time=24:00:00
#SBATCH --export=ALL

echo "Submitting simulations..."

module purge

# Using Launcher_GPU on ARIES
module load foss/2020b Launcher_GPU OpenMPI 

# OpenMM loaded via conda
echo "Initiating conda environment:"
source $HOME/anaconda3/bin/activate
conda activate openmm

# This is for controling Launcher
export LAUNCHER_WORKDIR=`pwd`
export LAUNCHER_JOB_FILE=$PWD/launcher_jobs_sim
export LAUNCHER_BIND=1

echo "Job started on " `date`
echo "Running on hostname" `hostname`
echo "Job $SLURM_JOB_ID is running on: $SLURM_NODELIST"
echo "Job SLURM_SUBMIT_DIR is $SLURM_SUBMIT_DIR"
echo "Running on $SLURM_NNODES nodes"
echo "Running on $SLURM_NPROCS processors"
echo "CPUS per task is $SLURM_CPUS_PER_TASK"
echo "LAUNCHER_WORKDIR: $LAUNCHER_WORKDIR"
echo "Number of replicas is $max_replicas"
df -h

# This will adjust the total number of runs to nodes*8
max_replicas=$((SLURM_NNODES*8))

rm $LAUNCHER_WORKDIR/launcher_jobs_sim &> /dev/null

# Create Launcher_job_file needed by $LAUNCHER_DIR/paramrun
for i in `seq 1 $max_replicas`
do
	echo "python run_code.py input_$i output_${i}.log" >> $LAUNCHER_WORKDIR/launcher_jobs_sim
done

# This line launches the jobs in the launcher_jobs_sim file
$LAUNCHER_DIR/paramrun

echo "My job finished at:" `date`

Additional Information

Acknowledgments

Please acknowledge the use of this cluster in your publications if you used the cluster:

This work was made possible by the donation of critical hardware and resources from AMD COVID-19 HPC Fund.

The following image can be used for poster and presentations.