Skip to content

Latest commit

 

History

History
219 lines (152 loc) · 6.9 KB

README.md

File metadata and controls

219 lines (152 loc) · 6.9 KB
Raw

Barracuda on AWS ParallelCluster

Barracuda on AWS ParallelCluster provides steps and code samples to build and run Barracuda Virtual Reactor on AWS using AWS ParallelCluster.

Barracuda On AWS

Barracuda Virtual Reactor simulates the 3D, transient behavior in fluid-particle systems including the multiphase hydrodynamics, heat balance and chemical reactions. It is a product from CPFD Software, visit their webpage for more information.

Architecture

Barracuda ParallelCluster Architecture

Deploying Barracuda on AWS

AWS CloudShell

AWS CloudShell is a browser-based, pre-authenticated shell that you can launch directly from the AWS Management Console. You can run AWS CLI commands against AWS services using your preferred shell, such as Bash, PowerShell, or Z shell. And, you can do this without needing to download or install command line tools.

You can launch AWS CloudShell from the AWS Management Console, and the AWS credentials that you used to sign in to the console are automatically available in a new shell session. This pre-authentication of AWS CloudShell users allows you to skip configuring credentials when interacting with AWS services using AWS CLI version 2. The AWS CLI is pre-installed on the shell's compute environment.

Launch AWS CloudShell

Prerequisites

Let start by downloading the Barracuda repository containing the Infrastructure as Code on your AWS CloudShell.

On the AWS CloudShell, run the script below to install the prerequisited software:

wget https://github.com/aws-samples/awsome-hpc/archive/refs/heads/main.tar.gz
mkdir -p AWSome-hpc
tar -xvzf main.tar.gz -C AWSome-hpc --strip-components 1
cd AWSome-hpc/apps/barracuda
bash ./scripts/setup/install_prerequisites.sh

The script will install the following on the Cloud9 instance:

Install AWS ParallelCluster

Create your Python3 virtual environment

python3 -m venv .env
source .env/bin/activate

Install AWS ParallelCluster

pip3 install aws-parallelcluster==3.4.1

Deploy AWS ParallelCluster with Barracuda

Create the AWS ParallelCluster Configuration file. Instances that will be used are p4de.24xlarge.

. ./scripts/setup/create_parallelcluster_config.sh

Create the Barracuda Cluster

CLUSTER_NAME="barracuda-cluster"
echo "export CLUSTER_NAME=${CLUSTER_NAME}" >> ~/.bashrc
pcluster create-cluster -n ${CLUSTER_NAME} -c config/barracuda-cluster.yaml --region ${AWS_REGION}

Connect to the cluster

pcluster ssh -n ${CLUSTER_NAME} -i ~/.ssh/${SSH_KEY_NAME} --region ${AWS_REGION}

Download Barracuda

wget -P /shared https://cpfd-software.com/wp-content/uploads/2022/11/barracuda_virtual_reactor-22.1.0-Linux.tar.gz

Extract archive

tar -xvzf /shared/barracuda_virtual_reactor-22.1.0-Linux.tar.gz -C /shared

Install Barracuda

/shared/barracuda_virtual_reactor-22.1.0-Linux/barracuda_virtual_reactor-22.1.0-Linux.run install --default-answer --accept-licenses --confirm-command --root /shared/Barracuda/22.1.0
echo "export PATH=/shared/Barracuda/22.1.0/bin:$PATH" >> ~/.bashrc

Run Barracuda

In this section, you will go through the steps to run test case(s) provided by Barracuda on AWS ParallelCluster.

Gasifier

In this section, you will learn how to run Barracuda on a Gasifier test case.

Setup

Download Sample case.

wget -P /shared https://cpfd-software.com/wp-content/uploads/2023/02/barracuda_sample_case.zip

Add your license file in /shared/ls.rlmcloud.com.lic

Create submission script that will run the simulation on one p3.2xlarge Amazon EC2 instance using NVIDIA V100 GPU.

cat > barracuda-gasifier-sub.sh << EOF
#!/bin/bash

#SBATCH --job-name=barracuda-gasifier
#SBATCH --output=%x_%j.out
#SBATCH --error=%x_%j.err
#SBATCH --partition=gpu-od-queue
#SBATCH --ntasks=1
#SBATCH --gpus=v100:1
#SBATCH --constraint=p3

# Set WORK_DIR as scratch if local storage exist.
# Otherwise use tmp
export WORK_DIR=/scratch/\$SLURM_JOB_ID

if [ ! -d /scratch ]; then
    export WORK_DIR=/tmp/\$SLURM_JOB_ID
fi

echo \$WORK_DIR
unzip -j /shared/barracuda_sample_case.zip -d \${WORK_DIR}
cd \${WORK_DIR}

export cpfd_LICENSE="/shared/ls.rlmcloud.com.lic"
/shared/Barracuda/22.1.0/bin/cpfd.x -ow -cc -ct -cbc -cic -qmdp -qll -qfe -gpu -d0 -fallback quit gasifier.prj

tar -czf /shared/barracuda-gasifier-results.tar.gz \${WORK_DIR}
EOF

You can also run on p4d.24xlarge or p4de.24xlarge Amazon EC2 instances by modifying the submission script above. As an example for p4de, you should replace --gpus=v100:1 by --gpus=a100:1 and --constraint=p3 by --constrant=p4de to run p4de.24xlarge EC2 instance.

Submit test case to Slurm

sbatch barracuda-gasifier-sub.sh << EOF

The job should complete in ~4 hours on one p3.2xlarge Amazon EC2 Instances.

Visualize Gasifier results

Once the simulation is completed, you can visualize the results. Extract the results archive

tar -xvzf /shared/barracuda-gasifier-results.tar.gz

Install the xdg-utils package.

sudo yum install -y xdg-utils

Let's exit the head node of AWS ParallelCluster to return to AWS Cloud9 environment.

exit

To visualize the results of the Gasifier test case, you will create remote visualization session using DCV

pcluster dcv-connect -n ${CLUSTER_NAME} --key-path ~/.ssh/${SSH_KEY_NAME} --region ${AWS_REGION}

You should obtain a response like this. DCV link

Copy and Paste the https link to a new tab of your web brower. It will create a remote visualization session.

Open a terminal. Launch Barracuda by typing barracuda in the terminal.

Open the gasifier project file, gasifier.prj. Barracuda Open

Visualize the results by selecting Post-processing > View results. Barracuda Visualization

You can now visualize at the results of the gasifier simulation. Barracuda Results

Cleanup your cluster

To avoid unexpected charges to your account relative to the Barracuda cluster, make sure you delete the cluster and associated resources.

Delete the cluster.

pcluster delete-cluster -n ${CLUSTER_NAME} --region ${AWS_REGION}

The steps below are optional if you plan to deploy a cluster with Barracuda in the future.

Delete remaining components of the Barracuda solution

. ./scripts/cleanup/cleanup_solution_components.sh