Scientific Computing Examples on Google Cloud

This repository provides a collection of examples for running scientific computing workloads on Google Cloud Platform (GCP). The examples cover a wide range of topics, from setting up HPC clusters to running specific applications in various scientific domains.

Repository Structure

The repository is organized into the following top-level directories:

• apptainer: Examples of using Apptainer (formerly Singularity) for containerizing scientific applications.
• custom-workbenches: Instructions for creating custom Vertex AI Workbench instances for scientific machine learning.
• eda-examples: Examples of running Electronic Design Automation (EDA) workloads on GCP.
• fluxfw-gcp: Scripts and Terraform configs for deploying the Flux Framework on GCP.
• fsi: Examples for the Financial Services Industry (FSI), including Monte Carlo simulations.
• hcls: Examples for the Health and Life Sciences (HCLS) domain, including genomics and molecular dynamics.
• higgs: An example of rediscovering the Higgs boson using GCP, based on the 2012 CERN experiment.
• ibm-symphony-on-gcp: A blueprint for deploying an IBM Spectrum Symphony cluster on GCP.
• notebooks: Jupyter notebooks for demonstrating various GCP services, such as Google Cloud Batch.
• python-batch: Python scripts for interacting with the Google Cloud Batch API.
• qwiklabs: Materials for Qwiklabs, including an AlphaFold3 lab.
• slurm-cookbook: A cookbook for setting up and using Slurm on GCP, including running Docker containers.

Getting Started

To get started with the examples in this repository, you will need a Google Cloud Platform account with billing enabled. You will also need to have the following tools installed and configured:

• Google Cloud SDK
• Terraform
• Git

Most of the examples use the Google Cloud HPC Toolkit to deploy the necessary infrastructure. Please follow the installation instructions to set it up.

Examples

Here is a curated list of examples from different domains:

• Apptainer:
  • Building Apptainer (SIF) Images on Google Cloud
  • Apptainer Enabled Slurm Clusters
  • OpenFOAM Simulation and Visualization
• Custom Workbenches:
  • Creating Custom VertexAI Workbench Instances for Scientific Machine Learning
• EDA:
  • EDA Workloads on Google Cloud
• Flux Framework:
  • Flux Framework on Google Cloud
• FSI:
  • NVIDIA CUDA Samples -- MonteCarloMultiGPU - Financial Services Industry
• HCLS:
  • Running NAMD APOA1 Benchmark using GPUs with Google Cloud Cluster Toolkit
  • Google Batch Submission for NVIDIA Clara Parabricks Workloads
• Higgs Boson Discovery:
  • Rediscovering the Higgs boson using the Google Cloud Platform
• IBM Spectrum Symphony:
  • IBM Spectrum Symphony on Google Cloud
• Python Batch:
  • GCP PyBatch
• Slurm:
  • HPC Toolkit Support for Slurm and Docker on Google Cloud

Contributing

We welcome contributions to this repository. Please see the CONTRIBUTING.md file for more information.

License

This repository is licensed under the Apache 2.0 License. See the LICENSE file for more information.