Overview

Powerful HPC infrastructure

Enable your team to run your most intensive workloads on the latest and greatest infrastructure. CPUs from Intel, AMD, and Arm. GPUs from NVIDIA, including the A100 and H100. High-performance storage options spanning object, block, and file storage.

Advanced HPC services and tools

Deploy HPC quickly with the Cloud HPC Toolkit's prebuilt best-practices HPC blueprints. Deploy preconfigured modules for Compute Engine, Google Kubernetes Engine, Batch, or individual VMs.

Access data on Parallelstore, Filestore, Cloud Storage, or partner storage offerings.

Speed up tightly coupled workloads with compact placement policies, gVNIC, and the HPC VM image.

Cost-optimized HPC

Manage costs as you scale with budgets and committed use discounts. Save up to 91% with Spot VMs for flexible, check-pointable, and fault-tolerant workloads. Report on costs at a granular level easily with built-in labels in the Cloud HPC Toolkit.

How It Works

Google Cloud’s HPC solutions are easy to use, built on the latest technology, and cost-optimized to provide a flexible and powerful HPC foundation.

The Cloud HPC Toolkit enables you to easily launch new HPC environments.

Read the docs

Common Uses

Drug discovery

Financial services

Electronics design automation

Computer-aided engineering

Weather forecasting

Life sciences and genomics

Energy

Cloud HPC Toolkit

Learn how the toolkit works, then deploy a blueprint

Cloud HPC Toolkit Overview

Read the latest HPC blogs

Google Cloud HPC blogs

FAQ

The best Google Cloud service for running HPC workloads depends on your specific needs. There are a number of factors to consider when architecting your HPC environment. Some of those factors include:

Control: How much control do you need over your HPC environment?

Scalability: How scalable does your HPC environment need to be?

Cost: How much are you willing to spend on your HPC environment?

Ease of use: How easy do you need your HPC environment to be to use?

Once you have considered these factors, you can choose the Google Cloud service that is best for you. Here is a brief overview of each service, and how they relate to the above factors:

Compute Engine: Compute Engine is an infrastructure as a service (IaaS) offering that provides virtual machines (VMs) that can be used to run HPC workloads. Compute Engine gives you the most control and scalability over your HPC environment.

Google Kubernetes Engine: Google Kubernetes Engine is a managed Kubernetes service that can be used to run containerized HPC workloads. Google Kubernetes Engine is a good option if you want to use containerized applications, or want the ease of use that Kubernetes brings to managing your compute resources.

Batch: Batch is a managed service for running batch jobs. Batch is a good option if you want to run on Compute Engine, have a large number of HPC jobs that you need to run on a regular basis, and don't need deep customization of the infrastructure or scheduling policies.

Cloud Run: Cloud Run is a serverless platform that can be used to run small, simple HPC workloads. Cloud Run is a good option if you want to run HPC workloads without having to manage infrastructure. See the Cloud Run resource limits to understand the limitations.

Don't hesitate to reach out to the Google Cloud HPC team to discuss your requirements in depth.

The choice between the Cloud HPC Toolkit and an HPC as a service platform partner depends on your specific needs. Here are some factors to consider when choosing between the Cloud HPC Toolkit and an HPC as a service platform partner:

Control: How much control do you need over your HPC environment?

Ease of use: How easy do you need your HPC environment to be to use?

Cost: How much are you willing to spend on your HPC environment?

Expertise: How much expertise do you have in HPC?

Once you have considered these factors, you can choose the option that is best for you. Here is a brief overview of each option:

Cloud HPC Toolkit

The Cloud HPC Toolkit is a set of open source tools that can be used to deploy and manage HPC workloads on Google Cloud. The toolkit provides a number of features, including being open source, Terraform and Cloud Foundation Toolkit based, composable, and integrated with Google Cloud services and popular HPC tooling and applications. The Cloud HPC Toolkit can be used through a web-based user interface with the open front end.

The Cloud HPC Toolkit is a good option if you want to have a high degree of control over your HPC environment. The Cloud HPC Toolkit was built to provide better ease of use than building DIY HPC environments. It both provides more configurability and requires more configuration than an HPC as a service platform from a partner, and therefore is best suited for users with more HPC expertise.

HPC as a service platform partner

An HPC as a service platform partner is a third-party company that provides a managed HPC platform on Google Cloud. These platforms typically provide a number of features, including preconfigured HPC environments, user-friendly interfaces, and technical support.

An HPC as a service platform partner is a good option if you want to get started with HPC quickly and easily, or want to provide users a simple, GUI-based user experience. However, they can be less flexible, or include extra costs.

In general, the Cloud HPC Toolkit is a good option for users who have a high degree of expertise in HPC and want a high degree of control over their HPC environment. HPC as a service platform partners are a good option for users who want to get started with HPC quickly and easily.

Didn't find your answer? Contact the Google Cloud team