![[Google Cloud Documentation]](https://www.gstatic.com/devrel-devsite/prod/v0f3a5b8b76e4a120372a69044e02756e703b6140f9f1812c6e4a0407b19864b7/clouddocs/images/lockup_full_color.svg){kind=link}
![[]](https://docs.cloud.google.com/_static/clouddocs/images/icons/products/ai-hypercomputer-color.svg){kind=link}
To run artificial intelligence (AI), machine learning (ML), or high performance computing (HPC) workloads, you can deploy AI-optimized Compute Engine instances and clusters that use A4X, A4, A3 Ultra, A3 Mega, and A3 High (8 GPUs) machines. For more information about the features of these machines that enable you to run large-scale AI and ML clusters, see Cluster management overview.
Note: The Cluster Director and Cluster Toolkit guides only apply to the A4X, A4, A3 Ultra, A3 Mega, and A3 High (8 GPUs) machine types. We recommend that you consult the relevant documentation for your choice of machine type. To identify the compute instance and cluster creation guide for other machine types, see Choose a deployment strategy.You can create A4X, A4, A3 Ultra, A3 Mega, and A3 High (8 GPUs) instances directly from Compute Engine, or through other services that run on Compute Engine instances like Cluster Toolkit or Google Kubernetes Engine.
For the most appropriate option to create your compute instances or clusters for your use case, choose one of the following:
| Option | Use case |
|---|---|
| Cluster Director | You want a fully managed service that automates the setup and configuration of your Slurm clusters. Cluster Director helps you configure compute, networking, and storage resources for your clusters to maximize performance and minimize downtimes. To learn more, see Create a fully managed Slurm cluster for AI workloads. |
| Cluster Toolkit | You want to use open-source software that simplifies the process for you to deploy both Slurm and GKE clusters. Cluster Toolkit is designed to be highly customizable and extensible. You can use blueprints to provision networking and storage resources. To learn more, see the following: |
| GKE | You want maximum flexibility in configuring your Google Kubernetes Engine cluster based on the needs of your workload. To learn more, see the following: |
| Use Compute Engine | You want full control of the infrastructure layer so that you can set up your own orchestrator. To learn more, see the following:
|
What's next
- Learn about the performance-optimized infrastructure of AI Hypercomputer: