GPU instance types available for ROSA

With the increase in adoption of Kubernetes, it has become popular to run containerized applications using GPUs, which are specialized to process data, especially for artificial intelligence and machine-learning workloads. ROSA provides a managed Red Hat OpenShift service running on AWS Cloud for customers to quickly build, deploy and manage containerized applications.

Now, you can use GPU instance types with Red Hat OpenShift Service on AWS (ROSA) as well as Red Hat OpenShift Data Science to accelerate your deployment of AI/ML workloads. Support for the GPU instance type software stack is provided by AWS.

With GPU instance types now enabled for ROSA, you can develop, test and run AI/ML workloads that rely on accelerated instance-types from AWS. Additionally, with ROSA being a managed OpenShift service and OpenShift Data Science providing core machine learning tooling, customers can focus on accelerating the deployment of AI/ML workloads without the need to manage the complexities of the underlying Kubernetes infrastructure. 

Check to see the instance types available to you with the ROSA CLI, by running `rosa list instance-types`. 

The specific instances being made available are as follows:

• p3.2xlarge
• p3.8xlarge
• p3.16xlarge
• p3dn.24xlarge
• p4d.24xlarge
• g4dn.xlarge thru g4dn.metal
• g5.xlarge thru g5.48xlarge
• dl1.24xlarge (Intel)
  
  

The following are additional resources on this topic:

• About GPU Accelerated Amazon Web Services : https://www.nvidia.com/en-us/data-center/gpu-cloud-computing/amazon-web-services/
• About the NVIDIA GPU Operator: ​​https://docs.nvidia.com/datacenter/cloud-native/gpu-operator/platform-support.html 
• AWS and NVIDIA: https://aws.amazon.com/nvidia/
• Red Hat and NVIDIA:  http://redhat.com/nvidia 

Or try our learning path on the developer portal to learn how to create a Jupyter notebook to use GPUs for AI/ML modeling