Actions: GPU enabled GitHub-hosted runners (Limited Beta)

[github-product-roadmap]

Summary

GitHub Action runners will now include the option to choose GPU runners to enable Machine learning Ops workloads to be run on GitHub actions. These new GPU runners come pre-loaded with the drivers and tools developers/data scientists need to run their models more efficiently as part of their CI/CD process, empowering teams to do complete application testing, including the ML components, in Actions.

Intended Outcome

Enable ML Ops workloads to fully embrace GitHub Actions without the overhead of managing their own infrastructure or paying a 3rd party provider to manage infrastructure for them.

How will it work?

Users will be able to enable GPU machines as part of their runner groups. From here their team will be able to pick up and use these runners simply by changing their runs-on: key within their workflow YAML. These are fully managed by GitHub and have all the drivers required for the users to just 'pick up and go' with their ML workloads

To sign up, please visit this link.

[ankneis]

🚢 This has shipped. Sign-up here: https://resources.github.com/devops/accelerate-your-cicd-with-arm-and-gpu-runners-in-github-actions/