Intel GPU resource driver

CAUTION: This is an beta / non-production software, do not use on production clusters.

Glossary

• DRA https://github.com/kubernetes/enhancements/tree/master/keps/sig-node/3063-dynamic-resource-allocation
• CDI https://github.com/container-orchestrated-devices/container-device-interface/
• K8s https://github.com/kubernetes/kubernetes.git

About resource driver

Intel GPU resource driver is a better alternative for Intel GPU device plugin, facilitating workload offloading by providing GPU access on Kubernetes cluster worker nodes.

Supported GPU devices (with Linux kernel Intel i915 GPU driver):

• Intel® Data Center GPU Max Series
• Intel® Data Center GPU Flex Series
• Intel® Arc A-Series
• Intel® Iris® Xe MAX
• Intel® Integrated graphics

About Dynamic Resource Allocation

Dynamic Resource Allocation (DRA) is a resource management framework in Kubernetes (1.26+), that allows management of special resources in cluster (typically HW accelerators) by vendor-provided resource drivers (typically a controller and a node-agent / kubelet-plugin) in a common way.

Resource drivers are meant to handle discovery, allocation, accounting of specific resources as well as their preparation for Pod before Pod startup, and cleanup after the Pod has completed successfully and the resource is no longer needed. More info is in the KEP

Intel GPU resource driver consists of the controller and kubelet plugin. Controller makes allocation decisions and kubelet plugin ensures that the allocated GPUs and SR-IOV Virtual Functions are prepared and available for Pods.

Requirements

• Kubernetes 1.28+, with DynamicResourceAllocation feature-flag enabled, and other cluster parameters
• Container runtime needs to support CDI:
  • CRI-O at least v1.23.0
  • Containerd at least v1.7 (any release candidate will do)

Supported Kubernetes Versions

Supported Kubernetes versions are listed below:

Branch	Kubernetes branch/version	Status
v0.1.0-beta	Kubernetes 1.26 branch v1.26.x	unsupported
v0.1.1-beta	Kubernetes 1.27 branch v1.27.x	unsupported
v0.2.0	Kubernetes 1.28 branch v1.28.x	unsupported
v0.3.0	Kubernetes 1.28+	supported
v0.4.0	Kubernetes 1.28+	supported

Documentation

• How to setup a Kubernetes cluster with DRA enabled
• How to deploy and use Intel GPU resource driver
• Optional: How to build Intel GPU resource driver container image

Release process

Project's release cadence is quarterly. During the release process the issue is created in Github to track progress based on release task template.

Once the content is available in the main branch and validation PASSes, release branch will be created (e.g. release-v0.2.0). The HEAD of release branch will also be tagged with the corresponding tag (e.g. v0.2.0).

During the release creation, the project's documentation, deployment files etc. will be changed to point to the newly created version.

Patch releases (e.g. 0.2.1) are done on a need basis if there are security issues or minor fixes for specific supported version. Fixes are always cherry-picked from the main branch to the release branches.