Compute
OpenShift Virtualization introduces support for virtual machines as a native capability to OpenShift, Red Hat’s enterprise application platform offering based on Kubernetes. The concept of using virtual machines in a “Kubernetes way” may seem strange to those that are used to traditional hypervisors. However, they will find that many of the same features and capabilities that are expected of any enterprise class virtualization solution, are available in OpenShift Virtualization.
GPU/vGPU access
If your virtual machines need to utilize (v)GPU resources, ensure that you follow the documentation for enabling access to mediated devices and enabling other features (IOMMU) as needed.
Hardware/PCI Passthrough
The PCI passthrough feature allows you to present direct access to hardware devices of host machines to the virtual guests themselves. This can be useful for a number of reasons primarily related to resource isolation for the virtual machine and the workload it is hosting. Examples include virtual machines running AI/ML based workloads, which can be granted direct access to GPU devices on the host, or perhaps a guest that requires a large amount of network bandwidth can be attached directly to a physical network adapter instead of sharing the SDN with other virtual machines. For more information related to device passthrough in OpenShift Virtualization, see the documentation here.