Document URL: https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html/virtualization_deployment_and_administration_guide/sect-device-gpu#sect-device-GPU-asignment Section Number and Name: 17.7.1. GPU PCI Device Assignment Describe the issue: NVIDIA now produces Tesla cards that make use of a proprietary hardware interconnect known as NVLink. We do not currently support assignment of NVLink based GPUs therefore let's make the documentation more clear that we're identifying a set of PCI Express devices as supported. Suggestions for improvement: The section currently states: "Red Hat Enterprise Linux 7 supports PCI device assignment of the following GPU devices as non-VGA graphics devices:" I would suggest: "Red Hat Enterprise Linux 7 supports PCI device assignment of the following PCIe-based GPU devices as non-VGA graphics devices:" Additional information: "PCIe" and "PCI-e" are official abbreviations of "PCI Express" per wikipedia, mix and match as you see fit. We use "PCI device assignment" correctly here because the overall technology is not limited to PCIe, but our support for this class of device is limited to PCIe.
Similar statement in RHEV docs noted in bug 1575755