Multi-vGPU and P2P#

Multi-vGPU™ and Peer-to-Peer (P2P) CUDA® transfers are related capabilities for multi-GPU VM workloads. Use this hub for the shared hypervisor matrix, then open the concept or support page you need.

Table 23 Multi-vGPU and P2P pages#

Page

Use when you need

Multi-vGPU

Concepts and compatibility rules for attaching several vGPUs to one VM

Multi-vGPU board support

Architecture-specific board and vGPU combinations for Multi-vGPU

Peer-to-Peer (P2P) CUDA Transfers

NVLink™ P2P between vGPUs in the same VM, profile tables, and the A100 / UVM caveat

Hypervisor Platform Support for Multi-vGPU and P2P#

The hypervisor support below applies to both Multi-vGPU and P2P features.

Table 24 Hypervisor Platform Support for Multi-vGPU and P2P#

Hypervisor Platform

NVIDIA® AI Enterprise Infra Release

Supported vGPU Types

Documentation

Red Hat Enterprise Linux with KVM

All active NVIDIA AI Enterprise Infra Releases

All NVIDIA vGPU for Compute with PCIe GPUs; on supported GPUs, both time-sliced and MIG-backed vGPUs are supported.

Setting up Multi-vGPU VMs on RHEL KVM

Ubuntu with KVM

All active NVIDIA AI Enterprise Infra Releases

All NVIDIA vGPU for Compute with PCIe GPUs; on supported GPUs, both time-sliced and MIG-backed vGPUs are supported.

Adding One or More vGPUs to a Linux with KVM Hypervisor VM

VMware vSphere

All active NVIDIA AI Enterprise Infra Releases

Time-sliced multi-vGPU on supported GPUs. MIG-backed multi-vGPU requires VMware Cloud Foundation (VCF) 9.1 or later.

Setting up Multi-vGPU on VMware vSphere 8

Note

P2P CUDA transfers are not supported on Windows. Only Linux OS distros as outlined in the Support Matrix are supported.

Note

MIG-backed multi-vGPU on VMware vSphere requires VMware VCF 9.1 or later. It is not supported on VCF 9.0P01 or on standalone ESXi without VCF 9.1. Linux KVM hypervisors continue to support MIG-backed multi-vGPU on supported GPUs.