Licensing

You can refer to the NVIDIA License System documentation and the NLS FAQ.

Refer to the NVIDIA Virtual GPU Licensing Guide.

First, determine whether you will use an Express CLS or require a Custom CLS instance. An Express CLS installation simplifies the process by automatically installing and binding the CLS instance to the NVIDIA License server. With a Custom CLS installation, these steps must be performed manually. Still, they may be needed based on the existing configuration or in cases in which legacy NVIDIA vGPU Software Licensing servers are present. Refer to this document for more detailed installation instructions on how to install either the Express CLS instance or the Custom CLS section.

The DLS instance can be installed in one of three ways: as a virtual appliance, container, or installable package. For detailed instructions on how to install and what hypervisors, container management, or operating systems are supported, please reference the Installing and Configuring the DLS Virtual Appliance documentation.

For additional details, including how to scale and size your environment, refer to the Sizing Guidelines for a DLS Appliance.

A client configuration token is a file that contains information about the NVIDIA vGPU licensing server. This file is used by the NVIDIA vGPU software driver on a virtual machine (VM) to connect to the licensing server and obtain a license. For more information, see Generating a Client Configuration Token.

The client configuration token contains information required for vGPU clients to connect to CLS/DLS service instances, including URLs (IP or FQDN) and port numbers. This information is digitally signed to ensure it has not been altered, and clients validate the signature to maintain integrity and security.

Yes. Refer to the packaging, pricing, and licensing guide for detailed information on licensing.

Yes, evaluation licenses are available for 128 users for 90 days on all NVIDIA vGPU software products.

Yes. NVIDIA vGPU software is supported on several cloud services with bring your own license (BYOL) licensing and licensing provided by the cloud service. Check out some of our CSPs for more information:

• Alibaba Cloud, the license is provided by the cloud service

• AWS, the license may be provided by the cloud service or supplied by the user (BYOL)

• Baidu AI Cloud, the license may be provided by the cloud service or supplied by the user (BYOL)

• Google Cloud, the license may be provided by the cloud service or supplied by the user (BYOL)

• IBM Cloud, the license must be provided by the user (BYOL)

• Microsoft Azure, the license is provided by the cloud service

• Oracle Cloud, the license must be provided by the user (BYOL)

• Tencent Cloud, the license may be provided by the cloud service or supplied by the user (BYOL)

How NVIDIA vApps licenses are enforced depends on the type of deployment for which the license is being used:

• For A-series NVIDIA vGPU deployments, one vApps license is required for each concurrent user, but licensing for only one of these users is enforced through software. Licensing for the remaining users is enforced through the end-user license agreement (EULA).

• Licensing for the following deployments is enforced through the EULA:

  • GPU pass-through

  • Bare metal

  • Microsoft DDA

  • VMware vDGA

It is not recommended to assign licenses that have not yet reached their start date to your license server. If you do so, the server will display both active and inactive licenses, but the co-termed license will remain unusable until its start date.

No. The license server will only use licenses that are currently active. The active license must remain on the license server until the co-termed license reaches its start date.

600 vApps licenses are needed. One vApps license is needed for each concurrent user.

For most customer use cases, a vApps license would be recommended for app streaming and session based solutions. Each concurrent user accessing the application would require a vApps license. If an application requires any of the RTX vWS features, you will need a RTX vWS license for each user.

For most customer use cases, a vApps license would be recommended for app streaming and session based solutions. Each concurrent user accessing the application would require a vApps license. If an application requires any of the RTX vWS features, you will need a RTX vWS license for each user.

300 vPC or vWS licenses are needed. Each virtual desktop requires a license after it is started and running. One vPC or vWS license is needed for each concurrent VM.

One vPC or vWS license for each VDI desktop is needed. Both NVIDIA Virtual PC and RTX vWS include a per-user entitlement for vApps.

When licensing is enforced through software, the vGPU or physical GPU assigned to the VM operates at full capacity for the first 20 minutes. If the VM fails to obtain a license within this period, the performance of the vGPU or physical GPU is degraded, affecting the user experience.

For deployments where licensing is enforced via the EULA only, no license is checked out and no performance enforcement is applied.

For more information, see the NVIDIA Client Licensing User Guide.

One NVIDIA vApps license is needed for each concurrent user session.

NVIDIA does not support or recommend such a configuration for commercial or education use.

Professional graphics users who require advanced features (e.g., app certification, higher frame rates, CUDA support, rendering, or CAD) require one RTX vWS license per running VM. Knowledge workers using standard office applications (e.g., Microsoft Office, Adobe Acrobat, web browsers) require only a vPC license per running VM.

Autodesk AutoCAD, Autodesk Revit, Solidworks 3DExperience, Siemens NX, PTC Creo, Dassault Systèmes CATIA, Esri ArcGIS, Adobe Creative Cloud, Schlumberger Petrel, Ansys Discovery Live, medical PACs apps, and RTX Rendering Apps.