This site requires Javascript in order to view all its content. Please enable Javascript in order to access all the functionality of this web site. Here are the instructions how to enable JavaScript in your web browser.

NVIDIA VIRTUAL GPU TECHNOLOGY

Accelerate Every Workload with NVIDIA vGPUs

GPU-ACCELERATED PERFORMANCE FOR TODAY’S DEMANDING WORKFLOWS

NVIDIA virtual GPU (vGPU) technology uses the power of NVIDIA GPUs and NVIDIA virtual GPU software products to accelerate every virtual workflow—from AI to virtual desktop infrastructure (VDI). By making GPU performance possible for every virtual machine (VM), vGPU technology enables users to work more efficiently and productively.

CPU-only VDI vs GPU-Powered VDI with NVIDIA Grid vPC

See why more customers have adopted NVIDIA vGPU-accelerated virtualization >

HOW IT WORKS

In a VDI environment powered by NVIDIA virtual GPU, the NVIDIA virtual GPU software is installed at the virtualization layer along with the hypervisor. The NVIDIA virtual GPU software creates virtual GPUs that enable every virtual machine (VM) to share a physical GPU installed on the server or allocate multiple GPUs to a single VM to power the most demanding workloads. The NVIDIA virtualization software includes a driver for every VM. NVIDIA Quadro^® Virtual Data Center Workstation (Quadro vDWS) includes, for example, the powerful Quadro driver. Because work that was typically done by the CPU is offloaded to the GPU, the user has a much better experience, and demanding engineering and creative applications can be supported in a virtualized and cloud environment.

FEATURES AND BENEFITS

MONITORING AND INSIGHTS

End-to-end insights across your entire virtualization stack let you create, deploy, and support at large scale with lower TCO.

LEARN MORE

Watch Video

LIVE MIGRATION

Enable live migration of GPU-enabled VMs using Citrix XenMotion and VMware vMotion support with NVIDIA vGPU.

LEARN MORE

MULTI-vGPU

Use the power of up to four GPUs per VM to enable new workflows for rendering, simulation, and design.

LEARN MORE

A PRODUCT FOR EVERY WORKLOAD

NVIDIA virtual GPU software is available in four editions to meet the needs of multiple virtualization use cases.

NVIDIA vComputeServer

For AI, deep learning, and data science workloads; includes an NVIDIA compute driver.

Download vComputeServer datasheet

LEARN MORE

NVIDIA Quadro vDWS

For professional graphics applications; includes an NVIDIA Quadro driver.

Download Quadro vDWS datasheet

LEARN MORE

NVIDIA GRID vPC

For virtual desktops delivering standard PC applications, browser, and multimedia.

Download GRID vPC datasheet

LEARN MORE

NVIDIA GRID vApps

Use with Citrix XenApp or other RDSH solutions like VMware Horizon Apps.

Download GRID vApps datasheet

LEARN MORE

FREQUENTLY ASKED QUESTIONS

Where do I get the NVIDIA drivers for my virtual GPU deployment?

NVIDIA virtual GPU solutions require both hardware and software licensing components. The NVIDIA compute, graphics, or Quadro drivers are available with your NVIDIA virtual GPU software license and can be downloaded from the customer portal.
What GPU cards are supported by NVIDIA virtual GPU software and how do I know which is right for me?

Find a list of NVIDIA GPUs supported by NVIDIA virtual GPU software here. Depending on the application, user density, and server form factor requirements, one GPU may be recommended over the others. See our line card for more information. Be sure to also check with your OEM to see which GPU is supported on their specific platform.
What software is required for the NVIDIA virtual GPU solution?

There are four editions of software products for NVIDIA virtual GPU: NVIDIA GRID® Virtual Applications (GRID vApps), NVIDIA GRID Virtual PC (GRID vPC), NVIDIA Quadro vDWS, and NVIDIA Virtual Compute Server (vComputeServer). A Support, Update, and Maintenance Subscription (SUMS) must be purchased along with the NVIDIA virtual GPU software license. Refer to our Licensing Guide to learn what features are supported with each software edition.
How do I properly size my VDI environment?

Sizing largely depends on the types of workloads the customer is running and the display requirements. For instance, a 1GB vGPU profile may be sufficient for a Windows 10 VDI user with general-purpose applications, but an Autodesk AutoCAD designer with three 4K resolution displays may need 2GB or more of framebuffer. Reading the performance guides by third party ISVs to understand the exact requirements of each application is highly recommended. NVIDIA Quadro Virtual Data Center Workstation Application Sizing Guides are also available for specific applications.
How do you quantify performance gains from using NVIDIA virtual GPU? How can we accurately measure the impact of using GPU acceleration?

Performance gains are achieved for each type of virtual GPU solution. For designers, running an application without GPU acceleration would be exceedingly slow in comparison. For knowledge workers,GPU acceleration increases computational density and improves user experience. Gains can be quantified using tools like VMware vRealize Operations (vROps) or the GPU Profiler. You can also refer to our whitepaper which measures user experience of a GPU-enabled VDI environment over a CPU-only VDI environment using metrics like latency, remoted frames, and image quality.
How does running AI, deep learning, and data science workloads virtualized with NVIDIA vComputeServer compare to running the same workloads in a bare-metal environment without virtualization?

NVIDIA vComputeServer software brings several benefits, including increased security for multi-tenancy environments; ease of management with virtualization tools and infrastructure; monitoring at the host, app, and guest levels; live migration for improved uptime and more efficient use of data center resources; and the ability to scale GPU resources from a fraction of a GPU to multiple GPUs with scheduler options for provisioning granularity.
Is there a performance difference when running compute-intensive workloads on vComputeServer versus on bare-metal servers?

Running a compute workload on bare metal will always yield the most optimal performance, however vGPU performance compared to bare metal is typically within a 10 percent performance delta, and usually less than 5 percent. Results may vary depending on the workload and number of GPUs being used.

CONNECT WITH vGPU EXPERTS

LEARN MORE

WHERE TO BUY

SEE PARTNERS

NVIDIA VIRTUAL GPU TECHNOLOGY

GPU-ACCELERATED PERFORMANCE FOR TODAY’S DEMANDING WORKFLOWS

HOW IT WORKS

FEATURES AND BENEFITS

A PRODUCT FOR EVERY WORKLOAD

FREQUENTLY ASKED QUESTIONS

Where do I get the NVIDIA drivers for my virtual GPU deployment?

What GPU cards are supported by NVIDIA virtual GPU software and how do I know which is right for me?

What software is required for the NVIDIA virtual GPU solution?

How do I properly size my VDI environment?

How do you quantify performance gains from using NVIDIA virtual GPU? How can we accurately measure the impact of using GPU acceleration?

How does running AI, deep learning, and data science workloads virtualized with NVIDIA vComputeServer compare to running the same workloads in a bare-metal environment without virtualization?

Is there a performance difference when running compute-intensive workloads on vComputeServer versus on bare-metal servers?

CONNECT WITH vGPU EXPERTS

WHERE TO BUY