English French German Italian Japanese Korean Portuguese Spanish

NVIDIA vGPU Upgrades Virtual Environments for High-Resolution Graphics

Nvidia vgpu workstation

NVIDIA virtual GPU (vGPU) software improves the user experience and the performance of new GPUs that run graphics, AI and high performance computing workloads in virtualised environments. vGPU delivers the resolution required to work with cinema-quality video in the field, for example, or design products in 3D from a home office.

When used with the Quadro Virtual Data Center Workstation (Quadro vDWS) license, the software now supports up to two 8K displays. Users can see more details in the context of the surrounding image, without the need to zoom in and out, and can also use multiple or widescreen displays. NVIDIA has also updated GRID virtual PC (GRID vPC) software, made for mainstream corporate VDI users, to now include support for 5K displays.

Higher Resolution, More Performance

Nvidia vgpu maya

Nvidia vgpu arnold RenderView

Autodesk Maya (top image) and Arnold RenderView in Maya (lower image), streamed to the desktop.

With Quadro vDWS and GRID vPC, people who work with high-resolution graphics can access more performance from GPUs, even when they are using multiple 4K, 5K or 8K displays in a virtual environment. In effect, through virtualisation, they can speed up workflows and make more efficient use of GPU investment.

When powered by NVIDIA RTX 8000 and RTX 6000 GPUs in a data centre, NVIDIA vGPU software accelerates Quadro virtual workstations used to produce 3D visualisations, achieving and maintaining photoreal quality. NVIDIA Quadro vDWS or vComputeServer vGPU software can provision virtual graphics workstations and compute-intensive workloads from a single RTX Server, extending RTX capabilities to designers and artists working on most types of device.

In particular, NVIDIA vComputeServer software supports the V100S GPUs, which are processors built for AI, to accelerate deep learning, data science and HPC workflows in a virtualised environment. In terms of efficiency, vGPU software makes it possible for underused GPU resources to run other workloads, such as virtual desktops with NVIDIA GRID, or AI and HPC with vComputeServer.

Nvidia vgpu adobe

NVIDIA Quadro vDWS running Adobe Wetbrush.

Benchmarking and Analysing the VDI Experience

Measuring the quality of the virtualisation user experience has been a challenge - possibly slowing down user adoption or contributing to failed VDI initiatives, according to NVIDIA. Lakeside Software, a software company that generates performance data and analytics for physical and virtual desktops, is collaborating with NVIDIA to produce a clearer description of users’ experience of the virtual environment. 

NVIDIA’s benchmarking tool nVector measures the main aspects like end-user latency, framerate, image quality and server utilisation. Instead of focussing on the response time on the virtual desktop, nVector determines the experience of the user on the endpoint device, which results in a better estimation of how fast and smooth the experience feels, as well as image quality. This type of information helps set up relevant utilisation thresholds, which the IT team uses to architect and size the VDI infrastructure.

Nvidia vgpu visualisation

Streaming video for visualisation applications.

Sizing VDI balances the end-user experience with infrastructure costs. It must also account for the fact that software and user needs change over time, usually calling for more resources and a better quality of experience. To continuously optimise the system, VDI administrators need tools that accurately assess current user experience and identify the causes of poor performance.

Combined with nVector benchmarking, Lakeside’s digital experience monitoring platform SysTrack collects about 10,000 data points (facts about the experience) on the endpoint, every 15 seconds. The granularity of collection makes it suitable for finding the root causes of user experience problems discovered by nVector. SysTrack shows when the problem started, what causes it and the best way to resolve it.   www.nvidia.com