Today’s user desktop experience is rapidly changing back to a server-centric ideology. This mindscape has taken on many shapes and sizes over the last couple of years. It has a variety of names: virtual desktop infrastructure (VDI), cloud computing, client consolidation infrastructure (CCI), hosted virtual desktops (HVD), and virtual machine (VM). As these environments have grown and evolved, their basic low-level graphics have remained the same. This is due, historically, to a variety of infrastructure constraints. These constraints may have stemmed from a lack of technology to support higher graphics. They may stem from the lack of network infrastructure to transport data. Regardless, there is finally a solution to fit users’ needs. Infrastructure capabilities have matured enough to remotely render and deliver high-performance graphics to users’ desktops via server side GPUs.
Nvidia has developed a new technology called GRID. The GRID technology from Nvidia delivers a high-performance visual experience remotely to the end user. The idea is to distribute the graphics workloads across GRID GPUs with vGPU technology and then direct only the resulting pixels to the users. With this vGPU technology, users get the high-performance graphics experience of a physical, local desktop in the cloud, which can run programs that demand high-performance graphics like AutoCAD, SolidWorks, RenderMan, or Blender. Nvidia GRID is currently supported in most virtual environment’s hypervisors. These include Citrix, VMWare, Microsoft, and Nice. The GRID hardware is in the form of an extended video card (like the K1 and K2), and is available from most hardware manufacturers, Lenovo, Cisco UCS, Dell, and HP.
You can find more information at the following links: