The introduction of the GPU revolutionised the graphical elements of computing in 1993 and this is only set to continue as GPUs become more and more powerful. A couple of weeks ago, at the annual GPU Technology Conference, NVidia introduced the first generation of GPUs for the cloud. NVidia has been working on the technology necessary to ensure GPUs can be used in the cloud for the past five years and have achieved success with their Kepler GPU architecture. This reduces ‘lag’ and will ring the changes for both virtualization and gaming alike, as previously, GPUs haven’t been able to cope with virtual environments due to the way in which they process information.
Kepler uses built-in streaming which allows for multiple users and its virtualization capabilities are well suited for use in large data centres. According to NVidia, costs are likely to be maintained at a low level due to power efficiency and processing density enhancements.
The first commercially available GPU, based on Kepler technology, has already been introduced by NVidia, who call it the “world’s first virtualized GPU” and is named NVidia VGX; it also allows users to run Windows or any applications from their connected device.
This means that a virtualized, or hosted, desktop can be implemented which has the power and graphics of a PC or workstation. NVidia say that the VGX platform allows for “a responsive experience for the full spectrum of applications previously only available on an office PC.”
In other words, whilst the gaming community may be getting excited about the possibilities offered by the VGX, so should the business community, as it offers even more options than they may have already considered when it comes to moving to the cloud.
Whilst of course, there are already plenty of hosted desktop choices available, the introduction of NVidia VGX means that businesses who require a lot of graphic processing power, such as CAD designers for example, can now also consider virtualization.
Previously, graphically intensive applications have been too much for a hosted desktop to handle. The new technology also has benefits for those running BYOD schemes, as the VGX platform can be used with any device, from tablet to laptop.
NVidia VGX has been made possible by three main technological breakthroughs: 4 NVidia VGX Boards, which allow for hosting of a large number of users and the first of which is configured with 4 GPUs and 16GB of memory and uses a standard PCI Express interface.
The GPU is virtualized using a software layer which integrates into commercial hypervisors, such as the Citrix XenServer, and there is a manageability option, USM, which allows enterprises to configure the graphics for each individual user.
"Desktop virtualization is rapidly becoming mainstream for enterprises. By leveraging the NVIDIA VGX platform, combined with XenDesktop and HDX technologies, we are enabling enterprise customers to virtually deliver graphics-intensive apps beyond power users and designers,” said Citrix’s Sumit Dhawan.
“Now, they can also serve users who require only occasional access to graphics-intensive apps, which previously would have been cost-prohibitive. The combined virtual desktop solution for serving these users can be reduced by up to 80 percent, while enabling users to securely access GPU-accelerated apps from any device."
For gaming purposes, NVidia’s GPU technology leverages Kepler’s virtual capabilities by reducing input latency by 30ms. This has been made possible by the creation of GeForce GRID technology and NVidia are keeping pretty stum about how they have gone about it. However, we expect that the gaming community will be less interested in how and more concerned with when they can begin gaming in the cloud with little or no lag – it doesn’t seem that it will be long before there will be more companies popping up offering GaaS.
Of course, the technology has much more value than just gaming, as demonstrated above, and this could mean even more options when it comes to the cloud and hosted desktops, especially for those who work with graphically intensive applications.