Bringing the GPU to the Cloud

PEER 1 and bluegfx have announced a partnership that brings the GPU to the cloud, giving bluegfx’s customers access to high-powered graphics on pay-per-process basis.

Using NVIDIA’s Fermi-based Tesla GPU’s and RealityServer hosted in PEER 1’s global data centres, bluegfx’s customers in the games, design, architectural, film, broadcast and education industries will benefit from high performance 3D web services.  It delivers a 3D rendering and animation package that works remotely over the Internet onto the GPU cloud to demonstrate the concept of remote hybrid computing, with the GPU processing being done off the PC.

Hosting the GPUs and software within the cloud simplifies customer implementation and provides an inexpensive entry point for new users. Businesses only pay for the processing they need and can upgrade or downgrade as required.

Amanda Dunn, Cloud GPU Manager EMEA, PEER 1 Hosting adds: “We have the only hosted NVIDIA GPU service, which brings simple, cost-effective graphical processing power to the media industry on a pay-as-you-use basis. bluegfx has excellent experience in rich media, and working together we can offer high performance computing in a variety of areas such as  rendering, transcoding and modelling.”

The move of the GPU to the cloud seems to be popular – Amazon and NVIDIA announced a similar service earlier in the month.  It certainly sounds like a great idea – if you need to do 3D rendering, don’t buy expensive GPUs but instead use the pay-per-process 3D web services.  Obviously some GPU processing doesn’t need to happen in real-time, e.g rendering a special effects sequence, but would it be possible to use a tablet to present a 3D model in real-time, with all the hard work being done in the data centre?  If it was possible over Wi-fi or 4G data speeds, I can think of some pretty amazing applications, particularly in the medical space.