Following on demonstrations at the GPU Technology Conference last month, NVIDIA is introducing the RealityServer platform for cloud computing, due to be made available on November 30, 2009.

The platform is based on NVIDIA's Tesla RS server, equipped with multiple GPUs, that can stream 3D applications to a wide array of clients - desktops, notebooks, smartphones - via the web.

Combining with software provided by mental images, a wholly-owned subsidiary of NVIDIA, the platform enables client-side customers to use the parallel processing power of the server to interact with complex 3D models.

By keeping much of the heavy-lifting of the raytracing rendering on the server side, client machines needn't be powerful.

We can see scope for the RealityServer to folk who need to tap into GPU-accelerated rendering without having to own house the necessary hardware in multiple locations.

Learn more here.