Nvidia Pushes Kepler for Cloud Gaming With GeForce Grid

Although the majority of focus at Nvidia's GPU Technology Conference is on the enterprise, scientific and business uses for graphics processors, the company didn't forget that many people buy its products for gaming.

Today, Nvidia announced what it calls the GeForce Grid cloud gaming platform, which is all about streaming next-generation games to virtually any device, without the amount of lag that hampers current offerings.

Using the Kepler architecture, GeForce Grid GPUs minimize power consumption by simultaneously encoding up to eight game streams. This allows providers to cost-effectively scale their service offerings to support millions of concurrent gamers.

Featuring two Kepler architecture-based GPUs, each with its own encoder, the processors have 3,072 CUDA technology cores and 4.7 teraflops of 3D shader performance. This enables providers to render highly complex games in the cloud and encode them on the GPU, rather than the CPU, allowing their servers to simultaneously run more game streams. Server power-consumption per game stream is reduced to about one-half that of previous implementations, an important metric for data centers.

At GTC, Nvidia and Gaikai demonstrated a virtual game console, consisting of an LG Cinema 3D Smart TV running a Gaikai application connected to a GeForce Grid GPU in a server 10 miles away. What appeared was relatively lag-free play through just an Ethernet cable and wireless USB game pad connected to the TV.

At the post-event keynote Q&A, Nvidia CEO Jen-Hsun Huang in response to a question clarified some of the latency claims. Nvidia's slide claims that it's able to drop more than 100ms from current cloud services and be even better than today's local consoles.

Digging a bit deeper into those numbers, Huang said that today's consoles are using technology that's 6-7 years old. With that hardware stretched to the very limit, machines are already straining to deliver the most latency-free play. In comparison, he says, Kepler can deliver 60 frames-per-second without breaking a sweat. Furthermore, images rendered by Kepler now skip the framebuffer, further cutting down time in the hardware.

Much of the speed increase comes right from the faster hardware in Kepler, with the external factor of the display remaining static. There's no doubt in our minds that GeForce Grid – which is essentially a brand of Nvidia Kepler-powered cloud gaming hardware – is faster than what we had yesterday, but then again, what we will have tomorrow will always be faster than what we have today.

The real interesting possibility to come out of the savings afforded by  more advanced hardware is what Nvidia hopes to become a Netflix-like economy of scale. Instead of devoting one GPU to one user, Kepler in servers can split the load to achieve bigger cost and power efficiencies. Nvidia envisions this to lead to a flat $10 monthly games subscription, and that's something everyone would want.

Read more about what Nvidia's CEO announced today at the keynote.

_{Read more from @MarcusYam on Twitter.}