At CES 2013 last week, Nvidia CEO Jen-Hsun Huang succinctly laid out the challenge facing the games industry at large in his opening remarks at the Nvidia press conference on Sunday evening: "It isn't possible for you to enjoy the same video game on any device."
Where the iPod, the Kindle, and the cloud have enabled consumers to consume music, books, and movies whenever and however they pleased, Huang said the challenge for the consumer tech companies is to invent the technology to make this happen with video games.
Huang's opening remarks may have been intended to offer context for Nvidia's subsequent product announcements, but they also conveniently framed the way game hardware manufacturers at CES are approaching the next year.
From chipset giants like Nvidia to smaller accessory developers like Mad Catz, everyone is trying to find a way to sell consumers on gadgets that will recreate their preferred game experience on any device they own -- but there are different approaches to making this happen.
One way to answer this challenge is to build a cloud streaming game service like Gaikai or OnLive, and literally offer players the same game on any device by offloading the heavy processing work to a central server facility and streaming the gameplay to a much lower-power device (a tablet, smartphone, lower-end PC, or smart TV chipset, for example). That way, the client device only needs to be powerful enough to handle an HD streaming video feed and the player's inputs, not the entire game's processing demands.
Cloud streaming game services and infrastructure were scattered all across CES: smart TVs from LG and Vizio with built-in Google TV support, including OnLive clients; budding streaming service Agawi announced a partnership with system-on-a-chip manufacturer Marvell to extend the reach of streaming game services in smart TVs and set-top boxes; and Nvidia announced Grid, a rack-mounted server meant specifically for cloud streaming game service providers.
Though OnLive sells games at near-retail prices (while also offering three- and five-day rentals for some titles) other providers monetize this model by selling access to games as a service, not unlike a premium TV channel. Nvidia sells the hardware to providers like Playcast and Agawi, and the providers negotiate with publishers to build a library of streaming games that end-users can pay a monthly fee to access from a compatible mobile device, set-top box/cable box, or smart TV. It's not yet clear which business model, if any, will win.
This approach certainly has its advantages -- it's relatively cheap for a new customer who owns a compatible device to sign up and start playing without having to buy any dedicated games hardware -- but the quality of the experience itself varies widely depending on the customer's broadband quality, distance from the server, and the stream provider's own configuration.
With this in mind, I asked Nvidia Grid senior project manager Andrew Fear about what they could do on their end to ensure a quality experience.
"We do everything we can to improve the latency by encoding the stream before we send it, but a lot of it will depend on the provider's network arrangement," says Fear. "There's certainly something to be said for having a smartly-distributed set of servers to reduce latency. The other thing is the type of games they're streaming; if one provider is only interested in serving Android casual games, they can serve more users per server. "
"We did change things in the hardware to improve latency, and when we measure it now, a traditional game console takes about 150ms to output to a console, and for Grid we're already at about the same level," Fear says.
"We don't control bandwidth, but we can anticipate trends. For us, when we send out our game streams it's about 5 mb/sec; in Seoul, that's a drop in the bucket. But I live in Texas, and every year my potential bandwidth is increasing. Certainly, there are always hiccups, but the infrastructure is building around Netflix for streaming video on demand, and our bandwidth demands are not that different."
A month or so before CES, I met with Playcast, one of Nvidia's streaming service partners, to try out its service. To be sure, the demo was promising; I tried the service out on both a smart TV and an Android tablet, and the service looked good and played well enough (especially considering that Playcast's US-side server deployment was fairly small at the time.) The system's on-the-fly video compression did appear to keep the game fairly responsive even when the network connection quality fluctuated.
But in order for cloud streaming games services to be a widely compelling product, each step in the delivery chain from streaming center to end-user needs to be up to par, and considering the tremendous variation in nationwide broadband quality in the U.S. alone, it's hard to tell when this kind of service will be able to muster the momentum it needs to succeed.
So while cloud streaming games may be able to deliver the same game on any device for a relatively low sticker price, there's no guarantee that the game experience would be a good one if you're living too far away from a server or your local broadband pipe doesn't have enough bandwidth or low enough latency.
What's more, even if the streaming conditions themselves were perfectly acceptable, playing a game with a touchscreen or average TV remote instead of the dual analog sticks or keyboard/mouse combo the game was designed for rarely makes for a quality game experience -- especially not for the core game audience who would be interested in subscribing to a streaming service in the first place.