Microsoft unveiled more about xCloud last week (after first announcing in October 2018); and, this week at the Game Developers Conference, Google officially announced Stadia.
We’re interested in game streaming because it’s essentially a highly specialized application of VDI. In January, Jack looked at the current state of cloud gaming and the different companies coming to market.
So, what do we know about Stadia and xCloud right now?
In late 2018, Google conducted tests of Project Stream and then went radio silent. Some thought maybe Google had quietly killed the project. But lo, it has returned.
Stadia runs on nearly every device, from TVs to Chromebooks, PCs, phones, and tablets. For the moment, Stadia requires Chrome, but Google suggested other browsers might eventually work, too. The game streaming platform runs on a Linux OS and uses Vulkan and a custom AMD GPU. They explained that they have purpose-built gaming racks in their datacenters capable of 10.7 teraflops—more than the Xbox One X and PS4 Pro combined. (As this console generation winds down, we’ll see how Stadia compares to the PS5 and the next Xbox console.)
So what kind of performance can users expect out of Stadia? The Project Stream tests locked performance at 1080p, 30fps, and stereo audio, but at launch Stadia will support 4K, 60fps, HDR, and surround sound. Eventually, Stadia will be capable of 8K and 120fps.
In addition to general game streaming, Google talked about a few features. Stream Connect will resurrect couch co-op (largely absent in current consoles), Crowd Play will allow users to queue up in a lobby system to play with their favorite YouTube streamer (just made me think of placing quarters on arcade machines to indicate who had next), and cross-play with all participating platforms. One other feature I found interesting is State Share, which allows players to share playable moments that others can then experience.
Google also showed off the Stadia Controller that will connect via Wi-Fi in the backend to the datacenter streaming the game and not the local device (using the controller makes it easier to jump between devices). This could reduce latency since it doesn’t have to connect to TV, tablet, etc. to relay user input. The controller is optional if playing on PC, but required when playing on TV, which requires Chromecast. Lastly, there’s a button for Google Assistant on the controller, because of course there is.
Another interesting aspect is that Stadia will have its own first-party game development studio, Stadia Games and Entertainment, headed by industry veteran Jade Raymond. Stadia will need some exclusive titles to draw in folk who might otherwise just stick with their console/PC. It also possibly shows Google’s belief in game streaming.
Stadia launches this year (no specific date yet) and Google said more details would be forthcoming in the summer (maybe at E3?).
Likely in an attempt to pre-empt coverage, Microsoft demoed xCloud a week before Google’s keynote, showing it running on a smartphone connected to an Xbox controller. Unfortunately, details remain a little sparse for xCloud.
Microsoft says to think of xCloud as more of an extension of the Xbox, rather than as a replacement. It offers people more options—maybe there’s an Xbox-exclusive title you’ve always wanted to play but don’t have the console or a gaming PC. Microsoft is clearly interested in gaming on the go (some markets, like Japan where Xbox perennially underperforms, prefer mobile gaming to consoles). Microsoft has been a little cagey about target resolution, but with a focus on mobile, 1080p seems like the goal.
Microsoft offers game developers a “Microsoft Game Stack” that looks to rope them into using Azure and other cloud services. xCloud runs in Azure datacenters, starting out as literal racks of Xbox One S consoles before moving on to a more datacenter-optimized version of the console.
xCloud remains a work in progress, with public trials beginning later this year.
Questions (with some answers) around Stadia/xCloud
The demos for both left us with some questions. The big question neither really answered during their respective presentations was what internet speeds will be required to enjoy Stadia or xCloud. Thankfully, Microsoft and Google did reveal a little afterward.
Kareem Choudhry, corporate VP of gaming cloud at Microsoft, spoke about how they want xCloud to run on low speeds, roughly around 10mbps or less. Meanwhile, Stadia will require more: There is a “worst case scenario” mode that simulates a 15mbps DSL connection for developers. Expect to need a consistent 25mbps internet connection to get 1080p and 60fps, with 30mbps needed to get 4K. Google explained that while Stadia uses less than 25mbps, that’s what they’re recommending for users.
Not the most taxing of requirements for either, but not everyone has access to 25mbps worldwide, let alone in the U.S.; infrastructure continues to lag behind demand. So, xCloud potentially has a leg up here.
What we still don’t enough about yet is what latency will be like for users. Reading through comment sections in the various articles leads me to believe that some believe Project Stream was good enough while others feel it was a laggy mess (and that’s not going into fuzzy resolution and screen tearing issues).
Eurogamer did some latency tests, but every user’s acceptance of input lag and general latency will be different. Some will be able to adjust to a little lag depending on what they’re playing, but if latency fluctuates throughout each session, it could render the experience frustrating.
A big question a lot of people will have is how much data will each consume? It’s nice to stream in 4K, but data caps are common and 4K is a hungry beast. Lastly, how will the services be priced? Will they be subscription services a la Xbox Game Pass, or more like Steam where you buy each game individually?
As we say goodbye to Google+, I do wonder how long Google will support game streaming if it struggles to gain much prominence early on. They do have a history of abandoning ideas. Still, my questions don’t diminish my excitement over the potential here.
How this relates to our space
Get ready for a lot of demos that are like remoting demos from 90s and 2000s. We already saw one for Stadia with a Google employee hopping between several different devices using the Stadia Controller (with maybe some terrible input lag witnessed, too). As Jack mentioned a couple months ago, a whole new industry is going to be getting conversive in VDI concepts.
We’ll hear all of the same arguments again: the benefits of delivering apps to different platforms, the security of remoting, and power available to games that run in the data center, patching handled by the company and not users, as well as the downside of needing a good connection, and on and on.
It’s easy to imagine that more than a few people in our industry, especially those focused on graphics, will find the opportunity to leap over to the much larger vertical. Video games are a multi-billion-dollar industry and many big names are already involved, like NVIDIA which has their own game streaming offering in beta. This could be the biggest application for VDI/DaaS ever!