Every year, the Game Developer’s Conference signals the direction the game industry is going in a variety of ways — particularly with regards to the technology and hardware powering the games with which players will eventually spend their time. But there’s also always an undercurrent of more urgent talks each year as developers and publishers discuss ways to make games in an increasingly budget-intensive and competitive field. And at this year’s GDC, those things all featured prominently, while also intersecting in curious ways.
One of the biggest announcements at this year’s show was Google’s Stadia platform, which purports to allow players to stream gameplay rather than downloading and installing games on dedicated systems. Ignoring the realities of the infrastructural challenges the internet presents to Google’s hopes for Stadia, one of the platform’s main selling points is the elimination of expensive hardware requirements. If Google has its way, players will never need to worry about specs or consoles again.
Ironically though, Google’s Stadia presentation gave us our first hints of the power threshold we can and should expect from the next generation of game consoles in development at both Microsoft and Sony. CPU and graphics hardware manufacturer and developer AMD is building the core technology for both the next generation PlayStation and Xbox devices, and Google also tapped the company for Stadia.
The resultant hardware, which will run each instance of a game that Stadia players spin up, is a custom AMD setup featuring 10.7 teraflops of graphics processing power, with a 2.7Ghz processor, and 16GB of very fast memory. This is all approximately six times as powerful as the original PlayStation 4 model, give or take some bandwidth here or there, which, as it happens, is the approximate jump made from the Xbox 360/PS3 era to the consoles that we’ve had in our homes since they launched in the Fall of 2013. It’s a safe assumption that the technological starting point for the next-generation of Xbox and PlayStation is where Google has now marked it. And, of course, both Microsoft and Sony are working on their own game streaming plans, making it highly likely that many of the instant-play features that Google is touting with Stadia will be present in some way on next-generation consoles.
The biggest gaming graphics developments at GDC were around ray tracing. Just two years ago, the prospect of ray tracing, in which the path of light in a virtual environment is realistically simulated, rather than being “faked”, leading to much more realistic shadows and lights in a scene, seemed like a pipe dream for video games. Companies like Dreamworks and Pixar have used ray tracing in their computer-animated films for years, which is a key reason for the exponentially higher level of verisimilitude in pre-rendered films over the real-time graphics in video games. But at GDC 2019, real-time ray tracing was on the lips of nearly every developer and publisher.
Graphics hardware-maker Nvidia has been the most publicly aggressive proponent of the technology, with a line of graphics cards for PCs with dedicated hardware launched last fall supporting it. Adoption initially seemed slow, but a snowball effect has been present, as both the Fortnite-powering Unreal Engine has rolled out an update integrating the technology, as well as rival engine developer Unity and Crytek’s CryEngine. The latter introduced support for ray tracing without dedicated hardware, and even Nvidia announced in March that it would expand access to ray tracing in games to its 2016-era GPUs in April.
As even midsized developers like Finland-based Remedy throw themselves into ray traced games with upcoming titles like “Control,” which uses the studios own proprietary tech, rather than building on the backs of others, it would seem that we’ve reached a tipping point with the technology — a tipping point that was clear in panel after panel as developers explained their workflows and processes to build ray tracing into their games so that other developers might benefit from those experiences. In addition, Microsoft aggressively pushed on ray tracing support in the DirectX development tools so critical to both Windows and Xbox game development.
The ubiquity of a technology like this at GDC is highly suggestive of an even heavier push we have yet to see. New console hardware is strongly rumored for a 2020 release, and while Nvidia is a successful company, its install base, particularly with ray tracing capable cards, is small. The development weight being thrown behind ray tracing indicates developers and publishers see a much, much bigger future in the technology — the kind of future install base that is really only offered by game console lifespans. More simply, the indications are present to suggest dedicated support of ray tracing in next-generation consoles, though how that will work is unknown. While Nvidia’s RTX GPUs have dedicated hardware for the tech — hardware that Nvidia itself says is necessary for the most efficient use of ray tracing — Crytek’s ray tracing demo was achieved on AMD GPUs without that kind of dedicated hardware.
Meanwhile, machine learning was also everywhere at GDC, and in a way that gamers will likely see have a profound effect on the look and feel of the games they play in the next five years. Nvidia’s DLSS technique uses machine learning to process hundreds of thousands of visual scenarios for specific games and applies that analysis to determine how best to tweak and adjust the image to look “best.” Results from this technique have been mixed so far, but promising. In addition, other publishers like Ubisoft put on panels dedicated to demonstrating how it uses so-called deep learning to catalog an enormous amount of visual assets that can then be searched and used by all of its studios spread across the planet. Machine learning, much like ray tracing, was a buzzword this year that will likely have a profound impact on game development, and in ways that are difficult to predict.