The gap between computers and consoles has been shrinking for decades -- and now, Xbox head Phil Spencer wants to eliminate it altogether. Microsoft is already taking steps to unify its Xbox One and Windows 10 experience, through features like game streaming across networks and cross-buy capability. Upgrading hardware, however, is something else entirely.
Nonetheless, upgraded hardware appears to be what Spencer meant. "We see on other platforms whether it be mobile or PC that you get a continuous innovation that you rarely see on console," Polygon reports(Opens in a new window) Spencer as saying. "Consoles lock the hardware and the software platforms together at the beginning of the generation. Then you ride the generation out for seven or so years, while other ecosystems are getting better, faster, stronger. And then you wait for the next big step function."
This graph shows the step function Spencer is referring to. PC performance increases over time at a fairly steady rate, consoles have long periods of static performance, followed by a jump.
"When you look at the console space, I believe we will see more hardware innovation in the console space than we've ever seen," Spencer said. "You'll actually see us come out with new hardware capability during a generation allowing the same games to run backward and forward compatible because we have a Universal Windows Application running on top of the Universal Windows Platform that allows us to focus more and more on hardware innovation without invalidating the games that run on that platform."
A new console paradigm?
Before we hit the software side of things, let's talk about long-term trends in console development. The truth is, the gap between consoles and PCs has been shrinking, bit by bit, ever since the Nintendo NES launched in the mid-1980s. In the 1990s, consoles adopted CD-ROM and DVD-ROM technology, even if they used customized discs and encoding schemes. Microsoft's Xbox was the first mainstream console to use an Intel CPU and Nvidia GPU; both the Xbox 360 and PlayStation 3 made integrated storage standard, even if Microsoft did technically sell a disc-only option. The PlayStation 3 could run Linux, until Sony patched it out.
Today, the Xbox One and PlayStation 4 are PCs in everything but name. They rely on commodity x86 hardware and consumer graphics cards. They're built around low-level APIs that have much in common with their PC brethren like Vulkan, DX12, and AMD's Mantle. This is slightly more obvious with the Xbox One, which literally runs a version of Windows, but there's nothing at the hardware level that would prevent the PS4 from doing so as well.
The Xbox One and PS4 are PCs with highly integrated SoCs and custom operating system and software support, combined with a few unique bits of hardware. That's it. From a hardware perspective, there's nothing stopping Microsoft or Sony from creating a similar ecosystem around these devices, in which each iteration of the product is a superset of the hardware that came before.
It's easy to imagine what this might look like. A hypothetical Xbox Two could offer 8GB of DDR4-3200 (if Microsoft wanted to take the lower-cost route) or 8GB of HBM2. The GPU could be substantially improved, from its current 768 shader cores to 1536 or more. A 14nm node transition would make it easy to redesign AMD's Jaguar for higher clocks, putting the new chip in the 2-2.5GHz range. Fix the core's half-speed L2 cache, and boom -- you've got a substantially faster Xbox Two built on cutting-edge 14nm technology that's still perfectly backwards-compatible with the XBox One.What about software and sales?
Historically, this has been the major sticking point. Conventional wisdom says that console games are as fast and competitive as they are thanks to extensive optimization for a single hardware SKU. If developers have to target multiple SKUs for optimization, the thinking is, such advantages would be lost.
It's not clear if this is still the case. First of all, previous consoles were completely different each and every generation. A studio that had specialized in designing games for the PlayStation 2 had to start from scratch when the PS3 shipped. If Microsoft pursued an iterative upgrade policy, there'd be no need for an industry-wide reset every 5-8 years. Backwards compatibility would be assured thanks to a combination of software engineering and intrinsic hardware capability.
The Sony PlayStation move -- or, as I like to call it, the PS Wii
The other major argument against console upgrades is the fact that add-ons for existing consoles tend to either sell poorly or ship with very few titles. Both the Kinect and PlayStation Move fell into this trap, as did numerous peripherals in previous generations. At best, console add-ons become niche products with fringe appeal in specific titles. At worst, they're expensive, overhyped doorstops with all the game-improving power of a dead manatee.
Again, however, one of the reasons developers never targeted these half-steps in great numbers is because they've almost always been either a half-measure to extend the performance of a platform in need of refresh, or a gimmick that would work well in a handful of titles, at best.
With the Xbox One already standardized on Windows 10 and the DX12 API, Microsoft could conceivably define a feature level that corresponded to each version of the Xbox, with corresponding resolution, frame rate, and quality targets. While this would still leave developers targeting more than one platform, it would be orders of magnitude simpler than the current PC ecosystem.
One could argue that this trend towards a periodic refresh cycle began with the PS3 and Xbox 360. Consoles have always evolved over time, but before last-gen, those evolutions were subtle, and tended to focus on smaller form factors. The Xbox 360 and PS3, in contrast, evolved significantly between launch and retirement.
Another advantage to this model would be the ability to respond more quickly to shifts in consumer demand. A traditional console launch is an extremely heavy lift. Microsoft invests years of work into launch lineups, content creation, hardware design, marketing, partnerships, and long-term software support.
A fast refresh cycle would give Microsoft the flexibility to iterate on VR, or AR, or any emerging consumer technology without asking developers to throw huge resources at hardware they knew had little staying power. Instead of building bleeding-edge hardware every 5-8 years, Microsoft and Sony could iterate on solid (but not outrageous) technology advances every 2-3 years. Give gamers guaranteed backwards compatibility for both peripherals and software, and I think rapid-iteration consoles could be a win for everyone involved. If Microsoft is serious about this plan, it would explain why AMD has been rumored to already be working on next-gen designs for a possible 2018 launch date.
My colleague, Grant Brunner, isn't nearly as bullish on this possibility as I am. He lays out his reasoning in a different piece published this morning. Where do you stand on the concept of upgradeable consoles?
