There's a rumor making the rounds that AMD is prepping new solutions for Sony and Microsoft with an estimated 2018 release date and up to 5x improved performance-per-watt. Sony and Microsoft haven't said much about these plans (and won't, to avoid cannibalizing sales of existing platforms), but that doesn't stop us from examining them. Could we see new hardware in-market by that date?
We could -- but there are significant questions about how the market would react to it. If you plot game console releases over time, the Xbox 360 and PS3 were clear outliers, at eight years and seven years, respectively. The historical trend going back to the original Nintendo NES has been a 4-6 year cadence. The problem with a release cycle this fast, however, is that game development cycles now take longer than they used to, and you run the risk of alienating gamers who dropped significant cash on a platform they thought would last as long as the Xbox 360 / PS3 cycle did.
The other major question is one of technology. In the past, console technology tended to leap forward relatively quickly -- compare the evolution of Lara Croft from her earliest appearance on the PlayStation to the 2013 title, Tomb Raider.
The evolution of Lara Croft. Click to enlarge.
In this diagram, the first three versions of Lara are from the PS1 era, the fourth is the PS2, and 5-7 are the PS3. The last is the PS4 version of the Tomb Raider reboot. Is this last Lara better than the versions that appeared on the PS3? Objectively, yes -- the character model is more detailed, her clothes have better textures and more realistic wrinkles, and her skin tone is much improved. Keep in mind, however, that the PS3 had a seven-year lifespan, from 2006 to 2013. It's therefore fair to ask if the model of Lara Croft from the PS4 era is as dramatic an improvement over the PS3 version as the PS3 version (2006) is compared to Lara in Tomb Raider: The Last Revelation? (1999).
The answer, obviously, is no. Tomb Raider Legend's Lara has a rectangular neck, an actual square jaw, and a square stomach. Part of the problem with these kind of judgments (and one reason why the PS4 and Xbox One don't look like huge leaps over the last generation) is that there's a diminishing marginal return from adding more polygons. Compare the models below -- each one contains 10x the triangles of the previous.
The diminishing return of higher polygon counts.
The gap between 600 and 6000 triangles is much larger than the gap between 6,000 and 60,000. As improvements become more subtle and fine-grained, they become more difficult to notice. There are other ways that games can use more advanced hardware, of course, like increasing map sizes, cutting load times, or improving shadows and lighting in a title, but we see diminishing marginal returns in these areas as well.
In other words: The chance that shifting up to a new generation of consoles in 2018 will provide a greater improvement over the PS4 and Xbox One than those consoles show over the PS3 and Xbox 360 is relatively small.The integration counter-argument
There's a counter-argument to be made here, however. The PS4 and Xbox One are not particularly aggressive in terms of their technologies or hardware. Microsoft and Sony could've built more powerful hardware than they actually shipped in 2012, and they could opt to reposition themselves more aggressively on the price/performance curve with next-gen hardware. High Bandwidth Memory, or HBM, is a major game-changing technology over existing GDDR5, and by 2018 AMD should have no problem building it into an APU-style device.
Integrating HBM into an APU would offer a huge performance increase.
A console equipped with HBM could offer huge amounts of memory bandwidth -- 4-8x as much as the current PS4 or Xbox One. If Samsung and TSMC keep their foundry schedules, meanwhile, they should be building chips on 10nm technology by that point. Flash memory should also be cheap enough to use it as primary storage, skipping hard drives entirely and adopting solid-state memory for the entire console. Backwards compatibility with the Xbox One and PS4 could be easily achieved if the two companies keep an x86 CPU in their respective platforms. Even if they adopt ARM, an ARM + GCN-derived GPU would be a better emulation platform than trying to imitate Cell or the Xbox 360. Microsoft may have gotten the latter working, but Eurogamer's coverage of the capability indicates mixed results(Opens in a new window). (Fallout 3, however, apparently runs quite well(Opens in a new window)).
The other argument in favor of a short console cycle is this: Microsoft and Sony arguably missed some of the major technological trends of the last few years because they focused on coming to market in 2013 with relatively modest hardware. There's no hope of gaming in 4K on either console (at least, not in a AAA title). Neither platform incorporates 4K Blu-ray support, and neither is designed to use higher-speed NAND flash as opposed to conventional hard drives. Sony is working on a VR solution, but neither the PS4 nor the Xbox One have the hardware to really support VR experiences the way top-end PCs can. Rather than trying to shoe-horn these capabilities into existing hardware, it may be simpler to start from scratch.
The long and short is this: It's possible that Microsoft and Sony could be looking ahead to a 2018 refresh cycle precisely because the technology expected to be available at that point offers enough potential to deliver a true generational gain, even if the industry as a whole is subject to diminishing marginal return. If nothing else, a further incremental boost over the existing Xbox One / PS4 would finally deliver gains well above even the best of the PS3 / Xbox 360 era. 60 FPS @ 1080p would be an easy target for either company, and 30 FPS @ 4K should be plausible, even on relatively limited console hardware.
Given the source(Opens in a new window) of this rumor, we'd take it with a huge grain of salt -- but it's not impossible that Microsoft and Sony are working on new hardware for a 2018 - 2019 timeframe.
