Last week, my favoured gaming news site, VGC, asked former US PlayStation boss Shawn Layden whether he thought the pursuit of more powerful consoles was still the way to go for the video games industry. His answer was not what I expected.
“We’ve done these things this way for 30 years, every generation those costs went up and we realigned with it. We’ve reached the precipice now, where the centre can’t hold, we cannot continue to do things that we have done before … It’s time for a real hard reset on the business model, on what it is to be a video game,” he said. “We’re at the stage of hardware development that I call ‘only dogs can hear the difference’. We’re fighting over teraflops and that’s no place to be. We need to compete on content. Jacking up the specs of the box, I think we’ve reached the ceiling.”
This surprised me because it seems very obvious, but it’s still not often said by games industry executives, who rely on the enticing promise of technological advancement to drum up investment and hype. If we’re now freely admitting that we’ve gone as far we sensibly can with console power, that does represent a major step-change in how the games industry does business.
So where should the industry go now?
Once upon a time, nobody could ever need more than 640kB of RAM. Every “hardware ceiling” ends up being a temporary plateau. How long that plateau lasts is anyone’s guess. They’ll chase “content” for awhile, and then some form of content will demand more power for something either new or evolved, and it’ll be back to hardware races.
Either way, as long as a market exists for dedicated gaming consoles (hi, I’m that market, zero desire to maintain a PC after 25 years in IT) they’ll stick around.
I think we are entering a different era.
Once upon a time shrinking nodes came with cost reductions for the same amount of compute.
With the new bleeding edge nodes, this is not so true, you can increase compute density, but the cost of new nodes is astronomical, so prices go up too.
Many improvements recently are more architectural in nature, like zen ccds to decrease costs.
The architectural improvements will continue to scale, but node improvements are slowing, we are right on the edge of what is physically possible with silicon.
The improvements in games have slowed a ton too.
Each new generation of consoles has started to reach diminishing returns for graphics. Ray tracing seems more like a technology that is being pushed to sell hardware, rather than actually improving graphics efficiently.
The next high compute case might need more creative solutions other than throwing more compute at it. Like eye tracking for VR which reduces compute demand greatly
Ray tracing seems more like a technology that is being pushed to sell hardware, rather than actually improving graphics efficiently.
If efficiently is the key word then I agree with you. Ray Tracing is definitely still extremely expensive as far as performance goes. But I do think we’ve also seen it actually add marked improvements to the graphical impact of games.
…but does it add anything to the experience of playing the game?
It certainly doesn’t affect the gameplay. You’ll still do the same things. It doesn’t enable a new game dynamic.
All it does is push the graphic fidelity up a bit. For me a good game can be enhanced marginally by good graphics, but a bad game is a bad game even if the graphics are stunning.
I don’t think consoles can play all games at 4k60fps yet right? Especially on a TV, 4k is a noticable difference. Maybe even go for 4k120? Seems like there still needs to be a bit more improvement. Not sure if 8k is worth it, need to find someone who knows more than i do for that.
I suspect the next avenue is going to be using AI to create more realistic and interactive NPCs, especially in open-world games.
I wonder how far away we are from just putting a prompt in for the kind of game you’d like to play and it just being procedurally generated for you.
Also when are we getting haptic suit porn games where my game station sucks my dick? Best I’ve got so far is lubing up the vaccuum while I stream Baywatch on my VR headset
Also when are we getting haptic suit porn games where my game station sucks my dick?
Asking the real questions.
Even using ai for environments. Plenty of open world games will have you looking at the exact same patch of grass and the exact same trees. Even getting ray traced lighting effects to work at life-like settings, running at at least 4k90, will take a lot of computing horsepower. The price to do these things is still prohibitive, but game devs will absolutely try doing those things in the future, so the console hardware will have to keep up.
What I can see them doing is taking a more passive approach, and waiting for the hardware to develop on its own, and just using “off the shelf” parts, instead of paying to have custom hardware developed.
There are so many features that could be enabled with more powerful compute. Of the top of my head are things like improved physics simulations and clipping geometry (e.g. weapons clipping through models or characters). More detailed hit boxes. Etc
But does that make the game more fun, or does it lower the barrier of entry for smaller studios to make high-quality games?
Arguably, ray-tracing does lower the barrier to entry. You place lights where they really are in a scene, boom, everything is light perfectly. Art assets and tuning up lighting are a huge time cost in current AAA games; making that much easier might benefit gaming in general.
Having improved physics modelling might improve physics-based games, but something like Angry Birds doesn’t need a supercomputer anyway, and for most games it’s just added prettiness that greatly increases the production cost
It should go towards getting rid of fucking garbage smeary TAA and developers that don’t want to put optimizations into their game and use DLSS as a cover-all bandaid for shit.
We shouldn’t need a 4090 to play 1080p games.
