Predict: Next gen console tech (9th iteration and 10th iteration edition) [2014 - 2017]
- Thread starter Shifty Geezer
- Start date
- Status
insanepunkclown
Newcomer
If the GPU for 9th gen are on mature 14nm then expect 2.5 more TFLOPS maximum for same price point(adjust it with inflation how much 400$ in 2013 would be in 2017-2018) which would be 440-450$.
Next gen consoleswill need the power not only for a generational leap in graphics, but also the ability to show off that graphical improvement at 120fps and the ultra high HD resolutions that is required for virtual reality systems. The PS4 had approximately ten times the GPU flops as the PS3 GPU. As we go up in graphics power, the law of diminishing returns kicks in so we will need at least 10 times the flops or even more for a similar gain in graphics. If the PS4 is 1.8 tflops, the PS5 gpu will need to be 18.0 tflops, minimum. Lets round that to 20 tflops. Some games today only run at around 30fps, so to increase that to a minimum of 60fps for VR (then hoping that some built in chip in the unit may add frames or something to smooth out judder) we need 40tflops. I would also add the extra power needed for two 1080p displays, but I'll keep it at 40tflops to stop from being crucified.
At a minimum, the PSV, to be a true generational leap with VR, needs 40tflops of processing power. This means that the GPU would need to be manufactured sub 10nm, probably even at 7nm or less. I also think the GPU will need to use more power than the PS4 GPU. Instead of a 100 watt budget for the console, maybe a 150 watt budget could be adopted.
Basically, we are talking about some serious GPU horsepower. But it is required if we want a graphics boost as visually apparent as the PS3-PS4 transition and high quality VR at the same time.
I don't see any way around this, unless Sony and Microsoft give up on a significant graphics bump and only allow for slightly better than today's quality games in VR.
I should also say that better bandwidth and other improvements other than flops could somewhat reduce the GPU flop total needed. But if we add a resolution boost to the 40 tflop figure, such bandwidth benefits are negated. Basically...
Good Graphics Leap - 20tflop GPU
Good Graphics Leap w/ high frame rate - 40 tflop GPU.
All of Above With Higher Resolution for VR - 50tflop? 60tflop?
Dedicated Hardware To Add Frames To Reduce Judder - Back down to 40 tflop?
My guess is the next generation of consoles.
2020 Launch
7nm 40tflop nVidia or AMD GPU
Integrated CPU on same chip
32gigs RAM
Totally VR Focused
Dedicated Hardware To Reduce GPU Burden
At a minimum, the PSV, to be a true generational leap with VR, needs 40tflops of processing power. This means that the GPU would need to be manufactured sub 10nm, probably even at 7nm or less. I also think the GPU will need to use more power than the PS4 GPU. Instead of a 100 watt budget for the console, maybe a 150 watt budget could be adopted.
Basically, we are talking about some serious GPU horsepower. But it is required if we want a graphics boost as visually apparent as the PS3-PS4 transition and high quality VR at the same time.
I don't see any way around this, unless Sony and Microsoft give up on a significant graphics bump and only allow for slightly better than today's quality games in VR.
I should also say that better bandwidth and other improvements other than flops could somewhat reduce the GPU flop total needed. But if we add a resolution boost to the 40 tflop figure, such bandwidth benefits are negated. Basically...
Good Graphics Leap - 20tflop GPU
Good Graphics Leap w/ high frame rate - 40 tflop GPU.
All of Above With Higher Resolution for VR - 50tflop? 60tflop?
Dedicated Hardware To Add Frames To Reduce Judder - Back down to 40 tflop?
My guess is the next generation of consoles.
2020 Launch
7nm 40tflop nVidia or AMD GPU
Integrated CPU on same chip
32gigs RAM
Totally VR Focused
Dedicated Hardware To Reduce GPU Burden
40 TF isn't possible unless either it's an insanely big, expensive console, or they wait 15 years before release.
Then we cant have a significant leap in graphics and VR. I think they will wait until they can have both. It does not make sense to invest in a console that cannot do both.
By the way, for the wonder of a console with GREAT near photorealistic VR, I don't think 1,000 dollars for a console and a headset would be outrageous.
I know. For a product that sales en masse, $1000 is likely too steep. As ever with discussion with you, the answer is for you to go buy $1000 PC and keep it up to date with hardware. Your graphical interests aren't going to be satisfied by cheap consoles.
The PS5 might not be a console, exactly. It could be a VR system. And 1000 dollars is not overly steep for what excellent VR could provide. Really good VR could become as addictive as a drug. Imagine being able to go on virtual vacations, anywhere. Or be in a community like Playstation Home with a virtual body. This goes beyond pure gaming. If you had a fast enough data feed and cameras, you could communicate with relatives far away and be there in real time.
VR only makes computing even more complicated. You have more settings and more to figure out. I think a VR box is the way to go.
VR only makes computing even more complicated. You have more settings and more to figure out. I think a VR box is the way to go.
Let's wait a few years to see how well VR will be adopted and what kind of support will get from non-gaming applications. For now I think that HDTV-based gaming will still be very big in 2019/2020 and beyond.
I don't quite understand all this talk of a PS5 having to be 20Gflops or whatever to be worth the leap. There are PC gamers today playing the current games at much higher settings and running a lot better - so I don't see why people get hung up on 'needs to be 10x PS4' (FTR I thought PS4 6-8x PS3?).
They will just need to cram the best they can into a box in 3-4 years time and that will have to do, it will be good enough for a noticeable improvement (don't forget TVs are getting bigger all the time) and run morpheus 1080p with nice effects and no framerate issues.
They will just need to cram the best they can into a box in 3-4 years time and that will have to do, it will be good enough for a noticeable improvement (don't forget TVs are getting bigger all the time) and run morpheus 1080p with nice effects and no framerate issues.
I don't think people would be satisfied if nextgen Morpheus has only 1080p screen. The starting point of Vive and Rift is 1080*1200 x2, and that resolution will easily get bumped several times over the next few years.
I presume the target for both Vive and Rift will be to as quickly move to two QXGA [2048x1536 4:3] screens or two QSXGA [2560x2048 5:4] screens.
I presume the target for both Vive and Rift will be to as quickly move to two QXGA [2048x1536 4:3] screens or two QSXGA [2560x2048 5:4] screens.
754,974,720 pixels per second at 120 Hz stereoscopy. That's a 6x increase over 1080p60. Scaling linearly, we'd need ~12 TF to keep the same graphics as now on future VR, and ~150x6 GB/s so about 1 TB/s bandwidth. To keep current quality.
There's really no point trying to decide on future specs to hit a VR quality mark. The deciding factor for any product is always economies. What product can you make for the price people will pay to make a successful business? If the available technology won't allow 20 TF and 1 TB/s bandwidth at the intended price, the consoles will aim lower and just compromise the experience. This is true of every mainstream product - the only no-compromise, best possible product is the super high end with a price tag to match.
That HQ setup sounds reasonable and achievable for 2019/2020 time frame. But we already have first hints of VR headsets that incorporate eye tracking. If successfully implemented, VR can switch to foveated rendering which will reduce sysreq possibly by a lot.
I think there is little chance that in 3-4 years we will still "brute force" render entire VR screen.
Last edited:
these TF estimations assume all pixels will be treated equal going forward. If we are sure VR will drive what we get predicting new software techniques to maximise VR pixel quality where required / seen, or screen technologies to deliver a better screen / lens without having to massively increase pixels / GPU load is just as need for estimating the resulting experience.
We had 1080p last gen but this gen a lot of games don't even hit that (and people seem more than happy with sub 1080p). I'm just suggesting that it need not be ultra high resolution if you can add loads of effects and have stable very high frame-rate.
That sounds terrible
Personally I don't think VR is most suitable for the hardcore or even core gamers, to me it's a niche feature at best. Can you actually imagine playing a game like COD or Bloodborne using VR for hours? You'd be long exhausted before you make past the first level. Yes it does sound like a perfect hardware for Quantic Dream games or some other experimental games, but I'm sure people would forgive if they're running at a lower res when in VR mode. I believe the main focus should be catering for 1080p games perhaps in 60fps with as many flops as possible. 4k is just too much to ask, it's 4 times the pixel count and to see generational graphics leap on top of that is not economical. 1080p gaming is already good enough in the living room where you're sit in a distance, it is at this well resolved high res that any further graphical improvement would be much more beneficial. Yes, I believe rendering 1080p with 20 teraflops of gpu is the right balance of resolution and pixel quality going forward.
I believe even a gpu twice as powerful as X390 should provide a reasonable leap given the flop count with new architecture and nextgen memory tech.
I believe even a gpu twice as powerful as X390 should provide a reasonable leap given the flop count with new architecture and nextgen memory tech.
Not gonna happen. The trend is 9000x9000 + 9000xAA. And it doesn't matter how game looks. And 60 fps.
Trend seems to be 9000p (actually being only 1000x9000, since only vertical resolution matters). Real AA with multiple samples is for pussies, a fancy post AA that makes the image look like vaseline is preferred because that is more cinematic. And 60 fps looks like cheap home made videos, 30 fps makes it so much better. You got other things right
Prophecy2k
Veteran
What's an X390?
And I also can't see much of a problem with playing bloodborne or COD in VR for hours. There's nothing about VR that precludes using the controller for movement input.
I think next-gen console HW will be designed to pack in as much processing grunt as is economical, given the end pricing, power, size and cost (i.e. BOM) contraints that the platform holders place on themselves.
An APU with 18-20 TFLOP GPU, full fat 4-8 core Zen-based CPU, with unified HBM would be a reasonable and safe assumption.
The VR perfromance of the given design will simply fall out of whatever they can get outta that, and will dictate whether they launch the box alongside an updated Morpheus2.0, or not.
- Status