Predict: Next gen console tech (9th iteration and 10th iteration edition) [2014 - 2017]

Status
Not open for further replies.
In addition to that, storage isn't getting 8x faster to match 8x more RAM. We already have games taking minutes to load. Factor in download times and install times...128 GB would be unusable. If you want to use that much storage, you'll need it to be persistent to spend time just once during 'installation'. So n TB HDD, 128 GB SSD/fast flash, and 24 GBs RAM, sort of thing.
To be fair isn't that mostly on the developer? I haven't played a game on Ps4 yet that takes that a minute to load and i'm using the stock HDD. The most egregious example I can think of is Just Cause 3 and that wasn't a finished product. It's gotta be open world games that you're thinking of.

I'm in agreement with you though, some of the cost of new consoles needs to go towards faster storage rather than that much memory. I'd even settle for less gpu power for an ssd to come standard, and not bother with HDD.
 
Last edited:
I have 0 understanding of this argument, but anyone is talking about tensor cores and even Huawei is including them in their next soc.
How can they improve the experience on a console? AI? Physic? Speech recognition?
 
I think having an SSD as standard in next gen consoles who still be cost prohibitive.
My hope that the next gen consoles cone with some kind of Flash drive as standard, they could make it removable/replaceable to mitigate fears of warranty swap outs due to failure.

Flash drives now are quite performant- see the Lexar P20 32GB for a cheap flash drive that has a 400mb/sec read speed.
 
In addition to that, storage isn't getting 8x faster to match 8x more RAM. We already have games taking minutes to load. Factor in download times and install times...128 GB would be unusable. If you want to use that much storage, you'll need it to be persistent to spend time just once during 'installation'. So n TB HDD, 128 GB SSD/fast flash, and 24 GBs RAM, sort of thing.
Yeah I'd really love that flash tiered storage, and there's so much more clever things to do with this if the developers control it directly. I never liked the integrated block-level cache of hybrid drives, they are not made for throughput, it's too small, and designed to help the random IO issues of bloated OS.

Somewhere between 100GB or even 200GB limit per game would be fine, we should have 1.5TB per platter by 2019, and the increased throughput would be more than twice the launch 500GB drives. Four times everything looks easily acheivable if we have some flash to help. Base drive would be whatever single platter we can have by then, so 1.5TB?

Which got me thinking that 24GB is in fact 4 times the memory if they reserve 4GB for the OS this time around. (20 vs 5 left for games). The same goes for a 1.5TB considering the crazy OS area they reserved on the 500GB.
 
Selling a console with include fast flash drive that simply plugs into USB allowing replacements would be a pretty ballsy move. You could even have two ports side-by-side under a flap where you could add a second USB drive to double up on transfer speeds in a RAID like configuration.
 
Yeah I'd really love that flash tiered storage, and there's so much more clever things to do with this if the developers control it directly. I never liked the integrated block-level cache of hybrid drives, they are not made for throughput, it's too small, and designed to help the random IO issues of bloated OS.

Somewhere between 100GB or even 200GB limit per game would be fine, we should have 1.5TB per platter by 2019, and the increased throughput would be more than twice the launch 500GB drives. Four times everything looks easily acheivable if we have some flash to help. Base drive would be whatever single platter we can have by then, so 1.5TB?

Which got me thinking that 24GB is in fact 4 times the memory if they reserve 4GB for the OS this time around. (20 vs 5 left for games). The same goes for a 1.5TB considering the crazy OS area they reserved on the 500GB.
24gb is not happening. only if prices would decrease drastically.
but as it looks right now prices are increasing.
By 2019/20 16gb would be realistic.

internal flash memory is still not cheap enough. But maybe they add a few gigs of fast flash memory just for the os.
 
How about approaching the question the other way around? Defining specs which would be considered a good enough improvement to justify a new console... And figure out what year we will have the tech to make it possible. If it's not enough in 2019, they would probably wait a year or even two.

2021:
36GB gddr6
384bits 18Gbps for 864GB/s
5nm EUV (???)
20TF
Zen plus moar edition

If it's that late, we might get sony and MS the same year for an epic battle.
 
PS5 in 2020 with:
- zen lite 2.5ghz
- 10TF gpu
- 16 GB gddr6
- 2gb ssd
- $400

X2X in 2021 with:
- zen lite 2.7ghz
- 12TF gpu
- 18-20 GB gddr6
- 2gb ssd
- $400

Rinse repeat until eternity
 
Depends on the native resolution policy of next gen I guess, if it's baseline native 4k then 9-12 tf of power is nowhere near enough to see a substantial leap in visuals. I'm thinking at least 15 tf and above to see wow worthy next gen visuals on top of a native 4k buffer. 4k CBR tho 12 tf should be a sweet spot. But with the way Microsoft is chest beating true 4k rendering I doubt Sony would rely on CBR as much as they would want.
 
Depends on the native resolution policy of next gen I guess, if it's baseline native 4k then 9-12 tf of power is nowhere near enough to see a substantial leap in visuals. I'm thinking at least 15 tf and above to see wow worthy next gen visuals on top of a native 4k buffer. 4k CBR tho 12 tf should be a sweet spot. But with the way Microsoft is chest beating true 4k rendering I doubt Sony would rely on CBR as much as they would want.

I don't think this is going to have any bearing on anything going forwards. As no matter what MS and Sony do, most devs on next gen hardware are going to be far more likely to use CBR than leave GPU performance on the table by rendering 4k natively.

XB1X (and PS4 Pro) is a mid-gen console with all games designed for a 1.3 TFLOPs XB1 as their baseline. PS5 and XBTuo won't be in the same situation, so developers are going to take the route that provides "good enough" IQ whilst using as much of the available next-gen GPU performance to make those pixels prettier.
 
I was thinking, if a large enough SSD is still out of the question, couldn't they just have a custom SSHD? Instead of the paltry 8gb of flash in current drives they could have 128gb+. That to me makes more sense than 2 separate storage drives ; both an HDD and SSD. I wonder why something like this doesn't exist in the PC space.

Still would rather see a large enough SSD stand alone, but it's probably wishful thinking.
 
That to me makes more sense than 2 separate storage drives ; both an HDD and SSD.
Why? Two drives is more flexible, faster, and likely cheaper, or certainly no more expensive, as you'd need to pay someone to make the custom SSHD. Just stick HDD and flash off the Southbridge and accessible as two drives, and you have the throughput of both together without bottlenecking the flash performance when writing to HDD. Put the flash into an external unit (thumbstick) and you have more upgradability to your console alongside replaceable HDD. The only clear advantage to a custom SSHD that I can see is the opportunity to charge a premium for proprietary replacements.

I wonder why something like this doesn't exist in the PC space.
Because two drives is better than one. ;) If you are using 128 GBs flash, buy a 128 GB SSD and run it alongside your HDD. SSHD's have a small amount of cache because it's the optimal balance between cost and performance benefit for caching the HDD. Bare in mind everything written to cache has to be written to HDD to, so you're limiting performance. It's only there for faster reads. There's no cross-over in components so you don't save anything fusing the two discrete devices into one package.
 
I don't think this is going to have any bearing on anything going forwards. As no matter what MS and Sony do, most devs on next gen hardware are going to be far more likely to use CBR than leave GPU performance on the table by rendering 4k natively.

XB1X (and PS4 Pro) is a mid-gen console with all games designed for a 1.3 TFLOPs XB1 as their baseline. PS5 and XBTuo won't be in the same situation, so developers are going to take the route that provides "good enough" IQ whilst using as much of the available next-gen GPU performance to make those pixels prettier.
I'm certainly fine with the CBR approach of course and it does make a lot of sense for the developers for how they could most effectively utilize the power. Although I highly doubt that would be the case for the marketing department, like period. By slapping a big "True Native 4k Next Gen graphics" brand on Sony's machine it'll effortlessly win the mainstream's recognition and poses a insurmountable threat to the competition assuming PS5 is first out. Sony's got a tricky move to make now and it's gonna be difficult to balance everything out.
 
They are using better assets and effects, it's not simply an upscaling of an old game.
True, though it'll be interesting to see how far devs get with FP16 post-processing, even if it ends up being only 20% less frametime at the end of the day.

On the raster side of things, hopefully we can get to the bottom of why MSAA appears to have some sort of cliff fall on Vega at the moment, though I wonder if devs would still be interested in combining MSAA with CBR anyway, given more flexibility in doing other wacky custom things.

something something conservative raster.
 
I'm certainly fine with the CBR approach of course and it does make a lot of sense for the developers for how they could most effectively utilize the power. Although I highly doubt that would be the case for the marketing department, like period. By slapping a big "True Native 4k Next Gen graphics" brand on Sony's machine it'll effortlessly win the mainstream's recognition and poses a insurmountable threat to the competition assuming PS5 is first out. Sony's got a tricky move to make now and it's gonna be difficult to balance everything out.
They won't do that. They don't need it. They'll have plenty of compelling games to show on their boxes. It'll be enough like it's enough now for them.

Great games > flashy logos.

Anyways "4K", "True 4K", "Dynamic 4K", "Native 4K", "Checkerboarded 4K", "True native 4K"...Most customers don't know the difference. Even the XB1s is labelled as a "4K" machine nowadays.
 
True, though it'll be interesting to see how far devs get with FP16 post-processing, even if it ends up being only 20% less frametime at the end of the day.

On the raster side of things, hopefully we can get to the bottom of why MSAA appears to have some sort of cliff fall on Vega at the moment, though I wonder if devs would still be interested in combining MSAA with CBR anyway, given more flexibility in doing other wacky custom things.

something something conservative raster.

How many console games even use MSAA now? To me it's a PC thing, mostly when people want to increase image quality on older games.
 
Status
Not open for further replies.
Back
Top