• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

The PS5 Pro Proves AI Upscaling Is the Future

Is it really mainstream if the Pro sells about 20million?,

The Pro is just the beginning. PS6 and the next Xbox (if there is one) will be designed around AI upscaling as standard.

sounds like it's gonna take Nintendo to make it mainstream with the next Switch.

90% of multiplatform AAA game devs don't even make games for Nintendo consoles. So Nintendo is mainstream but serves a different market.

So much wrong with that post.

Oh really?

NVIDIA represents 88% of the discrete GPU market share. That's a lot more than "only one out of 3".

Nvidia's dGPU market share is irrelevant. Most of the dGPUs they make are sold to datacentres for AI and crypto. The vast majority of PCs used for gaming don't even run dGPUs. They use integrated graphics.

Nice try though. Next time come back with some actual meaningful facts to share.

Yeah, no. The PS5 Pro will sell what, 13-17M units?

Again, the PS5 Pro is just a prelude (next-gen consoles will follow suit) and 13-17m units is multiple million units more than Nvidia will sell its highest end dGPUs.

How many 4080s do you think they sell? It's like a few hundreds of thousands to a handful of million chips max.

High end dGPUs are significantly lower volume than even a niche console like the Pro.

The Pro will expand the market for high-end AI upscaling massively, and then the next-gen consoles will make it ubiquitious.
 
Last edited:

shamoomoo

Member
They don't pay anyone for DLSS. Developers are free to use it. What you are thinking about are NVIDIA-sponsored titles where NVIDIA will assist the developers with their engineers to help implement NVIDIA technologies such as RTX, Reflex, and enhanced DLSS support. There likely is an exchange of money from NIVIDA to the developer as the developer slaps big NVIDIA logos on their splash screens, effectively advertising the GPUs. AMD does the same thing.

There are over 600 games with DLSS support and tens of millions of users with DLSS-capable GPUs. Safe to say, it has been mainstream for a long time and the Pro will not reach those numbers.
AMD and Nvidia are hardware companies, they only exist to sell their products.

There's no point in having a feature set if it isn't going to be used. So,each company are kinda incentives to get developers to use their features to sell more GPUs over their competitor.
 

sachos

Member
Next step is good artifactless frame generation. The ultimate goal is full neural rendering. Some people expect to fully generate games in real time but i expect we will use some form of basic engine rendering shapes and handling all the game systems and then using real time AI rendering on top of it to draw the photorealistic (or whatever) graphics you want.
 
Last edited:

Gaiff

SBI’s Resident Gaslighter
AMD and Nvidia are hardware companies, they only exist to sell their products.

There's no point in having a feature set if it isn't going to be used. So,each company are kinda incentives to get developers to use their features to sell more GPUs over their competitor.
Yeah, and what does that have to do with what I said?
 

Stooky

Member
>Been going on on PC for 4-5 years
>The PS5 Pro which has yet to be released proves AI upscaling is the future
Yeah, sounds like an IGN article, alright.

Also:



oh boy dunk GIF
meh Its a game changer. Its another tool, like every console having a hard drive. Now for next gen every console will have a dedicated chip just for a quality upscaler. hitting 4k native not as important, opens gpu to push other things.
 
Last edited:

Gaiff

SBI’s Resident Gaslighter
Nvidia's dGPU market share is irrelevant. Most of the dGPUs they make are sold to datacentres for AI and crypto. The vast majority of PCs used for gaming don't even run dGPUs. They use integrated graphics.
Good attempt with the dismissal but no. 88% are consumer-grade GPUs, not data centers.
Nvidia's dGPU market share is irrelevant. Most of the dGPUs they make are sold to datacentres for AI and crypto. The vast majority of PCs used for gaming don't even run dGPUs. They use integrated graphics.

Nice try though. Next time come back with some actual meaningful facts to share.
Imagine being this wrong with so much confidence. NVIDIA's discrete GPU market share is irrelevant? Then who the hell has it lol?
LBJRlZs.png


77% on Steam and this includes integrated graphics, which is why we're down to 77%. This doesn't include data centers that you tried to suggest were relevant to what we were talking about. 132 million monthly active users. 77.75% NVIDIA users among them so 102.63 million. Of those, 60% use RTX 20 GPUs and above, so there are 61.6 million DLSS-capable GPUs out there. And wrong, the vast majority of computers sold have iGPUs, not the majority of gaming PCs.
Again, the PS5 Pro is just a prelude (next-gen consoles will follow suit) and 13-17m units is multiple million units more than Nvidia will sell its highest end dGPUs.

How many 4080s do you think they sell? It's like a few hundreds of thousands to a handful of million chips max.

High end dGPUs are significantly lower volume than even a niche console like the Pro.
What the fuck does that have to do with the question? DLSS is supported on RTX 2060 and above GPUs. You don't need an RTX 4080 lmao.
The Pro will expand the market for high-end AI upscaling massively, and then the next-gen consoles will make it ubiquitious.
No, it freakin' won't with its 15 million (and that's being generous) units. The PS6 isn't coming out until late 2027/early 2028 and by then, AI upscaling will be a decade old and in use by over 100 million users. Get your facts straight before spouting falsehoods.
 
Last edited:

Seider

Member
"There are over 600 games with DLSS support and tens of millions of users with DLSS-capable GPUs. Safe to say, it has been mainstream for a long time and the Pro will not reach those numbers."

PSSR is going to obliterate that number on Ps5 Pro.
 
Last edited:
"There are over 600 games with DLSS support and tens of millions of users with DLSS-capable GPUs. Safe to say, it has been mainstream for a long time and the Pro will not reach those numbers."

PSSR is going to obliterate that number on Ps5 Pro.
Is that 600 a real stat? PSSR will have more games running on it than that number in 1 to 1.5 years. At the end of the day, this will lead to graphically better PC ports so I don't understand the arguments.
 

Gaiff

SBI’s Resident Gaslighter
"There are over 600 games with DLSS support and tens of millions of users with DLSS-capable GPUs. Safe to say, it has been mainstream for a long time and the Pro will not reach those numbers."

PSSR is going to obliterate that number on Ps5 Pro.
Looking forward to those 60 million PS5 Pros sold when the PS4 Pro only managed fewer than 15 million.
Is that 600 a real stat? PSSR will have more games running on it than that number in 1 to 1.5 years. At the end of the day, this will lead to graphically better PC ports so I don't understand the arguments.
Wouldn't bet on that.
 
Last edited:

shamoomoo

Member
Yeah, and what does that have to do with what I said?
Developers were asking for AI upscaling until Nvidia and AMD had hardware capable of it. Heck , even some of Nvidia's other tech weren't used much outside of sponsored titles,like PhysX,HBAO+, some forms of anti-aliasin and shadow techniques.
 

Gaiff

SBI’s Resident Gaslighter
Developers were asking for AI upscaling until Nvidia and AMD had hardware capable of it. Heck , even some of Nvidia's other tech weren't used much outside of sponsored titles,like PhysX,HBAO+, some forms of anti-aliasin and shadow techniques.
And what point are you trying to make?
 

Gaiff

SBI’s Resident Gaslighter
You don't think incorporating PSSR and pushing graphics tech helps PC ports? I'm just talking specifically first party.
I'm talking about 600 games in a year to a year and a half. A lot of devs won't bother and will just let their games run better on the Pro. The vast majority of games aren't that demanding and PSSR will mostly help the Pro in AAA/AA titles that push the system hard and there aren't that many every year. The only way they can reach that number is if all the major players go back and implement PSSR into their older titles. Otherwise, most big and mid-tier games from now, but I doubt there will be 600 of them within 18 months. Maybe 250-300?
 
Last edited:

James Sawyer Ford

Gold Member
I'm talking about 600 games in a year to a year and a half. A lot of devs won't bother and will just let their games run better on the Pro. The vast majority of games aren't that demanding and PSSR will mostly help the Pro in AAA/AA titles that push the system hard and there aren't that many every year. The only way they can reach that number is if all the major players go back and implement PSSR into their older titles. Otherwise, most big and mid-tier games from now, but I doubt there will be 600 of them within 18 months. Maybe 250-300?

Well arguably those few AAA titles are the most important ones, I don’t think people mind much that lower budget games don’t get much improvement.

I’m buying the pro for basically maybe 10 games per year on a good year, all mostly AAA
 

Gaiff

SBI’s Resident Gaslighter
Well arguably those few AAA titles are the most important ones, I don’t think people mind much that lower budget games don’t get much improvement.

I’m buying the pro for basically maybe 10 games per year on a good year, all mostly AAA
Yeah, that's my point. There aren't that many AAA titles every year. Certainly not hundreds. Little indie games don't need PSSR and likely won't implement it and there's a big chance old titles like Watch_Dogs Legion never get patched. Moving forward, PSSR should have a rapid adoption, comparable if not better than DLSS, but DLSS had a 6-year headstart. It will take a while for PSSR to catch up in terms of raw numbers.
 

James Sawyer Ford

Gold Member
Yeah, that's my point. There aren't that many AAA titles every year. Certainly not hundreds. Little indie games don't need PSSR and likely won't implement it and there's a big chance old titles like Watch_Dogs Legion never get patched. Moving forward, PSSR should have a rapid adoption, comparable if not better than DLSS, but DLSS had a 6-year headstart. It will take a while for PSSR to catch up in terms of raw numbers.

Yeah spot on

I am curious to see over time whether PSSR can get some kind of extra improvements by being a closed platform
 
I'm talking about 600 games in a year to a year and a half. A lot of devs won't bother and will just let their games run better on the Pro. The vast majority of games aren't that demanding and PSSR will mostly help the Pro in AAA/AA titles that push the system hard and there aren't that many every year. The only way they can reach that number is if all the major players go back and implement PSSR into their older titles. Otherwise, most big and mid-tier games from now, but I doubt there will be 600 of them within 18 months. Maybe 250-300?
From the latest Edge Magazine an anonymous dev with 25 years of experience...

“Developing a PS5 Pro version isn’t a pain in the arse at all,” he says. “Why would it be? It makes it easier to use your PC version [as the basis]. It’s just like adjusting the settings. Dealing with multiple hardware variations is only difficult when you have to spend the largest amount of time tuning and optimising for the lowest-spec hardware, such as Series S. For us, allowing for PS5 Pro [with our current project] is easy to do, and it would be like that for any developer, really."

If it's truly that easy then I expect at least 75% of the games after Nov. 7 to be using PSSR and there's already close to 3,000 games released on PS5 this generation.
 
Yeah, that's my point. There aren't that many AAA titles every year. Certainly not hundreds. Little indie games don't need PSSR and likely won't implement it and there's a big chance old titles like Watch_Dogs Legion never get patched. Moving forward, PSSR should have a rapid adoption, comparable if not better than DLSS, but DLSS had a 6-year headstart. It will take a while for PSSR to catch up in terms of raw numbers.
OK not 600 games but 600 AAA games gotcha well shit I doubt we'll see 600 AAA games all generation.
 

Gaiff

SBI’s Resident Gaslighter
From the latest Edge Magazine an anonymous dev with 25 years of experience...

“Developing a PS5 Pro version isn’t a pain in the arse at all,” he says. “Why would it be? It makes it easier to use your PC version [as the basis]. It’s just like adjusting the settings. Dealing with multiple hardware variations is only difficult when you have to spend the largest amount of time tuning and optimising for the lowest-spec hardware, such as Series S. For us, allowing for PS5 Pro [with our current project] is easy to do, and it would be like that for any developer, really."

If it's truly that easy then I expect at least 75% of the games after Nov. 7 to be using PSSR and there's already close to 3,000 games released on PS5 this generation.
I don't. 90% of games probably don't even need it. The majority of games you see are the top 10%. 75% of AAA games (more like 90%)? Absolutely. Those little indie puzzle and adventure games? No.
OK not 600 games but 600 AAA games gotcha well shit I doubt we'll see 600 AAA games all generation.
Let me be a bit clearer. You previously quoted that there are 3000 games on PS5 so far, but realistically, how many of those games would even benefit or need PSSR? 150? 200? Definitely nowhere near 3000. 600 may not seem like that big of a number, but that's because these are mostly AA/AAA games, and if you factor this in, then you realize that 600 is probably the majority of AA/AAA games since 2020. Nowadays, you'll be hard-pressed to find AA/AAA games without DLSS support. I think the same will happen with the Pro. AA/AAA games and even a good chunk of A games will support PSSR, but again, those probably don't even account for 25% of the total number of games. That's why reaching 600 will be a lot longer than you think, unless, as I said before, a bunch of devs go back and patch older games.
 

JackHanma

Neo Member
In the space of 5 years Sony has brought us the innovations of high speed SSD hard drives, 3 Dimensional Audio, and proper up scaling. The next generation doesn't begin until Sony says so.
 

Zathalus

Member
I know PC gamers don't like to hear it, but bespoke tech available to only one of the 3x PC desktop GPU IHVs might be the pioneer of a technology, but they don't prove it's the future until the consoles adopt it to make it mainstream.

After all, most game sales continue to be on console and therefore consoles continue to be the primary market for AAA game devs.
There are roughly 100 million PCs out there that can run DLSS, which is a far higher market that the PS5 Pro can ever hope to sell. Not to mention around 500 games already have DLSS, with almost every single new AAA release featuring it. It’s been known as the future of rendering and upscaling for quite a while now.
 

Silver Wattle

Gold Member
Next step is good artifactless frame generation. The ultimate goal is full neural rendering. Some people expect to fully generate games in real time but i expect we will use some form of basic engine rendering shapes and handling all the game systems and then using real time AI rendering on top of it to draw the photorealistic (or whatever) graphics you want.
We'll be downloading llm training data packs for each game so the local ai processor renders things the way the Devs want.
 

pasterpl

Member
Absolutely, but this also reminds me that we have now no less than 3 AI-based upscaling solutions and 2 hardware-agnostic ones. I can't imagine devs being happy with this many. Microsoft might also develop their own eventually or go with whatever solution AMD comes up with in the future. That's going to be yet another one added onto the pile.
Microsoft already has an AI based upscaling solution.

Microsoft auto super resolution - next Xbox might be an interesting machine.
 

PaintTinJr

Member
It's not. It's just switching between one set of pixels in one frame, and then another set of pixels in another frame. In a quincunx pattern.
It has some frame blending techniques to try to mask interpolation, but it is not creating new pixels.
I think you are just arguing semantics there, AFAIK DLSS at a given version with given settings for a give game scene produces identical results on any PC, so is just as deterministic and reversible as the quincunx pattern not producing new unique pixels IMHO because the AI model isn't training in a re-entrant way.
 

winjer

Gold Member
I think you are just arguing semantics there, AFAIK DLSS at a given version with given settings for a give game scene produces identical results on any PC, so is just as deterministic and reversible as the quincunx pattern not producing new unique pixels IMHO because the AI model isn't training in a re-entrant way.

A temporal upscaler produces new pixels, extrapolated from jittered samples of neighbouring pixels.
Interlacing, even if it's CB, only shifts rendering from one frame to the next.
 

Seider

Member
I'm talking about 600 games in a year to a year and a half. A lot of devs won't bother and will just let their games run better on the Pro. The vast majority of games aren't that demanding and PSSR will mostly help the Pro in AAA/AA titles that push the system hard and there aren't that many every year. The only way they can reach that number is if all the major players go back and implement PSSR into their older titles. Otherwise, most big and mid-tier games from now, but I doubt there will be 600 of them within 18 months. Maybe 250-300?
I think we are going to see PSSR even on Ps4 games. PSSR is free to apply, easy to update old games with this technology. If Ps5 Pro is a success, then PSSR is going to be everywhere on Playstation.
 
Last edited:
Top Bottom