Sony PlayStation 5 Pro

[I'mDudditz!]

So you can explain these results then and why there isn't a 'dramatic' increase in performance despite moving upscaling to AI hardware and away from compute.

Spoiler: Charts

View attachment 11001

View attachment 11002

EDIT: I've been really kind and run the CP2077 benchmark with both FSR2 and DLSS at matched quality levels for you. So once again, can you please explain why there isn't a 'dramatic' increase in performance despite moving upscaling to AI hardware and away from compute.

FSR2

View attachment 11003

DLSS

View attachment 11004

EDIT #2: Alan Wake 2 - Explain why there's no 'dramatic' increase in performance despite moving upscaling to AI hardware and away from compute.

FSR2

View attachment 11005

DLSS

View attachment 11006

Once again you have added absolutely nothing to this discussion or forum and have ben made to look extremely foolish.

Do you not understand the graphs you are posting.? No way you are this incompetent. You need to compare the native rendering to FSR2 and DLSS. You don't compare FSR2 and DLSS. And FSR2 doesn't use hardware based AI upscaling.

Just when I thought you couldn't be more foolish.

 

[davis.anthony]

Do you not understand the graphs you are posting.? No way you are this incompetent. You need to compare the native rendering to FSR2 and DLSS. You don't compare FSR2 and DLSS. And FSR2 doesn't use hardware based AI upscaling.

Just when I thought you couldn't be more foolish.

Native rendering has nothing to do with what I said, the AI based upscaling won't improve performance over existing upscaling, only the IQ.

It's really not a difficult concept to grasp but somehow you managed to not get it.

You also need to read what you wrote again,

What the heck are you talking about? Of course it increases performance quite dramatically by relying on AI upscaling instead of compute. Just like DLSS.

You stated that it shows a dramatic improvement moving from compute based upscaling (like FSR) to AI based (like DLSS) only to then have that claim shut down by my charts that show, that actually, there is no 'dramatic' performance increase moving from compute based upscalers (like FSR) to AI based (like DLSS) despite your claim.

So just to dumb it down for and make it as simple as I can.

PS5 Pro ML based upscaler will not allow for extra performance over the existing upscalers used on base PS5, only increased image quality like we have seen on PC for years.

 

[I'mDudditz!]

Native rendering has nothing to do with what I said, the AI based upscaling won't improve performance over existing upscaling, only the IQ.

It's really not a difficult concept to grasp but somehow you managed to not get it.

The purpose of AI upscaling over something like FSR2 isn't to improve performance. Both solve the performance issue. The difference is obviously in the image quality. So when you say AI upscaling will "only" improve image quality, your statement sounds very useless and ill-informed.

 

[davis.anthony]

The purpose of AI upscaling over something like FSR2 isn't to improve performance.

That's not what you said...

What the heck are you talking about? Of course it increases performance quite dramatically by relying on AI upscaling instead of compute. Just like DLSS.

It's pretty easy to see you claim that AI upscaling (like DLSS) gives a 'dramatic' increase in performance over compute based upscalers (like FSR)

Do me and everyone else a favour, decide what you're argument is so we can debate it, because as it stands you're contradicting yourself.

 

[Remij]

The purpose of AI upscaling over something like FSR2 isn't to improve performance. Both solve the performance issue. The difference is obviously in the image quality. So when you say AI upscaling will "only" improve image quality, your statement sounds very useless and ill-informed.

In this very post you say the purpose of AI upscaling isn't to improve perf over any other upscaling technique... then right after you argue that he's wrong for saying it will only improve image quality....

So the "performance issue" was already solved by FSR2, as YOU said... and the AI upscaler is meant to solve the image quality issue.... which is exactly what HE said.


But the issue he has, is that you said it would improve performance over solutions like FSR2.. because it's now "hardware accelerated" like DLSS. He showed however, that there isn't much difference in performance between FSR2 running on the shader cores, and DLSS running on the tensor cores.

 

[Cyan]

I don't think Microsoft should respond to Pro with another machine.

They've been absolutely wrecked by Sony this generation in terms of hardware sales and from that point of view the generation is already over.

So why waste money on a 3rd SKU when it's not really going to allow them to recover some ground?

Best attack would be to price cut Series-X as much as possible (which might actually be less of a cost than the R&D cost of a new machine) to get as many machines installed as possible and look to start next generation a year earlier than Sony.

1000% agreed, it would a waste of money and resources just to have better RT and nothing else to go with. It made sense with the Xbox One X, but Series X is plenty powerful, and the power battle doesn't matter here, it's a lost battle either way. So yeah, let them take their time to rethink another machine, but a future one with the Surface team.

 

[Remij]

I'll just add.. we all know that the performance improvement primarily comes from the reduction of shader workload (resolution).. NOT the efficiency of one upscaling algorithm over the other. The algorithm is always an additive cost. DLSS is something like ~1.5 to 2ms or something like that, right? Sounds like Sony's implementation will be similar to start with.. and in the end, even if they improve it drastically and cut it down to 1ms... we're still talking about only a 1ms improvement max.

It's unlikely that Sony's solution will provide better image quality than DLSS within the same frametime budget while simultaneously utilizing tangibly lower input resolutions. I think that's asking too much at this point. Not that it couldn't eventually get there in the future.. but for now.. it's doubtful.

 

[I'mDudditz!]

In this very post you say the purpose of AI upscaling isn't to improve perf over any other upscaling technique...

Yes, when he started to compare FSR2 vs AI upscaling and not Native vs AI upscaling. He is confusing DRS which is native rendering with FSR2 as shown here:

PS5 is already upscaling from very low resolutions.

DRS doesn't mean all PS5 games upscale using FSR2.

 

[techuse]

Because the very low rendering used in these tests and game performance modes in general is responsible for the poor image quality, not FSR or any upscaler per se!

Again, there is good evidence that FSR2 gives very good results in terms of image quality in practice in the Starfield game on Series X. It uses 3.7 megapixels to create something that closely resembles an 8.3 megapixel image, all with very little resource waste! Obviously, the quality sucks if you only have to work with, say, 1.5 megapixels... Furthermore, it's not a switch that you press and the image quality is ready, hello! It all depends on the correct implementation.

After what I've seen (a well-implemented FSR2), I'm so convinced of this that I'd question the need for the entire PSSR if it didn't involve unique frame generation. With frame generation, you can already have an advantage in a new hardware.

However, it would be foolish to think that MS and AMD won't have an answer for this...

FSR 2 quality looks far worse than IGTI performance. FSR 2 looks like absolute crap in Starfield. Perhaps you just aren’t bothered by IQ issues.

 

[davis.anthony]

Yes, when he started to compare FSR2 vs AI upscaling and not Native vs AI upscaling. He is confusing DRS which is native rendering with FSR2 as shown here:

DRS doesn't mean all PS5 games upscale using FSR2.

What nonsense are you even trying to talk now, I'm not confusing DRS at all thank you very much.

• Alan Wake 2 performance mode renders at 847p fixed resolution (Digital Foundry), which is a very low resolution
• Immortals of Aveum, 720p fixed resolution (Digital Foundry), which is a very low resolution
• Remnant 2 performance mode, 720p fixed resolution (Digital Foundry), which is a very low resolution

So where exactly is the confusion with DRS??

 

[Globalisateur]

Overall I think FSR2 simply isn't even trying to improve the IQ on the whole. It just improves some parts (higher perceptible resolution for textures and clean geometry) and worsen others parts (everything else). For me in the end the balance is highly negative. Like others techniques before (like quincunx used on PS3) it will soon die from its own incompetence and nobody will ever use it again.

 

[Remij]

Yes, when he started to compare FSR2 vs AI upscaling and not Native vs AI upscaling. He is confusing DRS which is native rendering with FSR2 as shown here:

No, he didn't confuse anything. What he said was that FSR2 and DLSS have similar frametime costs.. the fact that one is on shader cores and the other on hardware accelerated tensor cores is irrelevant. He said consoles already run at very low internal resolutions, so you're not going to get any tangible performance improvements with one upscaling algo vs the other. FSR2 and DLSS use the same input resolutions at the same qualities..

Thus you can expect similar performance... but as he said, with improved image quality.

 

[DavidGraham]

DLSS is something like ~1.5 to 2ms or something like that, right?

It used to be 1.5ms on a 2080Ti, it's much much faster on a 3090 or a 4090 now.

 

[metacore]

"no per title training needed"
"only accessible as library"
"replacement for game existing TAA upsamling"

That's exactly what doctor ordered in todays lazy devs environment (pardon I could not restrain myself )


Even with only 45% gpu performance uplift I'm quite optimistic for 60fps modes. Games targeting stable 30 for base console are running at about 40, unreal 5.4 cpu improvements, series s still on the table... It's not like this gen games are that hopelessly CPU limited, heck even lats gen internet said cpuuus... and majority of that consoles performance modes had resolution downgrades suggesting in fact more gpu/bandwidth bottleneck.

To me this might pan out much better than ps4pro for its purpose.

 

[iroboto]

Do you not understand the graphs you are posting.? No way you are this incompetent. You need to compare the native rendering to FSR2 and DLSS. You don't compare FSR2 and DLSS. And FSR2 doesn't use hardware based AI upscaling.

Just when I thought you couldn't be more foolish.

I spoke too soon it seems. My general advise here is to come back to a common ground on what is being discussed before you start getting prickly.

 

[Mitchings]

This would've been my perfect midgen lineup:

 

[Shifty Geezer]

Apologies for I'mDudditz's trashy attitude and well done for not rising to it. A 7 day reply ban is in effect to keep this channel clear of noise. Haven't time to edit out the attitude and I'd appreciate if people can carry on without responding further so I don't need to; conversation stability returns naturally, we hope.

 

[Remij]

This would've been my perfect midgen lineup:

I like the idea of an optional "universal optical drive". They should have done that for this generation IMO. PS5, PS5 Slim, and PS5 Pro, could have all been designed to be as small, compact, and cheap as possible.. and the "physical SKUs" of each of them could have simply bundled the add-on drive with it.

One of the biggest issues I've had with consoles over the years have been disc drives. LOL anyone who's old enough remembers that (at least the early) PS1's sometimes had to be tilted at an angle, or flipped upside down to read discs. PS2's disc read errors were famous.. and so on. Being able to use the same drive across multiple SKUs (Reg, Slim, Pro) would be nice.. as would being able to simply replace the drive if it fails.

MS had the right idea with the HD-DVD drive add-on

 

[davis.anthony]

I do wonder if Sonys ML upscaling solution is based on rumoured AMD solution that's releasing end of this year.

Maybe they worked on it together (maybe even with feedback from Sony studios) and Sony have tweaked it slightly to be a little more performant in a closed box, or it's something completely different.

But it would make sense for it to be shared technology and has many benefits for developing collaboratively such as shared development costs.

 

[Remij]

I do wonder if Sonys ML upscaling solution is based on rumoured AMD solution that's releasing end of this year.

Maybe they worked on it together (maybe even with feedback from Sony studios) and Sony have tweaked it slightly to be a little more performant in a closed box, or it's something completely different.

But it would make sense for it to be shared technology and has many benefits for developing collaboratively such as shared development costs.

Seems very likely. It can't be a coincidence that AMD are ready to start talking about their solution coming for the end of this year.. at the same time as the PS5 Pro is expected to launch.

Once Sony officially announces the Pro and what its capabilities are, I'm sure AMD will be able to finally speak about their implementation.