• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

The PS5 Pro Is Not Pro Enough

Did The PS5 Pro Fail As A Pro Device?

  • Yes

    Votes: 227 48.6%
  • Depends On The Game

    Votes: 107 22.9%
  • No

    Votes: 133 28.5%

  • Total voters
    467
It's strange to have to explain this again and again. But the Pro will not use 300 TOPs at once to do upscaling. It will only use a fraction of that per frame.
None of the machines on list uses all of its TOPS at once. Its the PROPORTION to each othet that is the question. PS5 Pro's INT8 TOPs is literally more than 10x of PS5"s FP16 number. Thats quite a proportional difference.
 
None of the machines on list uses all of its TOPS at once. Its the PROPORTION to each othet that is the question. PS5 Pro's INT8 TOPs is literally more than 10x of PS5"s FP16 number. Thats quite a proportional difference.

PSSR2 will never run on the base PS5, as it's too expensive. So it doesn't matter.
 
So that's why Sony jacked up INT8 capability for PS5 Pro? Just for upscaling? Oooooo...Kay....
If I had to take a guess I'd say that it is for upscaling as an immediate already proven application (even though their original approach with PSSR was lacking) but also to be able to experiment on things for PS6 which will require more throughput.
RT is handled by separate h/w unrelated to matrix ALUs. You do need a lot of shading power to denoise and use the results of tracing but INT8 won't help with that. So the only real application related to RT here would be ML assisted denoising, and it doesn't run currently that well even on RDNA4 which both supports FP8 in addition to INT8 and is overall faster.
 
how many games are worth being pro tho, is the question
VN1XtGa6WZ7sUM6g.jpg


ggPRq6Mj6UlV8JvK.gif
 
For me I'm just not seeing how this is a $1100 device. After taxes here that's 1243. Absolute insane price for the tech you're getting.
That's the entire industry though. It still holds the best value. A mid-tier PC with similar specs will cost you even more right now.
 
Ceny really fucked up with that RDNA1 features lock on PS5. Otherwise we already would have some form of ML upscaling from Sony studios, similar to XeSS at least.

He redeemed himself with pro but this console is insignificant compared to base PS5 (numbers).
At the time of the ps5 launch? I don't see how, looking at the TOPs number of Series X. To me seems Cerny was just realistic and perfectly knew AMD hadn't enough hardware to handle an AI upscaler so it wasn't wasted the budget in gimmick branded features.
 
Last edited:
At the time of the ps5 launch? I don't see how, looking at the TOPs number of Series X. To me seems Cerny was just realistic and perfectly knew AMD hadn't enough hardware to handle an AI upscaler so it wasn't waste the budget in gimmick branded features.

In 2020, AMD released CDNA with AI cores that were doing 184 TOPs of Int8.
The hardware was there, but at the time no one knew that Ai upscalers were going to be so big.
Still, if Cerny had chosen to have DP4A support on the PS5, we could have something like XeSS running on it with good performance and better image quality than FSR3.1 or TSR.
 
In 2020, AMD released CDNA with AI cores that were doing 184 TOPs of Int8.
The hardware was there, but at the time no one knew that Ai upscalers were going to be so big.
Still, if Cerny had chosen to have DP4A support on the PS5, we could have something like XeSS running on it with good performance and better image quality than FSR3.1 or TSR.
It's half of the ps5 pro TOPs number....and 2020 it's a year later of the ps5 hardware tech.
 
Last edited:
So you think have half of TOPs is meaningless? If you say.

Yes, because they are never used completely.
Also consider that the ML capabilities on the Pro are just 44 instructions added to the Shader Units. So they are deeply dependent on the CUs.
The AI Cores on CDNA1 were also dependent on the CUs, but not as much. This means that CDNA1 cores could get closer to achieve it's theoretical throughput, than the Pro, when the shader units are also in use, as there is less resource contention.
 
Yes, because they are never used completely.
Also consider that the ML capabilities on the Pro are just 44 instructions added to the Shader Units. So they are deeply dependent on the CUs.
The AI Cores on CDNA1 were also dependent on the CUs, but not as much. This means that CDNA1 cores could get closer to achieve it's theoretical throughput, than the Pro, when the shader units are also in use, as there is less resource contention.
It's always fascinating how casual user on forum read a couple of explanation on wikipedia about this stuff and they think automatically what theorically is possible always works flawless practically. Considered ps5 pro is still in an experimental stage, imagine release the ps5 a year later with this tech in such early stage with the cost consequences and the effective result.
 
Last edited:
It's always fascinating how casual user on forum read a couple of explanation on wikipedia about this stuff and they think automatically what theorically is possible always works flawless practically.

That in no way refutes what I said and confirms what you are trying to defend.
 
That in no way refutes what I said and confirms what you are trying to defend.
I'm not defending anything. I just said to you put a lot of theorically presumptiom is pointless. Neither AMD was capable to design an AI upscaler at the time but hearing you sony should achieved it easily with even less tech resources. I mean...
 
Last edited:
I'm not defending anything. I just said to you put a lot of theorically presumptiom is pointless. Neither AMD was capable to design an AI upscaler at the time but hearing you sony should achieved it with less tech available.

I'm just saying the hardware already existed, if Sony wanted it. Be it with the AI cores in CDNA1, or some variant of it. Be it with support for DP4A. The PS5 could have already had ML capabilities in 2020.
And AMD only designed the hardware. Drivers, firmware and even PSSR1 were all developed by Sony. So if Sony had decided to have Ml capabilities on the base PS5, they would also be the ones making the first upscalers.
 
I'm just saying the hardware already existed, if Sony wanted it. Be it with the AI cores in CDNA1, or some variant of it. Be it with support for DP4A. The PS5 could have already had ML capabilities in 2020.
And AMD only designed the hardware. Drivers, firmware and even PSSR1 were all developed by Sony. So if Sony had decided to have Ml capabilities on the base PS5, they would also be the ones making the first upscalers.
This tech was released a year later of the ps5 launch. Probably when ps5 was designed such tech wasn't even considered on the paper.
 
Last edited:
The tech was released a year later of the ps5 launch. Probably when it was designed this tech wasn't even on the paper.

Both CDNA1 and the PS5 were released in November 2020. So they were being developed at the same time. If Sony wanted it, they would just ask AMD.
And even if that wasn't the case, Sony could still go with DP4A.
So there were choices available. Sony decided to have none.
 
Both CDNA1 and the PS5 were released in November 2020. So they were being developed at the same time. If Sony wanted it, they would just ask AMD.
And even if that wasn't the case, Sony could still go with DP4A.
So there were choices available. Sony decided to have none.
You are free to have your convintions.
 
Last edited:
You are free to have your convintions. Already told you why it seems more obvious to choice.

Wouldn't you like the PS5 to have ML capabilities from the get go?
Even something basic like DP4A would allow Sony to have a good upscaler similar to XeSS, which is miles better than FSR3.
 
Wouldn't you like the PS5 to have ML capabilities from the get go?
Even something basic like DP4A would allow Sony to have a good upscaler similar to XeSS, which is miles better than FSR3.
Have you think for a second maybe AI upscaler wasn't even on the mind of Sony at the time because AMD wasn't exactly tempted to the AI tech, and probably they wait and see how AMD software upscaler worked before make their next move?
 
Last edited:
Have you think for a second maybe AI upscaler wasn't even on the mind of Sony at the time because AMD wasn't exactly tempted to the AI tech, and probably they wait and see how AMD upscaler worked before make their move?

That is not what I asked you.
Wouldn't you rather have something similar to XeSS or FSR3?
 
You should've bought it last year if you cared about it

Though shit

I payed around $450 trading in my original PS5 (Day 1)
Or be like me and spend $580 with no trade-ins for a 5070ti at the end of November and you could still have kept your base PS5 and gotten way better visuals in the new none exclusives. See other people can post deals that happened in the past as well.

The PS5 Pro was a turd when it came out compared to PS4 -> PS4 Pro. $699 launch price without a disk drive? Add on one year of base PS+ so you can play online for a single year and you are at $860 before buying a single game. At launch you had a handful of patches that let games get their performance mode frame rates but quality mode visuals. But even then, the games still had way worse visuals and mostly lower frame rates than a sub $1,000 PC would get you. Consoles are supposed to give you value. But the PS5 Pro didn't offer that.
 
Have you think for a second maybe AI upscaler wasn't even on the mind of Sony at the time because AMD wasn't exactly tempted to the AI tech, and probably they wait and see how AMD software upscaler worked before make their next move?

No one was thinking about upscaling other than Nvidia, and first version of DLSS was bad. But in 2020 (when consoles were in production) Nvidia released DLSS 2 that changed the landscape forever...
 
It wasn't Pro enough due to poor software implementation (PSSR). It has redeemed itself though.

And no, this isnt shitting on Sony, same goes for the 2080ti. DLSS when launched was a fucking joke, combined with the price increase, meh performance bump, and the huge price jump, the 2080ti at launch was laughed at. Thanks to DLSS improving I'd say the card has aged significantly better than the legendary 1080ti.

The PS5 Pro is now worthy of the Pro lable since the new PSSR launched.
 
Last edited:
It wasn't Pro enough sue to poor software implementation (PSSR). It has redeemed itself though.

And no, this isnt shitting on Sony, same goes for the 2080ti. DLSS when launched was a fucking joke, combined with the price increase, meh performance bump, and the huge price jump, the 2080ti at launch was laughed at. Thanks to DLSS improving I'd say the card has aged significantly better than the legendary 1080ti.

The PS5 Pro is now worthy of the Pro lable since the new PSSR launched.
People were criticizing PSSR 1.0 when it had basically been around for a year.

DLSS has been on the market since 2018.

It's like cheating at solitaire when they were making comparisons.

But it doesn't surprise me.
 
People were criticizing PSSR 1.0 when it had basically been around for a year.

DLSS has been on the market since 2018.

It's like cheating at solitaire when they were making comparisons.

But it doesn't surprise me.

DLSS as a temporal upscaler was only available in 2020.
 
People were criticizing PSSR 1.0 when it had basically been around for a year.

DLSS has been on the market since 2018.

It's like cheating at solitaire when they were making comparisons.

But it doesn't surprise me.
PSSR should rightly have been criticized from the get go. If it had launched in 2018 than for sure it would have been miles ahead of DLSS at the time, but it didn't. It launched end of 2024 and both DLSS and XeSS outclassed it at launch.

A modern upscaler doesn't have to launch in a poor state. XeSS was competent when it launched back in 2022 and FSR4 launched as an excellent example of a upscaler a scant few months after PSSR.

Fact is, the original PSSR was not very good and had numerous issues, which rightfully got pointed out. But that's all in the past. Current PSSR is really great and if anyone was to ask a question if the PS5 Pro is Pro enough today, I'd say the answer is a resounding yes.
 
At the time of the ps5 launch? I don't see how, looking at the TOPs number of Series X. To me seems Cerny was just realistic and perfectly knew AMD hadn't enough hardware to handle an AI upscaler so it wasn't wasted the budget in gimmick branded features.
Yeah you are not wrong,also PS5 has been the main platform for developers all generation,I wonder if MS said fk it about the upscaler because of that. Just some thoughts on the subject.
 
Last edited:
Come on, answer the question. Because if Sony had chosen to have support for DP4A on the PS5, this would be the reality.
Would you rather have FSR3 or XeSS?
Dude you literally changed your argument when I specifically told you what make you so sure that an AI upscaler was tech feasible and convenient by Sony in early 2020, when the whole AMD tech was quite behind and they just worked to software upscaler saying AI wasn't unnecessary or something like that. Sony doesn't manufacture gpu and they are still heavily tied to the AMD tech; highly probable they used an approach of wait and see, maybe hoping the AMD software upscaler was really sufficient, who knows and when realized AMD was just too much behind and took too much time, they started to think to their own solution with the AI.
 
Last edited:
PSSR should rightly have been criticized from the get go. If it had launched in 2018 than for sure it would have been miles ahead of DLSS at the time, but it didn't. It launched end of 2024 and both DLSS and XeSS outclassed it at launch.

A modern upscaler doesn't have to launch in a poor state. XeSS was competent when it launched back in 2022 and FSR4 launched as an excellent example of a upscaler a scant few months after PSSR.

Fact is, the original PSSR was not very good and had numerous issues, which rightfully got pointed out. But that's all in the past. Current PSSR is really great and if anyone was to ask a question if the PS5 Pro is Pro enough today, I'd say the answer is a resounding yes.
One could argue that by complaining about PSSR 1, we manifested the launch of PSSR 2. Now that's fucking teamwork.
 
Last edited:
People who want an accurate report of reality care.
The reality is just in your mind with all respect. You make appears everything easy and possible when 2020 wasn't like today for AMD. Why AMD would have wasted years and money to develop a software FSR if AI upscaler was already easy peasy in 2020. It's quite easy think such things worked like today at the time.
 
Last edited:
Dude you literally changed your argument when I specifically told you what make you so sure that an AI upscaler was tech feasible and convenient by Sony in early 2020, when the whole AMD tech was quite behind and they just worked to software upscaler saying AI wasn't unnecessary or something like that. Sony doesn't manufacture gpu and they are still heavily tied to the AMD tech; highly probable they used an approach of wait and see, maybe hoping the AMD software upscaler was really sufficient, who knows and when realized AMD was just too much behind and took too much time, they started to think to their own solution with the AI.

I changed nothing in my argument. I'm still stating that in 2020, AMD had the hardware for ML. Be it DP4A, for something similar to XeSS, or for something more advanced with CDNA1 AI cores.
The tech was there, and Sony chose not to implement it on the PS5.
Now please answer the question, would you not like to have an upscaler on the PS5 similar to XeSS, or would you prefer FSR3?
It's a simple question.
 
I changed nothing in my argument. I'm still stating that in 2020, AMD had the hardware for ML. Be it DP4A, for something similar to XeSS, or for something more advanced with CDNA1 AI cores.
The tech was there, and Sony chose not to implement it on the PS5.
Now please answer the question, would you not like to have an upscaler on the PS5 similar to XeSS, or would you prefer FSR3?
It's a simple question.
Why would not like the idea that AI upscaler wasn't in the sony mind at the time of the ps5 project? It answer to your question or you will continue to repeat to the infinite such unsensed question? Honestly I don't care to have the full RDNA 2 features and pay more for something similar to the series X when I don't see meaningful difference in performance to justify such waste of tech and money.
 
Last edited:
Or play a bunch of third party games with Pro upgrades. Why does everything have to be about first party?
Or even just play first party games that aren't Concord. You can't even play concord ffs so I'm not sure why you would "Pay $899 to play the best versions of Concord". Living rent free in people's minds even when the game doesn't exist anymore.
 
Why would not like the idea that AI upscaler wasn't in the sony mind at the time of the ps5 project? It answer to your question or you will continue to repeat to the infinite such unsensed question?

Why can't you admit that Sony made a mistake by not having ML capabilities on the PS5? Even the most basic solution, DP4A, would be e net positive for the console and was readily available at the time.
 
Top Bottom