• 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.

[Digital Foundry] New PS5 Pro GPU details emerge - including a 2.35GHz max boost clock

saintjules

Gold Member
Creating trailers is a massive workload for teams and games are already taking longer to develop.

The returns aren't worth it far in advance. Again they have the metrics on this.

While it would be my preference to have the old ways, times change.

I think I recall hearing/watching from TLOU2 Grounded that their trailers/reveals for one of the bigger events took like a year or something just to have it ready. Then there's other devs taking 6 months. Wild amounts of time nonetheless.
 
Yeah, developers graphs are one thing and actual performance is another. Avatar is one of those games where 6700 was faster than PS5, so no secret magic rendering here:

mtpRWLA.jpeg


It performs close to PC code despite not using PS/MS on PC. This game also runs good on RDNA2/3 so it's not typical Nvidia vs. AMD difference in RT (like Cyberpunk).
Avatar is a Ubisoft game. Last time I read an interview about console development they were saying they woud not use any bespoke hardware on any machine in order to keep their engines compatible between platforms.
Both PS5 and XBSX are early RDNA2 designs before AMD finalized RDNA2.
You can see the difference by looking at the CUs.

iImgo2M.jpeg


Regardless of how you guys feel, internally AMD, Sony and Microsoft had codename for these chips.
PS5:
Navi10 Lite = GFX1000/1001
Navi12 Lite = GFX1013/1014 (Major Revision)

XBSX:
Navi21 Lite = 1020

The GFX number can tell you where the chip lies in terms of development time frame.
KtQBUuy.jpeg


Notice RDNA1 GFX is in the 101X range while RDNA2 is in the GFX 103X range.

At the end of the day, the API is the determining factor the makes these chips different.
Yep XSX is not "true" RDNA2. Both PS5 and XSX are RDNA1.5
 
Last edited:

SlimySnake

Flashless at the Golden Globes
Avatar is a Ubisoft game. Last time I read an interview about console development they were saying they woud not use any bespoke hardware on any machine in order to keep their engines compatible between platforms.
actually, they use mesh shaders/primitive shaders on consoles and not on PC.
 

Loxus

Member
What exactly does this clear up?
Did you even read?

It clears up PS5 RT IP 1.0 vs RDNA2 RT IP 1.1 and possibly what RT IP 3.0 and RT IP 4.0 could be.

RT IP 1.0 = PS5
RT IP 1.1 = Xbox and RDNA2 (Minor bug fix)
RT IP 2.0 = RDNA3
RT IP 3.0 = RDNA3.5?
RT IP 4.0 = PS5 Pro?
RT IP 5.0 = RDNA4
 

ChiefDada

Gold Member
Did you even read?

It clears up PS5 RT IP 1.0 vs RDNA2 RT IP 1.1 and possibly what RT IP 3.0 and RT IP 4.0 could be.

RT IP 1.0 = PS5
RT IP 1.1 = Xbox and RDNA2 (Minor bug fix)
RT IP 2.0 = RDNA3
RT IP 3.0 = RDNA3.5?
RT IP 4.0 = PS5 Pro?
RT IP 5.0 = RDNA4

Yes, I read the tweet and his first words state that he has "no clue" on the technical differences between 1.0 and 1.1. And he goes on to say his guessing on the missing 3.0 and 4.0. Don't understand why you're getting so defensive about this but I appreciate you sharing all the same.
 

Loxus

Member
Yes, I read the tweet and his first words state that he has "no clue" on the technical differences between 1.0 and 1.1. And he goes on to say his guessing on the missing 3.0 and 4.0. Don't understand why you're getting so defensive about this but I appreciate you sharing all the same.
He goes on to say "AMD almost exclusively uses terms like .1 for minor bug fixes for partially unintended behavior." which is true.
So now we know there isn't any performance difference between 1.0 vs 1.1.

I don't think you've been reading this thread anyway. Cause you would have known that this clears up the small 1.0 vs 1.1 debate in this thread.
 
Last edited:

rnlval

Member
Avatar is a Ubisoft game. Last time I read an interview about console development they were saying they woud not use any bespoke hardware on any machine in order to keep their engines compatible between platforms.

Yep XSX is not "true" RDNA2. Both PS5 and XSX are RDNA1.5
Wrong, XSX's GPU supports DirectX12 Ultimate. AMD's Phoenix APU's RDNA 3 IGP is missing PC's discrete RDNA 3's infinity cache (L3 cache for GPU).

AMD's Rembrandt APU's RDNA 2 IGP is also missing PC's discrete RDNA 2's infinity cache (L3 cache for GPU).

AMD's incoming Strix Halo APU's IGP includes the 32 MB L3 infinity cache.
 
Last edited:

rnlval

Member
Did you even read?

It clears up PS5 RT IP 1.0 vs RDNA2 RT IP 1.1 and possibly what RT IP 3.0 and RT IP 4.0 could be.

RT IP 1.0 = PS5
RT IP 1.1 = Xbox and RDNA2 (Minor bug fix)
RT IP 2.0 = RDNA3
RT IP 3.0 = RDNA3.5?
RT IP 4.0 = PS5 Pro?
RT IP 5.0 = RDNA4
Both RDNA 2 and RDNA 3 have "BVH4 with 1 triangle" RT core design, hence the raw RT power is about the same. Path trace's RT saturation shows RX 6800's 60 CU and RX 7800 XT's 60 CU being similar.

RDNA 3's RT core has an early BVH culling, BVH transverse flags, and 1.5X RT instructions in flight improvements. RDNA 3's RT improvements should been in RDNA 2.
 

Ev1L AuRoN

Member
We only had one generation of pro consoles so far so in comparison to PS4 pro and XOX this console doesn't look that amazing:

PS4 1.8TF -> PS4P 4.2 - 2.3x more powerful

XB1 1.3TF -> XOX 6TF - 4.6x more powerful

PS5 Pro looks to be over 1.6x more powerful on paper and 1.45x in real life.

Jump in image quality between XB1 and PS4 games and PS4 pro and XOX was MASSIVE. Many Xbox one X games have higher resolution than current gen games. 2160cb on pro also looked excellent on 4k tvs vs 1080p on standard PS4.

This power jump isn't very impressive after 4 years (we had PS4 pro after 3)...
It was way easier to outperform the 8th gen. I bought a RX 580 under 200, and it was arguably even better than the One X.
 
I feel bad for today's generation of gamers, being a kid in the 80's and 90's was wild in terms of how quickly things progressed, in the time the PS5 has been out, we went from the Super NES to the Playstation, and had SuperFX chip and all sorts of other wild shit in-between. Now we wait 4 years and £200 more and have to watch a digital foundry video to spot the difference.

Don't get me wrong, I knew diminishing returns was a thing for the past few generations, but I think the PS5 Pro has really hammered home just how bad it's going to get. Not even throwing shade on Sony, the machine is probably the best you can put in a box for that price right now, but seeing "Pro" games that still have to run at a sub-1440p base resolution to get 60fps is downright depressing.

If you don't mind forking over serious cash for a monster PC you can probably get that same wow factor again, but with GPU prices being higher than inflation, and energy costs rising, even that is getting less attainable for casual hobbyists.

I miss being hyped about hardware.
 

Gaiff

SBI’s Resident Gaslighter
Didn't bugaga say the ps5 pro performance would be in the range of a 3060 or nvidia 2070ti???

🤣😂🤡
Well, the 2070 Ti isn’t a thing, so no. You probably mean the 2070 Super. As far as I’m aware, Alex never implied that the 3060 would be on par with a Pro since he did a few tests pitting the regular PS5 against the 2070S and they often were a close match with the PS5 pulling way ahead in exclusives. He might have said the 3060/2070S are close to a base PS5 and this isn’t too far off.

Rich also said that the Pro is closest to a 3070 Ti minus the VRAM where the 3070 Ti falls short. Maybe that’s what you’re thinking about?
 

Bojji

Member
Didn't bugaga say the ps5 pro performance would be in the range of a 3060 or nvidia 2070ti???

🤣😂🤡

Normal PS5 is above 3060 and around 2070S - 6700 range, Alex said that himself many times so I don't know what you are talking about.
 
Last edited:

mckmas8808

Mckmaster uses MasterCard to buy Slave drives
I agree with MLID here, Sony is underselling Pro performance for whatever reason. I'm willing to bet my account SM2, Ratchet, and TLOU are running circa 100-120fps uncapped in those new performance modes.



I think we all know why.....
 
I agree with MLID here, Sony is underselling Pro performance for whatever reason. I'm willing to bet my account SM2, Ratchet, and TLOU are running circa 100-120fps uncapped in those new performance modes.


They don't want to set unrealistic expectations that would have fed the warriors for years and that would have created years of (easy) negative coverage by the likes of DF : "Again PS5 Pro falls short of the 36Tflops specs". "Again this PS5 Pro game is clearly not showing 3x better RT".

They were right as DF can only concede with positive coverage that the results are better or same than what Sony first stated. Cerny played chess 4D against Microsoft shills and won. Because he is a genius unlike them.
 

Bojji

Member
They don't want to set unrealistic expectations that would have fed the warriors for years and that would have created years of (easy) negative coverage by the likes of DF : "Again PS5 Pro falls short of the 36Tflops specs". "Again this PS5 Pro game is clearly not showing 3x better RT".

They were right as DF can only concede with positive coverage that the results are better or same than what Sony first stated. Cerny played chess 4D against Microsoft shills and won. Because he is a genius unlike them.

Cerny is a good engineer but he is not a god of something.

Pro will perform like 16TF RDNA2 GPU in raster games, we know that since first leaks. There is no hidden power here outside of RT uplift that will be greater than 2x - but as we all know that not all games use RT.

You going to cry broken port but post terrible 3rd party ports that don't utilize the PS5's I/O | best in class data streaming?

🤣

Even some Sony games don't utilize PS5 I/O, is that your measure of good PS5 version of the game? You also have one broken ass port (that received many, many patches and don't look like what you posted anymore) vs. games with top of the line graphics, even with hardware RT. ND is stuck in the past with their games.
 
Last edited:

Zathalus

Member
Lol, you seriously use this broken port and in day one version to that?

2070S is crippled by low memory amount but in raw power it's still around PS5

x5QfzTp.jpeg
D7WEcQr.jpeg
RTv09Rf.jpeg
pf4RcrG.jpeg
MZAZKXh.jpeg
Don’t bother, Sony diehards will only bring up bad Sony ports as the true power of the PS5. Despite other Sony ports like Ghost, Returnal, and Horizon having none of these issues. Or literally every other single game in existence.
 

S0ULZB0URNE

Member
Avatar is a Ubisoft game. Last time I read an interview about console development they were saying they woud not use any bespoke hardware on any machine in order to keep their engines compatible between platforms.

Yep XSX is not "true" RDNA2. Both PS5 and XSX are RDNA1.5
Saying they aren't RDNA 2 was ban worthy at one point.
Id watch it.
 

S0ULZB0URNE

Member
Cerny is a good engineer but he is not a god of something.

Pro will perform like 16TF RDNA2 GPU in raster games, we know that since first leaks. There is no hidden power here outside of RT uplift that will be greater than 2x - but as we all know that not all games use RT.



Even some Sony games don't utilize PS5 I/O, is that your measure of good PS5 version of the game? You also have one broken ass port (that received many, many patches and don't look like what you posted anymore) vs. games with top of the line graphics, even with hardware RT. ND is stuck in the past with their games.
It is my measure as it uses what PC's with 2070S can't match unlike ports that are built for all types of variables.

Pro is over 30TF's and compared to what??? The 29TF 4070.
 

Gaiff

SBI’s Resident Gaslighter
Don’t bother, Sony diehards will only bring up bad Sony ports as the true power of the PS5. Despite other Sony ports like Ghost, Returnal, and Horizon having none of these issues. Or literally every other single game in existence.
And hilariously never bought up DLSS with those 2070/3060 comparisons, but they won’t shut up about PSSR.
 
Last edited:

Bojji

Member
It is my measure as it uses what PC's with 2070S can't match unlike ports that are built for all types of variables.

Pro is over 30TF's and compared to what??? The 29TF 4070.

You see that 35TF AMD GPU is weaker than 29TF Nvidia GPU?

jDld9su.jpeg


Pro is rumored to have 33.5TF. 4070 is measured from lower clock, it actually runs ~2700MHz in games so it's ~32TF, yet still performs 13% better than 35TF AMD GPU.
 

Killer8

Member
I agree with MLID here, Sony is underselling Pro performance for whatever reason. I'm willing to bet my account SM2, Ratchet, and TLOU are running circa 100-120fps uncapped in those new performance modes.



It's always going to be a difficult balancing act to sell the Pro console as the new hotness without making the existing PS5 look like a piece of shit in the process.

That and there isn't a whole lot of software that will show the big benefits yet. In a couple of years when ray-tracing sees more adoption and more UE5 projects start shipping, the PS5 Pro will make the PS5 look like Xbox Series X vs Series S.
 

simpatico

Member
I think I recall hearing/watching from TLOU2 Grounded that their trailers/reveals for one of the bigger events took like a year or something just to have it ready. Then there's other devs taking 6 months. Wild amounts of time nonetheless.
That's just shitty work ethic. You've seen fan trailers right? Ones people throw together for free in their spare time? They really aren't that far off from what AAA put out. Perfect example of why AAA gaming is just bleeding money. Sure you can spend a year making a trailer, but you can also spend 3 days. If you have a metrics guy who says he crunched the numbers and he knows there's ROI for spending a year making a trailer, walk him out of the building immediately.
 
Top Bottom