-
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.
PS5 Pro devkits arrive at third-party studios, Sony expects Pro specs to leak
- Thread starter Bernoulli
- Start date
HeisenbergFX4
Gold Member
I don't remember what the cooo cooo shit was even about honestly.
Rudius
Member
They don't care if you don't upgrade to a Pro and keep spending on PS5 software. They don't make (much) money selling consoles, but selling games, DLCs and subscriptions, the console is just a means to that end.
The Pro will exist as a way to keep the people that care about performance buying games on the PS5 ecosystem, instead of migrating to PC.
playsaves3
Member
You’re massively exaggerating how good the 4060 and 4070 are in pure rasterazatioj especially with its vram limitations. The 4080 isn’t that special in pure rasterization it’s its feature set like dlss and of course raytracing is where it really shines and where it will still beat the pro. You really don’t expect exclusives to like last of us or horizon to not run better than that card on here?
playsaves3
Member
Well that is as long as they make the pro good enough so hopefully they do
S0ULZB0URNE
Member
"Rasterization"
4070 is 29TF's
Best case scenario as I stated already is 4070ish.
As in slightly below it.
Again I hope I'm wrong and it's better.
As far as ram... the latest is saying Pro is only 16.
(3 years ago I was told 20)
Last edited:
Insane Metal
Member
That was really fun and that's how we started chatting right?
playsaves3
Member
So you are aware I wasn’t talking about raw hardware specs but performance in actual games right? The pro isn’t going to have the exact hardware grunt of the 4080 yes that doesn’t mean it won’t outperform it in this category on a large amount of games because of you know pc versions being heavier and running os etc, optimization, low level apis etc also the pro is going to have a somewhat newer feature set that may alleviate some of its tasks the 4080 won’t have access to. Based on how last of us for example runs worse on a 4080 than a ps5 is it really that unrealistic for the pro to outperform the 4080 there same with ratchet in pure raster
S0ULZB0URNE
Member
You said Rasterization.
Get back to me when you understand what this means.
playsaves3
Member
Yes I know what raster is I fully expect it to outperform the 4080 there at least in exclusive titles solely because of the pc vs console environment
Bry0
Member
You wanna bet? That’s pretty optimistic.
THE DUCK
Banned
Maybe first batches, but "up to" is a vague term. If Sony is break even now ps5, series x can't be far behind.
Ms can hire some of the best engineers in the world, it's a bit weird to say somehow Sony has the market cornered on hardware efficiency. Ps5 was a clever design, but let's not forget prior machines where no miracle.
EDMIX
Writes a lot, says very little
I mean...looking at how good RDR2 looks, I don't know why anyone would even doubt GTAVI.
PS5 and Series X are many times more powerful then PS4 and XONE.
TheRedRiders
Member
Coo-coo-coo-coo = 13 teraflops confirmed!
Tommy Fisher = TF = PS5 is 13 teraflops CONFIRMED
Then Tommy went on to get banned for leaking the PS5 controller which was actually just the DS4.
Man those were the days!
HeisenbergFX4
Gold Member
That was the place, when I really first starting chatting here
That "fat" controller was so funny and laying on that Sony Vaio laptop then Cerny gave the road to PS5 talk and what was he using?Coo-coo-coo-coo = 13 teraflops confirmed!
Tommy Fisher = TF = PS5 is 13 teraflops CONFIRMED
Then Tommy went on to get banned for leaking the PS5 controller which was actually just the DS4.
Man those were the days!
existensmaximum
Member
Well why does anyone ever post anything online?
Rudius
Member
Many games employ dynamic resolutions and give the option of unlocked framerates. They will automatically benefit from more power.
Rudius
Member
This is an interesting point. Unlike with used games, it is to Sony's benefit if people sell their used PS5s to get new Pros. The money for them is in software, not hardware.
Some of those will even be given for free to brothers, cousins etc. that would never have bought a PS5, but after being handed one will end up spending on games and subscriptions. I gave my PS4 Pro to a cousin when I bought a PS5; he acquired some games since then and still uses it.
Insane Metal
Member
Fun times! Can't wait for it to start all over again lol.Coo-coo-coo-coo = 13 teraflops confirmed!
Tommy Fisher = TF = PS5 is 13 teraflops CONFIRMED
Then Tommy went on to get banned for leaking the PS5 controller which was actually just the DS4.
Man those were the days!
Physiognomonics
Member
I think the 13 Tf comes from the fact that some leaker saw PS5 performing better than XSX in one UE5 game (or demo?) and so thought PS5 was above 12tf. But maybe I am missing and or mixing things here.
Last edited:
Insane Metal
Member
You're one of the kindest, nicest users here. Along with my palThat was the place, when I really first starting chatting here
CrustyBritches
HeisenbergFX4
Gold Member
I appreciate the kind words, have always enjoyed chatting with you as well my friend
Andrea Pessino had told a couple of people here who he knew the PS5 devkit was minimum 13TF and up to 15TF
A couple shared the chat with me they had with him since he claimed they had a devkit "under lock and key"
Physiognomonics
Member
Odd. VEGA GCN based devkit?
Rudius
Member
If they release a next gen Xbox two years before the PS6 it will bring them almost no benefit. The first 2 years of the current generation were almost all cross gen games, the biggest exceptions being a few PS5 exclusives. This is not 2005 when games took 1 or 2 years to make, it's tripple that now.Btw a couple of Tweets from Kepler who also got on the "plans changed" train along with my "predictions" in Aug
An early Xbox would probably get zero third party next gen exclusives and it might even help the PS6 to have a better start, since by it's release developers will be closer to abandoning the previous gen.
Now this would be very interesting and it might work. A console that is also a powerful PC would sell a ton, especially outside of places where PCs like that are cheaper. But it should not be just windows 11, but a new windows 12 with the next gen console for extra impact.
buenoblue
Member
? The pro will be lucky to out perform a 3070 lol. A PS5 is 2070/2060 super level at most. Most PS5 games run there performance modes close to 1080p.
S0ULZB0URNE
Member
So it's greater than 64TF's?
You confuse optimization with rasterization.
Skifi28
Member
Those nvidia TF numbers though are not comparable to the AMD ones from the consoles.
Wooxsvan
Member
well ya and i have my doubts. its def not beating my 7900xtx but if it can get look kinda close then it might stem some of my PC purchases
S0ULZB0URNE
Member
The nvidia numbers are not 64 I made a error but 76.
The actual TF's are 48 and change. Pro isn't matching or exceeding it in rasterization performance regardless.
Skifi28
Member
Oh, no argument there. We'll be lucky if it matches a base 4070.
HeisenbergFX4
Gold Member
4070 real world performance likely, not spec for spec
ChiefDada
Member
PS5 already competes with 4070 in many games, particularly Sony 1st party. Per DF, TLOU Pt. 1 runs locked native 1440p 70-80fps/High Settings. It's a good game to test because there is no DRS and PS5 can run uncapped. I found 4070 playthrough at 4K DLSS balanced (~1296p internal resolution) with High settings running in the 70s typically.
S0ULZB0URNE
Member
If you are joining in this convo.
I suggest you read all the posts from the beginning.
From your post It's obvious you haven't.
ChiefDada
Member
Lol why would I need to waste time doing this when there's no problem rebutting your comment as is? TLOU Pt. 1 is a pure raster based game. You said PS5 Pro would "likely" fall between 4060 and 4070 in raster performance. This example suggests otherwise.
S0ULZB0URNE
Member
RASTERIZATION PERFORMANCE
ChiefDada
Member
Lol what else are we measuring when we compare FPS count between GPUs on raster-only games???
angrod14
Member
Well, I hope you're right regarding performance, because people here are speculating the CPU will be barely improved and thus no chance of getting 60 fps in games that will be 30 on the base system (GTA VI for example). PS4 Pro was like that and it sucked ass.
James Sawyer Ford
Member
I wonder how significant the DLSS gains will be if it's competitive or better to what nvidia has?
Are we talking about a 30% efficiency gain with minimal tradeoffs?
Are we talking about a 30% efficiency gain with minimal tradeoffs?
playsaves3
Member
It actually could possibly trade blows/ beat it in rt specifically if the rdna 4 rat accelarwtor is true
playsaves3
Member
These people heavily exaggerate the raw Easter performance of these cards it’s they really think the ps5 pro won’t beat a 4060 in exclusive titles when it comes to raster?
Gaiff
SBI’s Resident Gaslighter
It doesn't. For one, you're not even doing a side-by-side so it's utterly pointless. The PS5 could be running at 61fps in this area for all we know. For two, this video is from 7 months ago when the game was still severely unoptimized on PC. Notice something in this screengrab I made?
Look at the GPU usage, a paltry 83% while the CPU, a 5600X, is at 90%. The game is CPU-limited and had severe threading issues that weren't fixed until months after.
A whopping 35% improvement in a CPU-bound scene. The usage went from 96% all the way down to 60% on a 5600X. The best data point we got where shots are lined up is the DF video where Alex compared the PS5 to a 2070S running High settings.
Here, the PS5 is outperforming the 2070S by 35% once it settles down. During the explosion, there is a massive fps spike on PC that causes the performance to tank and at that moment, the difference is 47% in favor of the PS5 but it only briefly happens. This would put the PS5's performance in line with a 2080 Ti/3070, nowhere near a 3080, let alone a 4080. Even using the 47% figure, you'd end up on the level of an RX 6800, again, short of a 3080.
So no, there aren't many games where the PS5 matches a 3080, in fact, there isn't a single one, which is expected since the 3080 is something like 60-80% faster in rasterization.
And I say 3080 because the 4070 and it are very close, similarly to the 3070 and 2080 Ti.
No, the 4080 isn't on par with the PS5 on TLOU. This is moronic to even suggest.
Last thing, I wouldn't even use TLOU as an example because even among Sony ports, it's an outlier for how abysmal it was (and still is lacking, even today). None of their other games exhibit anything close to that.
Also, the 4060 at times barely performs better than the 3060 which is a tad slower than the regular PS5 in general, so of course, the PS5 Pro will mop the floor with a 4060. I'm expecting it to land somewhere around a 7800 XT in rasterization. Jury is still out with RT because I haven't got the faintest idea of how these improvements will translate to the real world.
Last edited:
Gaiff
SBI’s Resident Gaslighter
Not sure I understand the question.
ChiefDada
Member
It doesn't. For one, you're not even doing a side-by-side so it's utterly pointless. The PS5 could be running at 61fps in this area for all we know. For two, this video is from 7 months ago when the game was still severely unoptimized on PC. Notice something in this screengrab I made?
Look at the GPU usage, a paltry 83% while the CPU, a 5600X, is at 90%. The game is CPU-limited and had severe threading issues that weren't fixed until months after.
A whopping 35% improvement in a CPU-bound scene. The usage went from 96% all the way down to 60% on a 5600X. The best data point we got where shots are lined up is the DF video where Alex compared the PS5 to a 2070S running High settings.
Here, the PS5 is outperforming the 2070S by 35% once it settles down. During the explosion, there is a massive fps spike on PC that causes the performance to tank and at that moment, the difference is 47% in favor of the PS5 but it only briefly happens. This would put the PS5's performance in line with a 2080 Ti/3070, nowhere near a 3080, let alone a 4080. Even using the 47% figure, you'd end up on the level of an RX 6800, again, short of a 3080.
So no, there aren't many games where the PS5 matches a 3080, in fact, there isn't a single one, which is expected since the 3080 is something like 60-80% faster in rasterization.
And I say 3080 because the 4070 and it are very close, similarly to the 3070 and 2080 Ti.
No, the 4080 isn't on par with the PS5 on TLOU. This is moronic to even suggest.
Whenever we get in these comparisons, you always point to comparison imperfections that is outside of anyone's control. By the time these PC patches get released, everyone has moved on from the game. I pull from wherever I can. You act as if I'm cherry-picking when in reality I'm not. This is one of the very few current gen/raster only title with uncapped performance mode so I chose. And per DF last video on this, PC port improved but there are still significant CPU bottlenecks due to how ND engine is crafted. That is a plus for the PS5 hardware decompression; it doesn't make the comparison unfair any more than DLSS allows for better image quality for cheaper cost on Nvidia cards.
Gaiff
SBI’s Resident Gaslighter
I'm not acting as if you're cherry-picking, I'm simply showing that you're using outdated data and I'm not blaming you since you might not have been aware of this.
The comparison you used wasn't "unfair", it was just outdated. You want to compare GPU performance, therefore you have to remove the CPU bottleneck as much as possible and when we see a 4070 hitting 80% usage on a CPU that hits 90% usage, we know the GPU is being held back. Per more recent metrics, the PS5 is ahead of the 2070S by 30-48%, which puts it somewhere around a 2080 Ti/6750 XT to a 6800 but typically closer to the former.
Not sure the CPU issues have anything to do with the way the engine is crafted or that ND were unfamiliar with PC development, perhaps both. Bottom line is, I don't think using a CPU-limited scenario of TLOU Part I as an indication of what the PS5 Pro will be capable of, even in optimized games. Unless of course, you expect more messes like TLOU Part I day 1 on PC.
I'd be really surprised if the PS5 Pro matches the 4080 in any game. 4070? Sure. 4070 Ti? Perhaps in select cases. 4080 is very likely out of reach unless we get a big surprise.
James Sawyer Ford
Member
Base PS5 has no DLSS equivalent.
With PS5 Pro, if you were to get it to produce the same resolution and framerate as PS5 Pro, how much LESS could you get away with in terms of TFLop utilization?
playsaves3
Member
The ps5 matches the 3080 in rift apart though tbf it’s the only game I’ve seen usually 3070 is best case. I also never said the ps5 performed like a 4080 literally where did I say that I said there are several games like last of us, rift apart some Ubisoft games where it already clearly outperforms the 4060 and matches a 4070 so why would a pro matching or dare I say edging the 4080 be some crazy idea. This isn’t for every game for referenceIt doesn't. For one, you're not even doing a side-by-side so it's utterly pointless. The PS5 could be running at 61fps in this area for all we know. For two, this video is from 7 months ago when the game was still severely unoptimized on PC. Notice something in this screengrab I made?
Look at the GPU usage, a paltry 83% while the CPU, a 5600X, is at 90%. The game is CPU-limited and had severe threading issues that weren't fixed until months after.
A whopping 35% improvement in a CPU-bound scene. The usage went from 96% all the way down to 60% on a 5600X. The best data point we got where shots are lined up is the DF video where Alex compared the PS5 to a 2070S running High settings.
Here, the PS5 is outperforming the 2070S by 35% once it settles down. During the explosion, there is a massive fps spike on PC that causes the performance to tank and at that moment, the difference is 47% in favor of the PS5 but it only briefly happens. This would put the PS5's performance in line with a 2080 Ti/3070, nowhere near a 3080, let alone a 4080. Even using the 47% figure, you'd end up on the level of an RX 6800, again, short of a 3080.
So no, there aren't many games where the PS5 matches a 3080, in fact, there isn't a single one, which is expected since the 3080 is something like 60-80% faster in rasterization.
And I say 3080 because the 4070 and it are very close, similarly to the 3070 and 2080 Ti.
No, the 4080 isn't on par with the PS5 on TLOU. This is moronic to even suggest.
Last thing, I wouldn't even use TLOU as an example because even among Sony ports, it's an outlier for how abysmal it was (and still is lacking, even today). None of their other games exhibit anything close to that.
Also, the 4060 at times barely performs better than the 3060 which is a tad slower than the regular PS5 in general, so of course, the PS5 Pro will mop the floor with a 4060. I'm expecting it to land somewhere around a 7800 XT in rasterization. Jury is still out with RT because I haven't got the faintest idea of how these improvements will translate to the real world.
playsaves3
Member
They constructed the cpu cost that way to help it run smoother on pc it’s not the cpu in that benchmark isn’t far better than the one in the ps5 anyway when doing this benchmark. So no if anything the results would be worse here if they used a ps5 equivalent cpu. Tbf I don’t like using last of us cause of the pc port was straight up bad so I use rift apart which is decent on pc the ps5 is only %10 slower there.
Crispy Gamer
Member
If anything we should hear something around CES there should be a lot of loose lips at that event
Ovech-King
Gold Member
I think the relative GPU power comparable of the PS5 is even less than a 2080TI at is gives very similar native 4k performance to the 2070S benchmarks when you look around the web. The 2080TI benchmarks always had it most of the time over 60 fps at 4k , at least in 2020 . Since we know the PS5 can do at least 40 fps (mode) at 4K but can't do 60 at this resolution it probably hovers in the 45 fps range tops . PS5 Pro if increase a 2x fold , it will be comparable to a 4070 Super (tbd) I suppose