• 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

Gaiff

SBI’s Resident Gaslighter
This is incorrect. It depends on split between raster and RT frame budget. I did the math assuming 45% raster and 2.5x RT boost. Show me your calculations if you disagree.

From my earlier post:

OTTekCT.jpg


1OAMi8i.jpg

Z4oeBwG.jpg
Yeah, except there's a big problem with this. You're overshooting the performance of the PS5 by a lot, especially in Performance Mode.

D49GNYK.png


According to DF, the 4060 is 19% faster than the PS5 in this game, and the 6700 XT is 6% faster than the 4060. Hell, the regular 6700 is 10% faster than the PS5 even. You're starting the PS5 at a baseline 26% faster than its real-world performance.

We could be looking at a CPU bottleneck on the PS5 but it's a bit unlikely because the PS5 and 2070S are neck-and-neck and the game's DRS is at its lower bound of 720p.

If you keep picking up cards that are over 25% faster than the PS5 as a baseline such as the 2080 Ti or the 6700 XT in this particular instance, your results will be skewed.
 
Last edited:

ChiefDada

Gold Member
Yeah, except there's a big problem with this. You're overshooting the performance of the PS5 by a lot, especially in Performance Mode.

In 1440p mode, true but in the same vein I'm almost certainly understating PS5/PS5 Pro at native 4k mode as well. 6700/xt will not outperform PS5 at native 4k in normal circumstances due to bandwidth constraint.

Edit: Actually no, I scale back some of my concession because you are showing internal 720p render. 6700/xt will overperform here at low resolutions thanks to faster clocks and infinity cache.
 
Last edited:
well yeah I assume a pro model would have premium specs. but that depends, I mean would old school (fat or slim) PS5 receive a price cut along with the launch of the "Pro" model? I'm not sure how they did this before, come to think of it the "PS5 Pro" and the "Nintendo Switch 2" should be launching around the same time, well google says "PS5 Pro" launches at the end of 2024 and "Nintendo Switch 2" comes at the beginning of 2025. what a time to be alive 😎
I mean extra premium most are expecting 599 but maybe if it’s 699 the specs and build quality will be even better
 

Gaiff

SBI’s Resident Gaslighter
In 1440p mode, true but in the same vein I'm almost certainly understating PS5/PS5 Pro at native 4k mode as well. 6700/xt will not outperform PS5 at native 4k in normal circumstances due to bandwidth constraint.
The 6700 XT most certainly will.
Edit: Actually no, I scale back some of my concession because you are showing internal 720p render. 6700/xt will overperform here at low resolutions thanks to faster clocks and infinity cache.
The PS5 won't claw back a 26% performance deficit. In Alan Wake 2 for instance, it gets back about a 10% performance improvement going from 1440p FSR Balanced to 2160p FSR Balanced.
Does Avatar still do RT in its performance mode?
Yes.
 

Audiophile

Gold Member
Zen 2 @ 3.85GHz is quite a disappointment. We'll be lucky to get a ropey 40fps mode in any CPU limited 30fps games (GTAVI for eg.). Was expecting 4GHz at a minimum for a Zen 2 setup. 3.85-4.0GHz is what I would have expected on an N4P Zen 4C.

If this is real then I think it's very clearly N6.

I was camp $599, but I'm thinking $499-$549 with a price-dropped base PS5 to $349/$399 now.

Hoping they at least get some extra cache in there.

As for RT, double throughput sounds nice; and with the 66% jump in compute that gives us a theoretical boost of 3.3x. I expect the extra 128GB/s of bandwidth and more efficient overall architecture would help get it further in certain cases.
 
Last edited:
Most of the games weren’t CPU bottlenecked anyway - I don’t see it as a major issue honestly.

1080p upscaled to 2160p through PSSR with a healthy amount of RT effects and likely 60 FPS? Count me in.
You're right, most games are GPU bound, hence the low resolution in many recent games (especially UE5 ones). Still a bummer though. I was hoping for at least a nice clock upgrade. Oh well. Can't get it all.
 
You're right, most games are GPU bound, hence the low resolution in many recent games (especially UE5 ones). Still a bummer though. I was hoping for at least a nice clock upgrade. Oh well. Can't get it all.

It should still smooth things out a bit. If we assume a linear 10% increase in CPU performance, under the most extreme CPU-limited cases, we should see 28 FPS games going to 30 FPS, 55 FPS games going to 60 FPS and 110 FPS games going to 120 FPS. Frametimes should get improved as well. A lot of PS5 games, which are CPU limited, are not that far away from having locked FPS in most cases, and VRR does a great job fixing stutter anyway.
 

clarky

Gold Member
It should still smooth things out a bit. If we assume a linear 10% increase in CPU performance, under the most extreme CPU-limited cases, we should see 28 FPS games going to 30 FPS, 55 FPS games going to 60 FPS and 110 FPS games going to 120 FPS. Frametimes should get improved as well. A lot of PS5 games, which are CPU limited, are not that far away from having locked FPS in most cases, and VRR does a great job fixing stutter anyway.
This get me about as wet Ghani's flip flop
 

Panajev2001a

GAF's Pleasant Genius
Zen 2 @ 3.85GHz is quite a disappointment.
Actually it is still 3.5 GHz, for the boost you mention they need to drop the GPU max clocks by about 1.5% (for an approx 1% GPU performance hit).

We'll be lucky to get a ropey 40fps mode in any CPU limited 30fps games (GTAVI for eg.).
True, but likely there are not many games in this scenario. For a lot of gamers if they figure out how to do frame-generation decently this will satisfy many of the complaints (especially for OLED panel people).

Was expecting 4GHz at a minimum for a Zen 2 setup. 3.85-4.0GHz is what I would have expected on an N4P Zen 4C.
I thought it was unlikely Sony would jump to N4, not a cheap jump for a mid-generation upgrade. Also unlikely to have the Slim on N6 and Pro on N4 about 1 year apart from each other. N6 was what they got from AMD and they went for size and power efficiency on one side and performance on the other (within limits, knowing the GPU and memory bandwidth were the main concerns).

If this is real then I think it's very clearly N6.
Yeah, but it was kind of to be expected. Zen 4c and a large manufacturing node jump were not (especially after the Slim was announced).Bigger chip but still N6 is denser than N7 they started with without spending money for a more expensive node like N4/N5.

Sony probably got combo deal here like for the PS4 Slim SoC redesign and the PS4 Pro. Essentially start from the base model, minor to moderate process improvement, and push one of the SoC’s further for the Pro model down the line changing components you think will make the most difference only (like the GPU for example).

RDNA3+ is also far less power hungry at the same clock speed as RDNA2 and they took advantage of that (more CUs and newer ones too, picking features from RDNA4).
 

Panajev2001a

GAF's Pleasant Genius
Starfield is CPU bound. So no it won’t. Maybe 40fps
That might be a game they push for the 3.85 GHz mode and can turn SMT on if the game benefits from it too. On XSX they need to decide whether to have SMT on or off to enable the high frequency mode. This is higher frequency and you can still enable SMT too.
 
Games that are CPU Bottlenecked aren't constantly bottlenecked by the CPU, it just means that occasionally the frametime will exceed the target for 30/60/120fps and cause a dip, if you give that CPU a slight buff it will make the target more often, that could mean we get GTA6 on PS5 at 30fps or 35/40 if it's unlocked, but 60 on PS5 Pro because the GPU will still be limiting the FPS even if the CPU is the thing that ultimately causes the big drops which then mean the dev team lower the target framerate so we don't know where the GPU limit is in reality.

If i put my 4090 in my old PC my average FPS will be way higher paired with the 9900k than it is with my 2080Ti that's in that machine, where as putting the 2080Ti in my machine with the 12900kf instead of the 9900k won't see any impact to average framerates, but might see the lows be slightly higher on average because the CPU is better.
 
Top Bottom