• 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.
  • Hey Guest. Check out the NeoGAF 2.2 Update Thread for details on our new Giphy integration and other new features.

Analysis Observer: System Redux is Simply Better on PS5 Than Xbox Series X

Agnostic2020

Member
Oct 24, 2017
4,672
19,827
805
and people still dont believe MS are behind with their dev tools? theyve had just as long as Sony so theres no excuses, but its clear as day that especially the dx12u raytracing side of things is behind schedule.
Ms chip was actually finalised h1 2019 vs h2 2019 for ps5 as per report. Phil Spencer has had xsx in his home since last November .
 

CAB_Life

Gold Member
Dec 12, 2019
1,026
2,348
520
Toronto
christianadrianbrown.com
Launch period is when a narrative is set for generation and many xbox fans called xbox api issues FUD 2 weeks ago. So why the change of heart ?? 👀😅

Why be so partisan and inflammatory when you could—going by the people who actually know things and aren’t just armchairing—be wrong? In the end, everyone who engages in this kind of behavior just looks stupid. I don’t care what side of the isle you’re on (and I’m on neither, since I flip-flop to whatever manufacturer/ device serves me best each gen).

And you all are bickering over literally a few dropped frames in most instances. It’s so minuscule as to be ridiculous. Countless threads on this nonsense when we should be celebrating the achievements of these consoles and their games. Oh, but Xbox has no games, sorry—there, I saved you the retort.
 

Monkeygourmet

Member
Jul 8, 2019
947
2,378
470
 

SegaSnatcher

Member
Jan 31, 2020
322
689
320
Scottsdale, Arizona
I don't want to come off as entitled, but what is the hold up with Digital Foundry making comparison analysis videos? Since launch we've only got one analysis video of comparing multiplats on the next gen consoles. Smaller channels have been cranking out solid analysis videos like every other day with great data. Are they seriously afraid to give PS5 so many wins early on? I'm not one for conspiracy theories or anything, but its just weird to me.
 

timmyp53

Member
May 26, 2013
2,008
803
660
USA
I don't want to come off as entitled, but what is the hold up with Digital Foundry making comparison analysis videos? Since launch we've only got one analysis video of comparing multiplats on the next gen consoles. Smaller channels have been cranking out solid analysis videos like every other day with great data. Are they seriously afraid to give PS5 so many wins early on? I'm not one for conspiracy theories or anything, but its just weird to me.
It's a new generation. New capabilities. DF have stated they are counting and going through making marks frame by frame for tearing. It'll get better as their tools mature so to speak. No joke intended.
 
Last edited:

OutRun88

Member
Jun 12, 2020
1,290
2,685
385
Well all of you arguing about which console it runs better on and why, I’m just wondering how I’ve never heard of this amazing looking game. Any word on if they plan on doing a physical release?
 

Monkeygourmet

Member
Jul 8, 2019
947
2,378
470
Simple facts here:

• MS said they had the most powerful machine

• MS has no big exclusives

They needed to show Series X was the best place to play multiplatform.

They haven’t.

The biggest draw is playing old exclusives on Series X yet they even had a problem and weaker exclusive line up compared to other platforms last gen. Simply zero reason apart from QOL improvements to own a series X at the moment.
 

Robins

Gold Member
Apr 21, 2016
144
205
425
I don't want to come off as entitled, but what is the hold up with Digital Foundry making comparison analysis videos? Since launch we've only got one analysis video of comparing multiplats on the next gen consoles. Smaller channels have been cranking out solid analysis videos like every other day with great data. Are they seriously afraid to give PS5 so many wins early on? I'm not one for conspiracy theories or anything, but its just weird to me.

They are too busy having fun playing next gen games to be giving a toss about making videos 😂
 
  • LOL
Reactions: Slothbeing1209

ToadMan

Gold Member
Dec 2, 2018
1,915
7,507
655
Did you notice that Cerny didn't give any benchmarks to show how higher clocks with lower Cu's is better?
Digital Foundry actually went and did the testing to get the benchmarks, and those benchmarks showed a performance advantage with a lower clock and more Cu's.

But Cerny did point out that using more parallelism wasn’t easy to get performance increases from ...


And here we are.
 
  • Like
Reactions: Yakuzakazuya

Aceofspades

is a 9 year old console warrior
Mar 31, 2015
3,705
7,491
865
Did we ever get the final say on what the PS5's GPU is? Is it pure RDNA 2, or did Sony take RDNA 2 as the base and make a load of tweaks?

There was something a long while back about how AMD and Sony collaborated on the PS5's GPU, with AMD planning on using the tech as the basis for RDNA 3?

RDNA1, or 2 or 10! Does it matter? The little thing is BEAST 🔥
 
Feb 15, 2013
8,897
7,581
855
London
So for the people around here to think this isn't some kind of GDK related issue, and that MIcrosoft should just close up shop because it's over, and that this platform which is fathoms easier to build games for and around won't have its issues quelled; is complete insanity.

You may have missed this but I posted an interview with the Technical Director of DIRT 5, and he says that MS dev tools are NOT behind Sony's:

David Spingate said:
(Are MS dev tools a bit further behind than Sony's?) I mean I wouldn't say that. That depends on your experience....thinking about tools, everything is coming in a bit hot.

(I am hearing that GDK is basacilly in its infancy and there is a lot of stuff in on the table?)
It's in its infancy, because it's new, but it's not particurly lacking...

(Like barebones?) It's not barebones, they brought a lot, no, most of the tools over from of SDK in terms of familitary and how it works, in fact stability and speed improved a lot on GDK, but yeah, I am happy with it, if anything there is very little small things I would like, I have weekly meetings with XBOX and my account manager get a lot of feedback from me, I am actually very happy with it.

Timestamped:


Watch it. This totally debunks your theory that MS tools are so behind. As people have been saying, it's likely a hardware design issue.
 

kuncol02

Member
Apr 4, 2020
1,321
1,455
455
But Cerny did point out that using more parallelism wasn’t easy to get performance increases from ...


And here we are.
And that never was a problem since shaders were invented. Always cards with more shaders and lower clocks were faster than cards with lower shader unit counts and faster clocks.
 

Valued Gamer

Banned
Oct 26, 2020
489
1,055
385
But Cerny did point out that using more parallelism wasn’t easy to get performance increases from ...


And here we are.
If you are trying to say that a GPU clocked at 2.23ghz with 36 Cu's is going to perform the same or better than one clocked at 1825ghz with 52 Cu's, using the same architecture, then you are wrong.
It does not work like that.
Go and watch the DF video where they actually do the testing.
 

spyshagg

Should not be allowed to breed
May 13, 2006
1,597
294
1,450
If you are trying to say that a GPU clocked at 2.23ghz with 36 Cu's is going to perform the same or better than one clocked at 1825ghz with 52 Cu's, using the same architecture, then you are wrong.
It does not work like that.
Go and watch the DF video where they actually do the testing.

Only if we assume its the exact same architecture. From the results, there must be differences.

The only data we have says XSX is full rdna2 and PS5 is rdna2 based. Thats all we can say.
 

Valued Gamer

Banned
Oct 26, 2020
489
1,055
385
Only if we assume its the exact same architecture. From the results, there must be differences.

The only data we have says XSX is full rdna2 and PS5 is rdna2 based. Thats all we can say.
There is more to it than the GPU. You can see a difference in games based on their APIs alone.
 

ToadMan

Gold Member
Dec 2, 2018
1,915
7,507
655
If you are trying to say that a GPU clocked at 2.23ghz with 36 Cu's is going to perform the same or better than one clocked at 1825ghz with 52 Cu's, using the same architecture, then you are wrong.
It does not work like that.
Go and watch the DF video where they actually do the testing.

Im not trying to say anything of the sort.

I am saying that PS5 and Xsex cannot be compared by Tflops because their architecture is different.

Increasing parallelism is not necessarily going to improve performance with developing technology.

How do I know that? Well the head to heads are out there.
 
Last edited:

ToadMan

Gold Member
Dec 2, 2018
1,915
7,507
655
I realize that many of you are having fun taking the piss, but we did hear about problems with Xbox tools months prior to launch, and Sony was said to have been leagues ahead with their kits. Didn’t someone from DF actually confirm this via Twitter, too? Maybe just calm down a bit and see how this shakes out after the initial launch period. We’re not looking at huge deltas in performance or resolution, either, we’re talking about a few frames here and there on about 4 games at this point (one of which, DMC, was titled in XSX’s favour).

Back before Craig no one was talking about poor dev environments.

In fact here’s an example of just how powerful, capable and smooth the transition to Xsex was supposed to be ...

A long list of love ins - go back to the post and check what all these developers had to say about their Xsex experience...

So far what I have gathered on what devs think of Xbox Series X|S
Colin Penty, Technical Art Director at Coalition

Daniele Galante and David Garcia, Ninja Theory Audio Team, Ninja Theory

Erwan LeCun, co-founder of Ape Tribe Games


Joel Baker, Technical Director at Hinterland Games

Gennadiy Korol, Director of Technology at Moon Studios

Kevin Floyer-Lea, Chief Technology Officer at Rebellion

Mike Rayner, Studio Technical Director at The Coalition

Alexandre Sabourin, Team Lead at Snowed In Studios


Grant Kot, working on a Real-Time Particle Physics based game for Xbox Series X and high end PC's

Richard Geldreich, worked at Space X, Valve and Ensemble

Chris Grannel, ex-Sony first party dev who worked on Formula One Series, Wipeout HD:Fury and Killzone 2 for Playstation.

Wojciech Piejko, Lead Game Designer and Jacek Zięba, Producer at Bloober Team

Johannes Kuhlmann, head of core technology at FishLabs

Ljubomir Peklar, designer at Ebb Software

Tor Frick
and Arcade Berg, co-founders of Neon Giant

David Springate, Technical Director at Codemasters

Manuel Fernández, Co-founder and Programmer at Out of the Blue

Tatiana Delgado, Co-founder at Out of the Blue

Marcin Makaj, CEO, Designer and Programmer at The Moonwalls

Jan Kavan, Co-Founder at CBE Software

Gavin Stevens, Co-Owner and Design Lead at Team Blur Games

Gabe Newell, Co-Founder and President of Valve

Tomas Sala, Developer of The Falconeer

Petr Kolář and David Kolečkář, Project Lead at Bohemia Interactive

Masayoshi Yokoyama, Yakuza Series Chief Producer at Ryu Ga Gotoku Studio

Anton Yudintsev, CEO at Gaijin Games

David Cage, CEO at Quantic Dream

It’s only post - Craig that people started to look for excuses. And of course that has picked up plenty of steam as these head to heads come out
 

megreotsugua

Member
Dec 3, 2019
1,454
3,449
400
You may have missed this but I posted an interview with the Technical Director of DIRT 5, and he says that MS dev tools are NOT behind Sony's:



Timestamped:


Watch it. This totally debunks your theory that MS tools are so behind. As people have been saying, it's likely a hardware design issue.
And that never was a problem since shaders were invented. Always cards with more shaders and lower clocks were faster than cards with lower shader unit counts and faster clocks.

It's a hardware design issue. Xsex is not actually wide in the strict sense. It's slow, yes, but it is not wide.

Xsex only has 4 shader arrays like the PS5. It's not wide. Xsex's shader arrays do have more CUs but the new RDNA2 GPUs from AMD show us that the optimal and ideal CU number per shader array for RDNA2 architecture is 10 and not 14. Xsex's design is for marketing to grab that TF advantage on paper but in actual practice those CUs are actually data-starved and are bottlenecked by the caches in the system.

You all gotta listen to geordiemp geordiemp , he's been saying these things for over a month now.
 

Valued Gamer

Banned
Oct 26, 2020
489
1,055
385
Im not trying to say anything of the sort.

I am saying that PS5 and Xsex cannot be compared by Tflops because their architecture is different.

Increasing parallelism is not necessarily going to improve performance with developing technology.

How do I know that? Well the head to heads are out there.
The head to heads are interesting to watch, and will be more interesting the longer they go on.
I am waiting for DFs ones because they have a ton more inside dev contacts that they can get the inside skinny from. And, they will tell what they hear without fear or favour.
My point is don't jump yet, there's still some water to pass under the bridge.
 

scydrex

Member
Aug 10, 2017
616
420
310
what surprised us even more is that even with ray tracing enabled, the PS5 version of Observer: System Redux still performs better than its Xbox Series X counterpart. There are some parts of the apartment block where the Xbox Series X seriously struggles, ruining the immersion a little. It’s perhaps at its worst when you enter the basement.

PS5



XSX




Because you are too ugly
 
  • LOL
Reactions: Bo_Hazem
Jan 29, 2019
5,511
5,864
495
Because there's nothing to admit? You actually think this is going to be the modus operandi going forward? You actually think the PlayStation 5 is going to be the "powerful" system?

Nah bro, just nah. This is a set of software kinks that will be resolved, and nothing more.
As of now bro. There is nothing to make us think your assumptions are correct, so please:
 

NXGamer

Member
Jan 7, 2016
716
2,045
670
wait, so these 2 systems run 2 completely different settings and people now argue about performance?

how can you compare 1080p + raytracing with 4k without raytracing and think you have any accurate understanding of which system runs the game better?

this also highly suggests an issue with development tools,
you can joke about it all you want, but that's extremely likely the case here as every single piece of evidence points to it being the case.
Ah, recall the hate I got for stating this issue when talking to DEVS on PS5 SX and now I see that is now a conversation piece.

For those that shoot down these kind of comments with no fact or feeling behind them, please, in future, recall this as you may then return to it in the future when it suits you to.
 

Panajev2001a

GAF's Pleasant Genius
Jun 7, 2004
18,268
11,050
2,110
Because there's nothing to admit? You actually think this is going to be the modus operandi going forward? You actually think the PlayStation 5 is going to be the "powerful" system?

Nah bro, just nah. This is a set of software kinks that will be resolved, and nothing more.

Since you also spread the talk that 18% was the minimum gap, are you still saying that or now the goal post moved to any win proves you right? :rolleyes:.
 
Feb 15, 2013
8,897
7,581
855
London
It's a hardware design issue. Xsex is not actually wide in the strict sense. It's slow, yes, but it is not wide.

Xsex only has 4 shader arrays like the PS5. It's not wide. Xsex's shader arrays do have more CUs but the new RDNA2 GPUs from AMD show us that the optimal and ideal CU number per shader array for RDNA2 architecture is 10 and not 14. Xsex's design is for marketing to grab that TF advantage on paper but in actual practice those CUs are actually data-starved and are bottlenecked by the caches in the system.

You all gotta listen to geordiemp geordiemp , he's been saying these things for over a month now.

Yes, good points.

There's two problems trying to get a rasterization advantage out of 52 slower CUs vs 36 faster CUS.

1. Filling all those CUs with work (triangles) is hard. To do that requires an extremely efficient graphics pipeline. If one other part of the pipeline encounters even a small bottleneck, this efficiency is not achieved
2. Because the clocks are so low comparative to PS5, it's not just rasterization units running slower, but everything else is running slower i.e processing command buffer, L2 and other caches (per Cerny's talk).
 

truth411

Member
May 19, 2014
2,034
817
530
So I did some math and actually, the difference in gpu clock speed between the ps5 and XSX is actually greater than the difference in Teraflops between the two machines. *roughly

XSX TF advantage : 17%*
PS5 gpu clockspeed advantage : 22.2%*

Initially I thought like @DynamiteCop! and just assumed it was the tools.

However if we think about it, even if the games on both machines are operating at the same resolution, the geometry pipeline and basically everything else is pretty much faster on ps5, meaning it might be that, in the best case the XSX can match ps5, but it makes sense that PS5 is doing better in these initial games.

I'm not saying the results we're seeing won't get better on the XSX side, but it could be the case that Ps5 will actually be a slightly better multiplat machine ; either offering similar performance at the same resolution or, clearly better performance/graphics at a lower resolution.

XSX might show its advantage in its console exclusive games, where games rely more on memory bandwidth or pure floating performance and less on clockspeed. Regardless, we've got more to see from the likes of DF and it's early days for sure, but it's becoming more clear that these machines are much more evenly matched than it appears.

On the bright side, XSX remains much more appealing to me due to its incredible BC implementation, and having skipped the Xbox one generation myself, among other things i'm not too thrilled with on Ps5, but I digress.

Fixed lol.
 

Armorian

Member
Jan 17, 2018
2,492
3,664
465
They are too busy having fun playing next gen games to be giving a toss about making videos 😂

What next gen games? Everything is cross platform... Only next gen game is DeS (and that's a remake...).

PS4 had much better tools than XONE.
Nothing changed in 2020.
MSFT is not the same, get used to it.

Didn't stop X1X from some massive wins. And this is a SOFTWARE company.
 
  • Fire
Reactions: Bo_Hazem

Md Ray

Member
Nov 12, 2016
2,160
7,099
585
India
Apparently, and this isn't confirmed but highly likely. The reason Sony doesn't have a fully featured RDNA 2 GPU is because instead of waiting for AMD to finish they took what was available, and then kind of went their own way building the remainder of the feature sets themselves under proprietary methods.

This allowed them to get their devkits out sooner, their tools, developers to get more in tune and comfortable with developing for it. On the Microsoft side it appears they waited until AMD's hardware was entirely done, which like the opposite of Sony explains the reported late arrival of devkits, spec targeted PC's being used and as you can see now software which appears to be having some issues.
And this is coming from the guy who claimed Series S doesn't exist.
 
Last edited:

Md Ray

Member
Nov 12, 2016
2,160
7,099
585
India
If you are trying to say that a GPU clocked at 2.23ghz with 36 Cu's is going to perform the same or better than one clocked at 1825ghz with 52 Cu's, using the same architecture, then you are wrong.
It does not work like that.
Go and watch the DF video where they actually do the testing.
And GPUs aren't just made up of CUs. There are other units and some of them are faster on PS5 than XSX.
 
Last edited:

truth411

Member
May 19, 2014
2,034
817
530
Apparently, and this isn't confirmed but highly likely. The reason Sony doesn't have a fully featured RDNA 2 GPU is because instead of waiting for AMD to finish they took what was available, and then kind of went their own way building the remainder of the feature sets themselves under proprietary methods.

This allowed them to get their devkits out sooner, their tools, developers to get more in tune and comfortable with developing for it. On the Microsoft side it appears they waited until AMD's hardware was entirely done, which like the opposite of Sony explains the reported late arrival of devkits, spec targeted PC's being used and as you can see now software which appears to be having some issues.
Is this a joke post?
XSX is pretty much a Vanilla RDNA2 because they want there games on the console and PC. In other words, Ease of development and porting. Thats not what Sony is doing, So they customized their RDNA2 GPU to get more performance out of the console. PC isnt the focus at all.
 

Aladin

Member
Jul 26, 2020
370
443
280
And GPUs aren't just made up of CUs. There are other units and some of them are faster on PS5 than XSX.
Go on, list all of those "other" units. Badly optimised game, you will see 8% difference, uncapped optimised game 10-15% difference. It is because clocks alone cant makeup for the large die in series x.
 

Md Ray

Member
Nov 12, 2016
2,160
7,099
585
India
Go on, list all of those "other" units. Badly optimised game, you will see 8% difference, uncapped optimised game 10-15% difference. It is because clocks alone cant makeup for the large die in series x.
ROPs for e.g. caches, rasterization have a 22% advantage on PS5. Games relying heavily on pixel fillrate for e.g. might pull ahead on PS5. Games relying on the floating-point will pull ahead on XSX.
 
Last edited:
  • Like
Reactions: PharaoTutAnchAmun